id	sid	tid	token	lemma	pos
fcis-26720	1	1	frontiers	frontier	NOUN
fcis-26720	1	2	in	in	ADP
fcis-26720	1	3	computing	computing	NOUN
fcis-26720	1	4	and	and	CCONJ
fcis-26720	1	5	intelligent	intelligent	ADJ
fcis-26720	1	6	systems	system	NOUN
fcis-26720	1	7	issn	issn	VERB
fcis-26720	1	8	:	:	PUNCT
fcis-26720	1	9	2832	2832	NUM
fcis-26720	1	10	-	-	SYM
fcis-26720	1	11	6024	6024	NUM
fcis-26720	1	12	|	|	NOUN
fcis-26720	1	13	vol	vol	NOUN
fcis-26720	1	14	.	.	PROPN
fcis-26720	2	1	10	10	NUM
fcis-26720	2	2	,	,	PUNCT
fcis-26720	2	3	no	no	INTJ
fcis-26720	2	4	.	.	NOUN
fcis-26720	2	5	1	1	NUM
fcis-26720	2	6	,	,	PUNCT
fcis-26720	2	7	2024	2024	NUM
fcis-26720	2	8	85	85	NUM
fcis-26720	2	9	tibetan	tibetan	ADJ
fcis-26720	2	10	speech	speech	NOUN
fcis-26720	2	11	emotion	emotion	NOUN
fcis-26720	2	12	recognition	recognition	NOUN
fcis-26720	2	13	based	base	VERB
fcis-26720	2	14	on	on	ADP
fcis-26720	2	15	capsule	capsule	NOUN
fcis-26720	2	16	network	network	NOUN
fcis-26720	2	17	and	and	CCONJ
fcis-26720	2	18	spatiotemporal	spatiotemporal	ADJ
fcis-26720	2	19	features	feature	NOUN
fcis-26720	2	20	rongzhao	rongzhao	PROPN
fcis-26720	2	21	huang	huang	PROPN
fcis-26720	2	22	1	1	NUM
fcis-26720	2	23	,	,	PUNCT
fcis-26720	2	24	ang	ang	PROPN
fcis-26720	2	25	chen	chen	PROPN
fcis-26720	2	26	1	1	NUM
fcis-26720	2	27	,	,	PUNCT
fcis-26720	2	28	menglin	menglin	PROPN
fcis-26720	2	29	bai	bai	PROPN
fcis-26720	2	30	1	1	NUM
fcis-26720	2	31	,	,	PUNCT
fcis-26720	2	32	bawangdui	bawangdui	NOUN
fcis-26720	2	33	bian	bian	PROPN
fcis-26720	2	34	1	1	NUM
fcis-26720	2	35	,	,	PUNCT
fcis-26720	2	36	2	2	NUM
fcis-26720	2	37	1	1	NUM
fcis-26720	2	38	college	college	NOUN
fcis-26720	2	39	of	of	ADP
fcis-26720	2	40	information	information	NOUN
fcis-26720	2	41	science	science	NOUN
fcis-26720	2	42	and	and	CCONJ
fcis-26720	2	43	technology	technology	NOUN
fcis-26720	2	44	,	,	PUNCT
fcis-26720	2	45	tibet	tibet	PROPN
fcis-26720	2	46	university	university	PROPN
fcis-26720	2	47	,	,	PUNCT
fcis-26720	2	48	lhasa	lhasa	PROPN
fcis-26720	2	49	tibet	tibet	PROPN
fcis-26720	2	50	850000	850000	NUM
fcis-26720	2	51	,	,	PUNCT
fcis-26720	2	52	china	china	PROPN
fcis-26720	2	53	2	2	NUM
fcis-26720	2	54	national	national	ADJ
fcis-26720	2	55	experimental	experimental	ADJ
fcis-26720	2	56	teaching	teaching	NOUN
fcis-26720	2	57	demonstration	demonstration	NOUN
fcis-26720	2	58	center	center	NOUN
fcis-26720	2	59	for	for	ADP
fcis-26720	2	60	information	information	NOUN
fcis-26720	2	61	technology	technology	NOUN
fcis-26720	2	62	,	,	PUNCT
fcis-26720	2	63	lhasa	lhasa	PROPN
fcis-26720	2	64	tibet	tibet	PROPN
fcis-26720	2	65	850000	850000	NUM
fcis-26720	2	66	,	,	PUNCT
fcis-26720	2	67	china	china	PROPN
fcis-26720	2	68	abstract	abstract	NOUN
fcis-26720	2	69	:	:	PUNCT
fcis-26720	2	70	speech	speech	NOUN
fcis-26720	2	71	emotion	emotion	NOUN
fcis-26720	2	72	recognition	recognition	NOUN
fcis-26720	2	73	is	be	AUX
fcis-26720	2	74	an	an	DET
fcis-26720	2	75	important	important	ADJ
fcis-26720	2	76	branch	branch	NOUN
fcis-26720	2	77	of	of	ADP
fcis-26720	2	78	natural	natural	ADJ
fcis-26720	2	79	language	language	NOUN
fcis-26720	2	80	processing	processing	NOUN
fcis-26720	2	81	that	that	PRON
fcis-26720	2	82	aims	aim	VERB
fcis-26720	2	83	to	to	PART
fcis-26720	2	84	automatically	automatically	ADV
fcis-26720	2	85	recognize	recognize	VERB
fcis-26720	2	86	and	and	CCONJ
fcis-26720	2	87	classify	classify	VERB
fcis-26720	2	88	emotional	emotional	ADJ
fcis-26720	2	89	information	information	NOUN
fcis-26720	2	90	in	in	ADP
fcis-26720	2	91	speech	speech	NOUN
fcis-26720	2	92	through	through	ADP
fcis-26720	2	93	computer	computer	NOUN
fcis-26720	2	94	technology	technology	NOUN
fcis-26720	2	95	.	.	PUNCT
fcis-26720	3	1	in	in	ADP
fcis-26720	3	2	the	the	DET
fcis-26720	3	3	specific	specific	ADJ
fcis-26720	3	4	language	language	NOUN
fcis-26720	3	5	environment	environment	NOUN
fcis-26720	3	6	of	of	ADP
fcis-26720	3	7	tibetan	tibetan	PROPN
fcis-26720	3	8	,	,	PUNCT
fcis-26720	3	9	due	due	ADP
fcis-26720	3	10	to	to	ADP
fcis-26720	3	11	relatively	relatively	ADV
fcis-26720	3	12	limited	limited	ADJ
fcis-26720	3	13	research	research	NOUN
fcis-26720	3	14	and	and	CCONJ
fcis-26720	3	15	some	some	DET
fcis-26720	3	16	existing	exist	VERB
fcis-26720	3	17	studies	study	NOUN
fcis-26720	3	18	appearing	appear	VERB
fcis-26720	3	19	cumbersome	cumbersome	ADJ
fcis-26720	3	20	and	and	CCONJ
fcis-26720	3	21	complex	complex	ADJ
fcis-26720	3	22	in	in	ADP
fcis-26720	3	23	feature	feature	NOUN
fcis-26720	3	24	extraction	extraction	NOUN
fcis-26720	3	25	steps	step	NOUN
fcis-26720	3	26	,	,	PUNCT
fcis-26720	3	27	a	a	DET
fcis-26720	3	28	new	new	ADJ
fcis-26720	3	29	network	network	NOUN
fcis-26720	3	30	model	model	NOUN
fcis-26720	3	31	has	have	AUX
fcis-26720	3	32	been	be	AUX
fcis-26720	3	33	proposed	propose	VERB
fcis-26720	3	34	.	.	PUNCT
fcis-26720	4	1	the	the	DET
fcis-26720	4	2	model	model	NOUN
fcis-26720	4	3	is	be	AUX
fcis-26720	4	4	based	base	VERB
fcis-26720	4	5	on	on	ADP
fcis-26720	4	6	a	a	DET
fcis-26720	4	7	capsule	capsule	NOUN
fcis-26720	4	8	network	network	NOUN
fcis-26720	4	9	and	and	CCONJ
fcis-26720	4	10	achieves	achieve	VERB
fcis-26720	4	11	lightweight	lightweight	ADJ
fcis-26720	4	12	design	design	NOUN
fcis-26720	4	13	.	.	PUNCT
fcis-26720	5	1	it	it	PRON
fcis-26720	5	2	only	only	ADV
fcis-26720	5	3	uses	use	VERB
fcis-26720	5	4	mel	mel	PROPN
fcis-26720	5	5	frequency	frequency	PROPN
fcis-26720	5	6	cepstral	cepstral	ADJ
fcis-26720	5	7	coefficients	coefficient	NOUN
fcis-26720	5	8	(	(	PUNCT
fcis-26720	5	9	mfcc	mfcc	NOUN
fcis-26720	5	10	)	)	PUNCT
fcis-26720	5	11	as	as	ADP
fcis-26720	5	12	its	its	PRON
fcis-26720	5	13	input	input	NOUN
fcis-26720	5	14	features	feature	NOUN
fcis-26720	5	15	,	,	PUNCT
fcis-26720	5	16	extracts	extract	VERB
fcis-26720	5	17	the	the	DET
fcis-26720	5	18	spatiotemporal	spatiotemporal	ADJ
fcis-26720	5	19	information	information	NOUN
fcis-26720	5	20	of	of	ADP
fcis-26720	5	21	mfcc	mfcc	NOUN
fcis-26720	5	22	through	through	ADP
fcis-26720	5	23	multiple	multiple	ADJ
fcis-26720	5	24	convolutional	convolutional	ADJ
fcis-26720	5	25	layers	layer	NOUN
fcis-26720	5	26	,	,	PUNCT
fcis-26720	5	27	and	and	CCONJ
fcis-26720	5	28	sends	send	VERB
fcis-26720	5	29	it	it	PRON
fcis-26720	5	30	into	into	ADP
fcis-26720	5	31	the	the	DET
fcis-26720	5	32	capsule	capsule	NOUN
fcis-26720	5	33	network	network	NOUN
fcis-26720	5	34	for	for	ADP
fcis-26720	5	35	deep	deep	ADJ
fcis-26720	5	36	analysis	analysis	NOUN
fcis-26720	5	37	.	.	PUNCT
fcis-26720	6	1	the	the	DET
fcis-26720	6	2	recognition	recognition	NOUN
fcis-26720	6	3	rate	rate	NOUN
fcis-26720	6	4	of	of	ADP
fcis-26720	6	5	81.52	81.52	NUM
fcis-26720	6	6	%	%	NOUN
fcis-26720	6	7	was	be	AUX
fcis-26720	6	8	achieved	achieve	VERB
fcis-26720	6	9	on	on	ADP
fcis-26720	6	10	the	the	DET
fcis-26720	6	11	self	self	NOUN
fcis-26720	6	12	-	-	PUNCT
fcis-26720	6	13	built	build	VERB
fcis-26720	6	14	tibetan	tibetan	ADJ
fcis-26720	6	15	language	language	PROPN
fcis-26720	6	16	emotion	emotion	NOUN
fcis-26720	6	17	corpus	corpus	PROPN
fcis-26720	6	18	tbsec001	tbsec001	PROPN
fcis-26720	6	19	.	.	PUNCT
fcis-26720	7	1	meanwhile	meanwhile	ADV
fcis-26720	7	2	,	,	PUNCT
fcis-26720	7	3	the	the	DET
fcis-26720	7	4	method	method	NOUN
fcis-26720	7	5	achieved	achieve	VERB
fcis-26720	7	6	an	an	DET
fcis-26720	7	7	unweighted	unweighted	ADJ
fcis-26720	7	8	accuracy	accuracy	NOUN
fcis-26720	7	9	(	(	PUNCT
fcis-26720	7	10	ua	ua	NOUN
fcis-26720	7	11	)	)	PUNCT
fcis-26720	7	12	of	of	ADP
fcis-26720	7	13	85.63	85.63	NUM
fcis-26720	7	14	%	%	NOUN
fcis-26720	7	15	and	and	CCONJ
fcis-26720	7	16	95.54	95.54	NUM
fcis-26720	7	17	%	%	NOUN
fcis-26720	7	18	respectively	respectively	ADV
fcis-26720	7	19	on	on	ADP
fcis-26720	7	20	the	the	DET
fcis-26720	7	21	emo	emo	NOUN
fcis-26720	7	22	-	-	PUNCT
fcis-26720	7	23	db	db	NOUN
fcis-26720	7	24	and	and	CCONJ
fcis-26720	7	25	ravdess	ravdess	PROPN
fcis-26720	7	26	public	public	NOUN
fcis-26720	7	27	corpora	corpora	PROPN
fcis-26720	7	28	,	,	PUNCT
fcis-26720	7	29	demonstrating	demonstrate	VERB
fcis-26720	7	30	the	the	DET
fcis-26720	7	31	method	method	NOUN
fcis-26720	7	32	's	's	PART
fcis-26720	7	33	effectiveness	effectiveness	NOUN
fcis-26720	7	34	.	.	PUNCT
fcis-26720	8	1	keywords	keyword	NOUN
fcis-26720	8	2	:	:	PUNCT
fcis-26720	8	3	speech	speech	NOUN
fcis-26720	8	4	emotion	emotion	NOUN
fcis-26720	8	5	recognition	recognition	NOUN
fcis-26720	8	6	;	;	PUNCT
fcis-26720	8	7	tibetan	tibetan	PROPN
fcis-26720	8	8	;	;	PUNCT
fcis-26720	8	9	mel	mel	PROPN
fcis-26720	8	10	-	-	PUNCT
fcis-26720	8	11	frequency	frequency	ADJ
fcis-26720	8	12	cepstral	cepstral	ADJ
fcis-26720	8	13	coefficients	coefficient	NOUN
fcis-26720	8	14	;	;	PUNCT
fcis-26720	8	15	deep	deep	ADJ
fcis-26720	8	16	learning	learning	NOUN
fcis-26720	8	17	;	;	PUNCT
fcis-26720	8	18	capsule	capsule	ADJ
fcis-26720	8	19	net	net	NOUN
fcis-26720	8	20	.	.	PUNCT
fcis-26720	9	1	1	1	X
fcis-26720	9	2	.	.	X
fcis-26720	9	3	introduction	introduction	NOUN
fcis-26720	9	4	as	as	ADP
fcis-26720	9	5	an	an	DET
fcis-26720	9	6	important	important	ADJ
fcis-26720	9	7	carrier	carrier	NOUN
fcis-26720	9	8	in	in	ADP
fcis-26720	9	9	human	human	ADJ
fcis-26720	9	10	communication	communication	NOUN
fcis-26720	9	11	,	,	PUNCT
fcis-26720	9	12	emotion	emotion	NOUN
fcis-26720	9	13	has	have	VERB
fcis-26720	9	14	remarkable	remarkable	ADJ
fcis-26720	9	15	subjectivity	subjectivity	NOUN
fcis-26720	9	16	and	and	CCONJ
fcis-26720	9	17	contextual	contextual	ADJ
fcis-26720	9	18	relevance	relevance	NOUN
fcis-26720	9	19	.	.	PUNCT
fcis-26720	10	1	different	different	ADJ
fcis-26720	10	2	individuals	individual	NOUN
fcis-26720	10	3	may	may	AUX
fcis-26720	10	4	adopt	adopt	VERB
fcis-26720	10	5	different	different	ADJ
fcis-26720	10	6	ways	way	NOUN
fcis-26720	10	7	to	to	PART
fcis-26720	10	8	express	express	VERB
fcis-26720	10	9	and	and	CCONJ
fcis-26720	10	10	interpret	interpret	VERB
fcis-26720	10	11	emotions	emotion	NOUN
fcis-26720	10	12	,	,	PUNCT
fcis-26720	10	13	which	which	PRON
fcis-26720	10	14	reflects	reflect	VERB
fcis-26720	10	15	the	the	DET
fcis-26720	10	16	complexity	complexity	NOUN
fcis-26720	10	17	and	and	CCONJ
fcis-26720	10	18	diversity	diversity	NOUN
fcis-26720	10	19	of	of	ADP
fcis-26720	10	20	emotional	emotional	ADJ
fcis-26720	10	21	communication	communication	NOUN
fcis-26720	10	22	[	[	X
fcis-26720	10	23	1	1	X
fcis-26720	10	24	]	]	PUNCT
fcis-26720	10	25	with	with	ADP
fcis-26720	10	26	the	the	DET
fcis-26720	10	27	continuous	continuous	ADJ
fcis-26720	10	28	progress	progress	NOUN
fcis-26720	10	29	of	of	ADP
fcis-26720	10	30	computer	computer	NOUN
fcis-26720	10	31	technology	technology	NOUN
fcis-26720	10	32	and	and	CCONJ
fcis-26720	10	33	the	the	DET
fcis-26720	10	34	increasingly	increasingly	ADV
fcis-26720	10	35	prominent	prominent	ADJ
fcis-26720	10	36	demand	demand	NOUN
fcis-26720	10	37	of	of	ADP
fcis-26720	10	38	human	human	ADJ
fcis-26720	10	39	-	-	PUNCT
fcis-26720	10	40	computer	computer	NOUN
fcis-26720	10	41	interaction	interaction	NOUN
fcis-26720	10	42	(	(	PUNCT
fcis-26720	10	43	hci	hci	NOUN
fcis-26720	10	44	)	)	PUNCT
fcis-26720	10	45	,	,	PUNCT
fcis-26720	10	46	people	people	NOUN
fcis-26720	10	47	frequently	frequently	ADV
fcis-26720	10	48	communicate	communicate	VERB
fcis-26720	10	49	and	and	CCONJ
fcis-26720	10	50	interact	interact	VERB
fcis-26720	10	51	with	with	ADP
fcis-26720	10	52	computers	computer	NOUN
fcis-26720	10	53	in	in	ADP
fcis-26720	10	54	daily	daily	ADJ
fcis-26720	10	55	life	life	NOUN
fcis-26720	10	56	and	and	CCONJ
fcis-26720	10	57	work	work	NOUN
fcis-26720	10	58	.	.	PUNCT
fcis-26720	11	1	if	if	SCONJ
fcis-26720	11	2	the	the	DET
fcis-26720	11	3	computer	computer	NOUN
fcis-26720	11	4	can	can	AUX
fcis-26720	11	5	automatically	automatically	ADV
fcis-26720	11	6	recognize	recognize	VERB
fcis-26720	11	7	the	the	DET
fcis-26720	11	8	users	user	NOUN
fcis-26720	11	9	emotional	emotional	ADJ
fcis-26720	11	10	state	state	NOUN
fcis-26720	11	11	,	,	PUNCT
fcis-26720	11	12	it	it	PRON
fcis-26720	11	13	can	can	AUX
fcis-26720	11	14	respond	respond	VERB
fcis-26720	11	15	to	to	ADP
fcis-26720	11	16	the	the	DET
fcis-26720	11	17	user	user	NOUN
fcis-26720	11	18	’s	’s	PART
fcis-26720	11	19	needs	need	NOUN
fcis-26720	11	20	in	in	ADP
fcis-26720	11	21	a	a	DET
fcis-26720	11	22	more	more	ADV
fcis-26720	11	23	intelligent	intelligent	ADJ
fcis-26720	11	24	way	way	NOUN
fcis-26720	11	25	,	,	PUNCT
fcis-26720	11	26	thereby	thereby	ADV
fcis-26720	11	27	improving	improve	VERB
fcis-26720	11	28	the	the	DET
fcis-26720	11	29	interactive	interactive	ADJ
fcis-26720	11	30	experience	experience	NOUN
fcis-26720	12	1	[	[	X
fcis-26720	12	2	2	2	NUM
fcis-26720	12	3	]	]	PUNCT
fcis-26720	12	4	.	.	PUNCT
fcis-26720	13	1	the	the	DET
fcis-26720	13	2	process	process	NOUN
fcis-26720	13	3	of	of	ADP
fcis-26720	13	4	automatically	automatically	ADV
fcis-26720	13	5	recognizing	recognize	VERB
fcis-26720	13	6	human	human	ADJ
fcis-26720	13	7	emotions	emotion	NOUN
fcis-26720	13	8	by	by	ADP
fcis-26720	13	9	computers	computer	NOUN
fcis-26720	13	10	is	be	AUX
fcis-26720	13	11	called	call	VERB
fcis-26720	13	12	speech	speech	NOUN
fcis-26720	13	13	emotion	emotion	NOUN
fcis-26720	13	14	recognition	recognition	NOUN
fcis-26720	13	15	(	(	PUNCT
fcis-26720	13	16	ser	ser	NOUN
fcis-26720	13	17	)	)	PUNCT
fcis-26720	13	18	.	.	PUNCT
fcis-26720	14	1	the	the	DET
fcis-26720	14	2	speech	speech	NOUN
fcis-26720	14	3	emotion	emotion	NOUN
fcis-26720	14	4	recognition	recognition	NOUN
fcis-26720	14	5	task	task	NOUN
fcis-26720	14	6	includes	include	VERB
fcis-26720	14	7	many	many	ADJ
fcis-26720	14	8	key	key	ADJ
fcis-26720	14	9	steps	step	NOUN
fcis-26720	14	10	,	,	PUNCT
fcis-26720	14	11	the	the	DET
fcis-26720	14	12	core	core	ADJ
fcis-26720	14	13	parts	part	NOUN
fcis-26720	14	14	of	of	ADP
fcis-26720	14	15	which	which	PRON
fcis-26720	14	16	are	be	AUX
fcis-26720	14	17	:	:	PUNCT
fcis-26720	14	18	building	build	VERB
fcis-26720	14	19	a	a	DET
fcis-26720	14	20	speech	speech	NOUN
fcis-26720	14	21	emotion	emotion	NOUN
fcis-26720	14	22	corpus	corpus	NOUN
fcis-26720	14	23	,	,	PUNCT
fcis-26720	14	24	extracting	extract	VERB
fcis-26720	14	25	speech	speech	NOUN
fcis-26720	14	26	emotion	emotion	NOUN
fcis-26720	14	27	features	feature	NOUN
fcis-26720	14	28	,	,	PUNCT
fcis-26720	14	29	model	model	NOUN
fcis-26720	14	30	training	training	NOUN
fcis-26720	14	31	and	and	CCONJ
fcis-26720	14	32	evaluation	evaluation	NOUN
fcis-26720	14	33	.	.	PUNCT
fcis-26720	15	1	the	the	DET
fcis-26720	15	2	task	task	NOUN
fcis-26720	15	3	first	first	ADV
fcis-26720	15	4	builds	build	VERB
fcis-26720	15	5	a	a	DET
fcis-26720	15	6	corpus	corpus	NOUN
fcis-26720	15	7	by	by	ADP
fcis-26720	15	8	collecting	collect	VERB
fcis-26720	15	9	and	and	CCONJ
fcis-26720	15	10	labeling	label	VERB
fcis-26720	15	11	emotion	emotion	NOUN
fcis-26720	15	12	-	-	PUNCT
fcis-26720	15	13	rich	rich	ADJ
fcis-26720	15	14	speech	speech	NOUN
fcis-26720	15	15	segments	segment	NOUN
fcis-26720	15	16	,	,	PUNCT
fcis-26720	15	17	and	and	CCONJ
fcis-26720	15	18	then	then	ADV
fcis-26720	15	19	extracts	extract	VERB
fcis-26720	15	20	features	feature	NOUN
fcis-26720	15	21	that	that	PRON
fcis-26720	15	22	can	can	AUX
fcis-26720	15	23	reflect	reflect	VERB
fcis-26720	15	24	emotional	emotional	ADJ
fcis-26720	15	25	states	state	NOUN
fcis-26720	15	26	from	from	ADP
fcis-26720	15	27	these	these	DET
fcis-26720	15	28	segments	segment	NOUN
fcis-26720	15	29	,	,	PUNCT
fcis-26720	15	30	sends	send	VERB
fcis-26720	15	31	them	they	PRON
fcis-26720	15	32	into	into	ADP
fcis-26720	15	33	the	the	DET
fcis-26720	15	34	model	model	NOUN
fcis-26720	15	35	as	as	ADP
fcis-26720	15	36	input	input	NOUN
fcis-26720	15	37	for	for	ADP
fcis-26720	15	38	training	training	NOUN
fcis-26720	15	39	and	and	CCONJ
fcis-26720	15	40	testing	testing	NOUN
fcis-26720	15	41	.	.	PUNCT
fcis-26720	16	1	in	in	ADP
fcis-26720	16	2	this	this	DET
fcis-26720	16	3	process	process	NOUN
fcis-26720	16	4	,	,	PUNCT
fcis-26720	16	5	the	the	DET
fcis-26720	16	6	extraction	extraction	NOUN
fcis-26720	16	7	of	of	ADP
fcis-26720	16	8	speech	speech	NOUN
fcis-26720	16	9	emotional	emotional	ADJ
fcis-26720	16	10	features	feature	NOUN
fcis-26720	16	11	and	and	CCONJ
fcis-26720	16	12	the	the	DET
fcis-26720	16	13	creation	creation	NOUN
fcis-26720	16	14	of	of	ADP
fcis-26720	16	15	efficient	efficient	ADJ
fcis-26720	16	16	models	model	NOUN
fcis-26720	16	17	constitute	constitute	VERB
fcis-26720	16	18	the	the	DET
fcis-26720	16	19	main	main	ADJ
fcis-26720	16	20	research	research	NOUN
fcis-26720	16	21	contents	content	NOUN
fcis-26720	16	22	of	of	ADP
fcis-26720	16	23	the	the	DET
fcis-26720	16	24	ser	ser	NOUN
fcis-26720	16	25	task	task	PROPN
fcis-26720	17	1	[	[	X
fcis-26720	17	2	3	3	NUM
fcis-26720	17	3	]	]	PUNCT
fcis-26720	17	4	,	,	PUNCT
fcis-26720	17	5	it	it	PRON
fcis-26720	17	6	is	be	AUX
fcis-26720	17	7	great	great	ADJ
fcis-26720	17	8	significance	significance	NOUN
fcis-26720	17	9	to	to	PART
fcis-26720	17	10	improve	improve	VERB
fcis-26720	17	11	the	the	DET
fcis-26720	17	12	accuracy	accuracy	NOUN
fcis-26720	17	13	and	and	CCONJ
fcis-26720	17	14	robustness	robustness	NOUN
fcis-26720	17	15	of	of	ADP
fcis-26720	17	16	emotion	emotion	NOUN
fcis-26720	17	17	recognition	recognition	NOUN
fcis-26720	17	18	.	.	PUNCT
fcis-26720	18	1	generally	generally	ADV
fcis-26720	18	2	speaking	speak	VERB
fcis-26720	18	3	,	,	PUNCT
fcis-26720	18	4	the	the	DET
fcis-26720	18	5	affective	affective	NOUN
fcis-26720	18	6	features	feature	NOUN
fcis-26720	18	7	of	of	ADP
fcis-26720	18	8	speech	speech	NOUN
fcis-26720	18	9	are	be	AUX
fcis-26720	18	10	divided	divide	VERB
fcis-26720	18	11	into	into	ADP
fcis-26720	18	12	three	three	NUM
fcis-26720	18	13	categories	category	NOUN
fcis-26720	18	14	:	:	PUNCT
fcis-26720	18	15	prosodic	prosodic	NOUN
fcis-26720	18	16	features	feature	VERB
fcis-26720	18	17	[	[	X
fcis-26720	18	18	4	4	NUM
fcis-26720	18	19	]	]	PUNCT
fcis-26720	18	20	,	,	PUNCT
fcis-26720	18	21	spectrum	spectrum	NOUN
fcis-26720	18	22	features	feature	VERB
fcis-26720	18	23	[	[	X
fcis-26720	18	24	5	5	NUM
fcis-26720	18	25	-	-	SYM
fcis-26720	18	26	7	7	NUM
fcis-26720	18	27	]	]	PUNCT
fcis-26720	18	28	and	and	CCONJ
fcis-26720	18	29	sound	sound	ADJ
fcis-26720	18	30	quality	quality	NOUN
fcis-26720	18	31	features	feature	NOUN
fcis-26720	18	32	[	[	X
fcis-26720	18	33	8	8	NUM
fcis-26720	18	34	]	]	PUNCT
fcis-26720	18	35	.	.	PUNCT
fcis-26720	19	1	in	in	ADP
fcis-26720	19	2	the	the	DET
fcis-26720	19	3	early	early	ADJ
fcis-26720	19	4	ser	ser	NOUN
fcis-26720	19	5	task	task	PROPN
fcis-26720	19	6	,	,	PUNCT
fcis-26720	19	7	the	the	DET
fcis-26720	19	8	above	above	ADJ
fcis-26720	19	9	three	three	NUM
fcis-26720	19	10	types	type	NOUN
fcis-26720	19	11	of	of	ADP
fcis-26720	19	12	features	feature	NOUN
fcis-26720	19	13	were	be	AUX
fcis-26720	19	14	mainly	mainly	ADV
fcis-26720	19	15	extracted	extract	VERB
fcis-26720	19	16	as	as	ADP
fcis-26720	19	17	the	the	DET
fcis-26720	19	18	input	input	NOUN
fcis-26720	19	19	of	of	ADP
fcis-26720	19	20	the	the	DET
fcis-26720	19	21	model	model	NOUN
fcis-26720	19	22	,	,	PUNCT
fcis-26720	19	23	which	which	PRON
fcis-26720	19	24	are	be	AUX
fcis-26720	19	25	also	also	ADV
fcis-26720	19	26	called	call	VERB
fcis-26720	19	27	manually	manually	ADV
fcis-26720	19	28	extracted	extract	VERB
fcis-26720	19	29	low	low	ADJ
fcis-26720	19	30	-	-	PUNCT
fcis-26720	19	31	level	level	NOUN
fcis-26720	19	32	descriptors	descriptor	NOUN
fcis-26720	19	33	(	(	PUNCT
fcis-26720	19	34	llds	lld	NOUN
fcis-26720	19	35	)	)	PUNCT
fcis-26720	19	36	.	.	PUNCT
fcis-26720	20	1	in	in	ADP
fcis-26720	20	2	fact	fact	NOUN
fcis-26720	20	3	,	,	PUNCT
fcis-26720	20	4	the	the	DET
fcis-26720	20	5	extraction	extraction	NOUN
fcis-26720	20	6	of	of	ADP
fcis-26720	20	7	these	these	DET
fcis-26720	20	8	features	feature	NOUN
fcis-26720	20	9	involves	involve	VERB
fcis-26720	20	10	a	a	DET
fcis-26720	20	11	lot	lot	NOUN
fcis-26720	20	12	of	of	ADP
fcis-26720	20	13	knowledge	knowledge	NOUN
fcis-26720	20	14	of	of	ADP
fcis-26720	20	15	speech	speech	NOUN
fcis-26720	20	16	signal	signal	NOUN
fcis-26720	20	17	processing	processing	NOUN
fcis-26720	20	18	.	.	PUNCT
fcis-26720	21	1	for	for	ADP
fcis-26720	21	2	researchers	researcher	NOUN
fcis-26720	21	3	,	,	PUNCT
fcis-26720	21	4	designing	design	VERB
fcis-26720	21	5	input	input	NOUN
fcis-26720	21	6	emotional	emotional	ADJ
fcis-26720	21	7	features	feature	NOUN
fcis-26720	21	8	is	be	AUX
fcis-26720	21	9	one	one	NUM
fcis-26720	21	10	of	of	ADP
fcis-26720	21	11	the	the	DET
fcis-26720	21	12	difficulties	difficulty	NOUN
fcis-26720	21	13	of	of	ADP
fcis-26720	21	14	ser	ser	NOUN
fcis-26720	21	15	task	task	PROPN
fcis-26720	21	16	.	.	PUNCT
fcis-26720	22	1	in	in	ADP
fcis-26720	22	2	recent	recent	ADJ
fcis-26720	22	3	years	year	NOUN
fcis-26720	22	4	,	,	PUNCT
fcis-26720	22	5	with	with	ADP
fcis-26720	22	6	the	the	DET
fcis-26720	22	7	improvement	improvement	NOUN
fcis-26720	22	8	of	of	ADP
fcis-26720	22	9	computer	computer	NOUN
fcis-26720	22	10	computing	computing	NOUN
fcis-26720	22	11	power	power	NOUN
fcis-26720	22	12	,	,	PUNCT
fcis-26720	22	13	deep	deep	ADJ
fcis-26720	22	14	learning	learning	NOUN
fcis-26720	22	15	has	have	AUX
fcis-26720	22	16	also	also	ADV
fcis-26720	22	17	risen	rise	VERB
fcis-26720	22	18	.	.	PUNCT
fcis-26720	23	1	the	the	DET
fcis-26720	23	2	proposal	proposal	NOUN
fcis-26720	23	3	of	of	ADP
fcis-26720	23	4	deep	deep	ADJ
fcis-26720	23	5	features	feature	NOUN
fcis-26720	23	6	based	base	VERB
fcis-26720	23	7	on	on	ADP
fcis-26720	23	8	deep	deep	ADJ
fcis-26720	23	9	learning	learning	NOUN
fcis-26720	23	10	has	have	AUX
fcis-26720	23	11	reduced	reduce	VERB
fcis-26720	23	12	the	the	DET
fcis-26720	23	13	difficulty	difficulty	NOUN
fcis-26720	23	14	for	for	SCONJ
fcis-26720	23	15	researchers	researcher	NOUN
fcis-26720	23	16	to	to	PART
fcis-26720	23	17	study	study	VERB
fcis-26720	23	18	speech	speech	NOUN
fcis-26720	23	19	emotion	emotion	NOUN
fcis-26720	23	20	recognition	recognition	NOUN
fcis-26720	23	21	to	to	ADP
fcis-26720	23	22	a	a	DET
fcis-26720	23	23	certain	certain	ADJ
fcis-26720	23	24	extent	extent	NOUN
fcis-26720	23	25	.	.	PUNCT
fcis-26720	24	1	in	in	ADP
fcis-26720	24	2	the	the	DET
fcis-26720	24	3	field	field	NOUN
fcis-26720	24	4	of	of	ADP
fcis-26720	24	5	deep	deep	ADJ
fcis-26720	24	6	learning	learning	NOUN
fcis-26720	24	7	,	,	PUNCT
fcis-26720	24	8	convolutional	convolutional	ADJ
fcis-26720	24	9	neural	neural	ADJ
fcis-26720	24	10	networks	network	NOUN
fcis-26720	24	11	(	(	PUNCT
fcis-26720	24	12	cnn	cnn	PROPN
fcis-26720	24	13	)	)	PUNCT
fcis-26720	24	14	and	and	CCONJ
fcis-26720	24	15	recurrent	recurrent	ADJ
fcis-26720	24	16	neural	neural	ADJ
fcis-26720	24	17	networks	network	NOUN
fcis-26720	24	18	(	(	PUNCT
fcis-26720	24	19	rnn	rnn	PROPN
fcis-26720	24	20	)	)	PUNCT
fcis-26720	24	21	were	be	AUX
fcis-26720	24	22	initially	initially	ADV
fcis-26720	24	23	applied	apply	VERB
fcis-26720	24	24	to	to	ADP
fcis-26720	24	25	speech	speech	NOUN
fcis-26720	24	26	emotion	emotion	NOUN
fcis-26720	24	27	recognition	recognition	NOUN
fcis-26720	24	28	tasks	task	NOUN
fcis-26720	24	29	and	and	CCONJ
fcis-26720	24	30	achieved	achieve	VERB
fcis-26720	24	31	good	good	ADJ
fcis-26720	24	32	results	result	NOUN
fcis-26720	24	33	.	.	PUNCT
fcis-26720	25	1	subsequently	subsequently	ADV
fcis-26720	25	2	,	,	PUNCT
fcis-26720	25	3	long	long	ADJ
fcis-26720	25	4	short	short	ADJ
fcis-26720	25	5	-	-	PUNCT
fcis-26720	25	6	term	term	NOUN
fcis-26720	25	7	memory	memory	NOUN
fcis-26720	25	8	(	(	PUNCT
fcis-26720	25	9	lstm	lstm	NOUN
fcis-26720	25	10	)	)	PUNCT
fcis-26720	25	11	surpasses	surpass	VERB
fcis-26720	25	12	traditional	traditional	ADJ
fcis-26720	25	13	speech	speech	NOUN
fcis-26720	25	14	recognition	recognition	NOUN
fcis-26720	25	15	classifiers	classifier	NOUN
fcis-26720	25	16	in	in	ADP
fcis-26720	25	17	performance	performance	NOUN
fcis-26720	25	18	by	by	ADP
fcis-26720	25	19	virtue	virtue	NOUN
fcis-26720	25	20	of	of	ADP
fcis-26720	25	21	its	its	PRON
fcis-26720	25	22	stronger	strong	ADJ
fcis-26720	25	23	context	context	NOUN
fcis-26720	25	24	association	association	NOUN
fcis-26720	25	25	ability	ability	NOUN
fcis-26720	25	26	[	[	X
fcis-26720	25	27	9	9	NUM
fcis-26720	25	28	]	]	PUNCT
fcis-26720	25	29	.	.	PUNCT
fcis-26720	26	1	in	in	ADP
fcis-26720	26	2	terms	term	NOUN
fcis-26720	26	3	of	of	ADP
fcis-26720	26	4	feature	feature	NOUN
fcis-26720	26	5	extraction	extraction	NOUN
fcis-26720	26	6	,	,	PUNCT
fcis-26720	26	7	researchers	researcher	NOUN
fcis-26720	26	8	began	begin	VERB
fcis-26720	26	9	to	to	PART
fcis-26720	26	10	try	try	VERB
fcis-26720	26	11	feature	feature	NOUN
fcis-26720	26	12	fusion	fusion	NOUN
fcis-26720	26	13	experiments	experiment	NOUN
fcis-26720	26	14	,	,	PUNCT
fcis-26720	26	15	by	by	ADP
fcis-26720	26	16	horizontally	horizontally	ADV
fcis-26720	26	17	stitching	stitching	NOUN
fcis-26720	26	18	llds	lld	NOUN
fcis-26720	26	19	of	of	ADP
fcis-26720	26	20	speech	speech	NOUN
fcis-26720	26	21	signals	signal	NOUN
fcis-26720	26	22	and	and	CCONJ
fcis-26720	26	23	sending	send	VERB
fcis-26720	26	24	them	they	PRON
fcis-26720	26	25	to	to	ADP
fcis-26720	26	26	the	the	DET
fcis-26720	26	27	network	network	NOUN
fcis-26720	26	28	.	.	PUNCT
fcis-26720	27	1	references	reference	NOUN
fcis-26720	27	2	[	[	X
fcis-26720	27	3	10	10	NUM
fcis-26720	27	4	]	]	X
fcis-26720	28	1	[	[	X
fcis-26720	28	2	11	11	NUM
fcis-26720	28	3	]	]	X
fcis-26720	28	4	[	[	X
fcis-26720	28	5	12	12	NUM
fcis-26720	28	6	]	]	PUNCT
fcis-26720	28	7	show	show	VERB
fcis-26720	28	8	that	that	SCONJ
fcis-26720	28	9	these	these	DET
fcis-26720	28	10	fused	fuse	VERB
fcis-26720	28	11	features	feature	NOUN
fcis-26720	28	12	have	have	AUX
fcis-26720	28	13	achieved	achieve	VERB
fcis-26720	28	14	better	well	ADJ
fcis-26720	28	15	performance	performance	NOUN
fcis-26720	28	16	in	in	ADP
fcis-26720	28	17	public	public	ADJ
fcis-26720	28	18	corpus	corpus	NOUN
fcis-26720	28	19	.	.	PUNCT
fcis-26720	29	1	due	due	ADP
fcis-26720	29	2	to	to	ADP
fcis-26720	29	3	the	the	DET
fcis-26720	29	4	problem	problem	NOUN
fcis-26720	29	5	that	that	PRON
fcis-26720	29	6	lstm	lstm	PROPN
fcis-26720	29	7	networks	network	NOUN
fcis-26720	29	8	can	can	AUX
fcis-26720	29	9	not	not	PART
fcis-26720	29	10	parallelize	parallelize	VERB
fcis-26720	29	11	computing	computing	NOUN
fcis-26720	29	12	,	,	PUNCT
fcis-26720	29	13	google	google	PROPN
fcis-26720	29	14	proposed	propose	VERB
fcis-26720	29	15	the	the	DET
fcis-26720	29	16	transformer	transformer	NOUN
fcis-26720	29	17	model	model	NOUN
fcis-26720	29	18	in	in	ADP
fcis-26720	29	19	2017	2017	NUM
fcis-26720	29	20	.	.	PUNCT
fcis-26720	30	1	its	its	PRON
fcis-26720	30	2	advent	advent	NOUN
fcis-26720	30	3	marks	mark	VERB
fcis-26720	30	4	a	a	DET
fcis-26720	30	5	great	great	ADJ
fcis-26720	30	6	success	success	NOUN
fcis-26720	30	7	of	of	ADP
fcis-26720	30	8	attention	attention	NOUN
fcis-26720	30	9	mechanism	mechanism	NOUN
fcis-26720	30	10	in	in	ADP
fcis-26720	30	11	the	the	DET
fcis-26720	30	12	field	field	NOUN
fcis-26720	30	13	of	of	ADP
fcis-26720	30	14	natural	natural	ADJ
fcis-26720	30	15	language	language	NOUN
fcis-26720	30	16	processing	processing	NOUN
fcis-26720	30	17	(	(	PUNCT
fcis-26720	30	18	nlp	nlp	NOUN
fcis-26720	30	19	)	)	PUNCT
fcis-26720	30	20	,	,	PUNCT
fcis-26720	30	21	and	and	CCONJ
fcis-26720	30	22	this	this	DET
fcis-26720	30	23	mechanism	mechanism	NOUN
fcis-26720	30	24	also	also	ADV
fcis-26720	30	25	has	have	VERB
fcis-26720	30	26	significant	significant	ADJ
fcis-26720	30	27	effects	effect	NOUN
fcis-26720	30	28	on	on	ADP
fcis-26720	30	29	ser	ser	NOUN
fcis-26720	30	30	tasks	task	NOUN
fcis-26720	31	1	[	[	X
fcis-26720	31	2	13	13	NUM
fcis-26720	31	3	-	-	SYM
fcis-26720	31	4	15	15	NUM
fcis-26720	31	5	]	]	PUNCT
fcis-26720	31	6	.	.	PUNCT
fcis-26720	32	1	literature	literature	NOUN
fcis-26720	33	1	[	[	X
fcis-26720	33	2	16	16	NUM
fcis-26720	33	3	]	]	PUNCT
fcis-26720	33	4	achieved	achieve	VERB
fcis-26720	33	5	better	well	ADJ
fcis-26720	33	6	classification	classification	NOUN
fcis-26720	33	7	performance	performance	NOUN
fcis-26720	33	8	by	by	ADP
fcis-26720	33	9	combining	combine	VERB
fcis-26720	33	10	attention	attention	NOUN
fcis-26720	33	11	mechanisms	mechanism	NOUN
fcis-26720	33	12	and	and	CCONJ
fcis-26720	33	13	lstm	lstm	NOUN
fcis-26720	33	14	networks	network	NOUN
fcis-26720	33	15	and	and	CCONJ
fcis-26720	33	16	modifying	modify	VERB
fcis-26720	33	17	their	their	PRON
fcis-26720	33	18	forgetting	forget	VERB
fcis-26720	33	19	gates	gate	NOUN
fcis-26720	33	20	.	.	PUNCT
fcis-26720	34	1	furthermore	furthermore	ADV
fcis-26720	34	2	,	,	PUNCT
fcis-26720	34	3	reference	reference	NOUN
fcis-26720	34	4	[	[	X
fcis-26720	34	5	17	17	NUM
fcis-26720	34	6	]	]	PUNCT
fcis-26720	34	7	improved	improve	VERB
fcis-26720	34	8	the	the	DET
fcis-26720	34	9	transformer	transformer	NOUN
fcis-26720	34	10	model	model	NOUN
fcis-26720	34	11	,	,	PUNCT
fcis-26720	34	12	and	and	CCONJ
fcis-26720	34	13	verified	verify	VERB
fcis-26720	34	14	the	the	DET
fcis-26720	34	15	feasibility	feasibility	NOUN
fcis-26720	34	16	and	and	CCONJ
fcis-26720	34	17	effectiveness	effectiveness	NOUN
fcis-26720	34	18	of	of	ADP
fcis-26720	34	19	the	the	DET
fcis-26720	34	20	method	method	NOUN
fcis-26720	34	21	on	on	ADP
fcis-26720	34	22	multiple	multiple	ADJ
fcis-26720	34	23	public	public	ADJ
fcis-26720	34	24	data	data	NOUN
fcis-26720	34	25	sets	set	NOUN
fcis-26720	34	26	by	by	ADP
fcis-26720	34	27	adjusting	adjust	VERB
fcis-26720	34	28	the	the	DET
fcis-26720	34	29	attention	attention	NOUN
fcis-26720	34	30	window	window	NOUN
fcis-26720	34	31	in	in	ADP
fcis-26720	34	32	its	its	PRON
fcis-26720	34	33	multi	multi	ADJ
fcis-26720	34	34	-	-	ADJ
fcis-26720	34	35	head	head	ADJ
fcis-26720	34	36	attention	attention	NOUN
fcis-26720	34	37	mechanism	mechanism	NOUN
fcis-26720	34	38	.	.	PUNCT
fcis-26720	35	1	although	although	SCONJ
fcis-26720	35	2	deep	deep	ADJ
fcis-26720	35	3	learning	learning	NOUN
fcis-26720	35	4	has	have	AUX
fcis-26720	35	5	made	make	VERB
fcis-26720	35	6	remarkable	remarkable	ADJ
fcis-26720	35	7	progress	progress	NOUN
fcis-26720	35	8	in	in	ADP
fcis-26720	35	9	the	the	DET
fcis-26720	35	10	field	field	NOUN
fcis-26720	35	11	of	of	ADP
fcis-26720	35	12	speech	speech	NOUN
fcis-26720	35	13	emotion	emotion	NOUN
fcis-26720	35	14	recognition	recognition	NOUN
fcis-26720	35	15	,	,	PUNCT
fcis-26720	35	16	the	the	DET
fcis-26720	35	17	research	research	NOUN
fcis-26720	35	18	of	of	ADP
fcis-26720	35	19	tibetan	tibetan	ADJ
fcis-26720	35	20	speech	speech	NOUN
fcis-26720	35	21	emotion	emotion	NOUN
fcis-26720	35	22	recognition	recognition	NOUN
fcis-26720	35	23	is	be	AUX
fcis-26720	35	24	still	still	ADV
fcis-26720	35	25	relatively	relatively	ADV
fcis-26720	35	26	lagging	lag	VERB
fcis-26720	35	27	behind	behind	ADV
fcis-26720	35	28	.	.	PUNCT
fcis-26720	36	1	as	as	ADP
fcis-26720	36	2	a	a	DET
fcis-26720	36	3	unique	unique	ADJ
fcis-26720	36	4	minority	minority	NOUN
fcis-26720	36	5	language	language	NOUN
fcis-26720	36	6	,	,	PUNCT
fcis-26720	36	7	tibetan	tibetan	PROPN
fcis-26720	36	8	faces	face	VERB
fcis-26720	36	9	multiple	multiple	ADJ
fcis-26720	36	10	challenges	challenge	NOUN
fcis-26720	36	11	in	in	ADP
fcis-26720	36	12	speech	speech	NOUN
fcis-26720	36	13	emotion	emotion	NOUN
fcis-26720	36	14	recognition	recognition	NOUN
fcis-26720	36	15	,	,	PUNCT
fcis-26720	36	16	including	include	VERB
fcis-26720	36	17	data	datum	NOUN
fcis-26720	36	18	scarcity	scarcity	NOUN
fcis-26720	36	19	,	,	PUNCT
fcis-26720	36	20	labeling	labeling	NOUN
fcis-26720	36	21	difficulties	difficulty	NOUN
fcis-26720	36	22	and	and	CCONJ
fcis-26720	36	23	differences	difference	NOUN
fcis-26720	36	24	in	in	ADP
fcis-26720	36	25	emotion	emotion	NOUN
fcis-26720	36	26	expression	expression	NOUN
fcis-26720	36	27	habits	habit	NOUN
fcis-26720	36	28	[	[	X
fcis-26720	36	29	18	18	NUM
fcis-26720	36	30	]	]	PUNCT
fcis-26720	36	31	.	.	PUNCT
fcis-26720	37	1	although	although	SCONJ
fcis-26720	37	2	some	some	DET
fcis-26720	37	3	researchers	researcher	NOUN
fcis-26720	37	4	have	have	AUX
fcis-26720	37	5	been	be	AUX
fcis-26720	37	6	involved	involve	VERB
fcis-26720	37	7	in	in	ADP
fcis-26720	37	8	the	the	DET
fcis-26720	37	9	field	field	NOUN
fcis-26720	37	10	of	of	ADP
fcis-26720	37	11	tibetan	tibetan	ADJ
fcis-26720	37	12	speech	speech	NOUN
fcis-26720	37	13	emotion	emotion	NOUN
fcis-26720	37	14	recognition	recognition	NOUN
fcis-26720	37	15	,	,	PUNCT
fcis-26720	37	16	they	they	PRON
fcis-26720	37	17	mainly	mainly	ADV
fcis-26720	37	18	fuse	fuse	VERB
fcis-26720	37	19	multiple	multiple	ADJ
fcis-26720	37	20	llds	lld	NOUN
fcis-26720	37	21	.	.	PUNCT
fcis-26720	38	1	however	however	ADV
fcis-26720	38	2	,	,	PUNCT
fcis-26720	38	3	these	these	DET
fcis-26720	38	4	methods	method	NOUN
fcis-26720	38	5	have	have	VERB
fcis-26720	38	6	some	some	DET
fcis-26720	38	7	problems	problem	NOUN
fcis-26720	38	8	,	,	PUNCT
fcis-26720	38	9	such	such	ADJ
fcis-26720	38	10	as	as	ADP
fcis-26720	38	11	too	too	ADV
fcis-26720	38	12	high	high	ADJ
fcis-26720	38	13	emotional	emotional	ADJ
fcis-26720	38	14	feature	feature	NOUN
fcis-26720	38	15	dimension	dimension	NOUN
fcis-26720	38	16	,	,	PUNCT
fcis-26720	38	17	too	too	ADV
fcis-26720	38	18	cumbersome	cumbersome	ADJ
fcis-26720	38	19	emotional	emotional	ADJ
fcis-26720	38	20	feature	feature	NOUN
fcis-26720	38	21	extraction	extraction	NOUN
fcis-26720	38	22	,	,	PUNCT
fcis-26720	38	23	complex	complex	ADJ
fcis-26720	38	24	network	network	NOUN
fcis-26720	38	25	design	design	NOUN
fcis-26720	38	26	and	and	CCONJ
fcis-26720	38	27	high	high	ADJ
fcis-26720	38	28	computing	computing	NOUN
fcis-26720	38	29	power	power	NOUN
fcis-26720	38	30	consumption	consumption	NOUN
fcis-26720	38	31	[	[	X
fcis-26720	38	32	19	19	NUM
fcis-26720	38	33	]	]	PUNCT
fcis-26720	38	34	.	.	PUNCT
fcis-26720	39	1	to	to	PART
fcis-26720	39	2	effectively	effectively	ADV
fcis-26720	39	3	address	address	VERB
fcis-26720	39	4	these	these	DET
fcis-26720	39	5	challenges	challenge	NOUN
fcis-26720	39	6	,	,	PUNCT
fcis-26720	39	7	an	an	DET
fcis-26720	39	8	86	86	NUM
fcis-26720	39	9	innovative	innovative	ADJ
fcis-26720	39	10	lightweight	lightweight	ADJ
fcis-26720	39	11	network	network	NOUN
fcis-26720	39	12	model	model	NOUN
fcis-26720	39	13	is	be	AUX
fcis-26720	39	14	proposed	propose	VERB
fcis-26720	39	15	in	in	ADP
fcis-26720	39	16	this	this	DET
fcis-26720	39	17	paper	paper	NOUN
fcis-26720	39	18	.	.	PUNCT
fcis-26720	40	1	this	this	DET
fcis-26720	40	2	model	model	NOUN
fcis-26720	40	3	combines	combine	VERB
fcis-26720	40	4	the	the	DET
fcis-26720	40	5	spatiotemporal	spatiotemporal	ADJ
fcis-26720	40	6	feature	feature	NOUN
fcis-26720	40	7	extraction	extraction	NOUN
fcis-26720	40	8	of	of	ADP
fcis-26720	40	9	mel	mel	PROPN
fcis-26720	40	10	-	-	PUNCT
fcis-26720	40	11	frequency	frequency	ADJ
fcis-26720	40	12	cepstral	cepstral	ADJ
fcis-26720	40	13	coefficients	coefficient	NOUN
fcis-26720	40	14	(	(	PUNCT
fcis-26720	40	15	mfcc	mfcc	NOUN
fcis-26720	40	16	)	)	PUNCT
fcis-26720	40	17	with	with	ADP
fcis-26720	40	18	the	the	DET
fcis-26720	40	19	powerful	powerful	ADJ
fcis-26720	40	20	vectorization	vectorization	NOUN
fcis-26720	40	21	capabilities	capability	NOUN
fcis-26720	40	22	of	of	ADP
fcis-26720	40	23	capsule	capsule	NOUN
fcis-26720	40	24	networks	network	NOUN
fcis-26720	40	25	,	,	PUNCT
fcis-26720	40	26	aiming	aim	VERB
fcis-26720	40	27	to	to	PART
fcis-26720	40	28	achieve	achieve	VERB
fcis-26720	40	29	accurate	accurate	ADJ
fcis-26720	40	30	recognition	recognition	NOUN
fcis-26720	40	31	of	of	ADP
fcis-26720	40	32	tibetan	tibetan	ADJ
fcis-26720	40	33	speech	speech	NOUN
fcis-26720	40	34	emotions	emotion	NOUN
fcis-26720	40	35	.	.	PUNCT
fcis-26720	41	1	in	in	ADP
fcis-26720	41	2	order	order	NOUN
fcis-26720	41	3	to	to	PART
fcis-26720	41	4	comprehensively	comprehensively	ADV
fcis-26720	41	5	evaluate	evaluate	VERB
fcis-26720	41	6	and	and	CCONJ
fcis-26720	41	7	verify	verify	VERB
fcis-26720	41	8	the	the	DET
fcis-26720	41	9	performance	performance	NOUN
fcis-26720	41	10	of	of	ADP
fcis-26720	41	11	the	the	DET
fcis-26720	41	12	model	model	NOUN
fcis-26720	41	13	,	,	PUNCT
fcis-26720	41	14	this	this	DET
fcis-26720	41	15	paper	paper	NOUN
fcis-26720	41	16	conducts	conduct	VERB
fcis-26720	41	17	experiments	experiment	NOUN
fcis-26720	41	18	in	in	ADP
fcis-26720	41	19	emo	emo	NOUN
fcis-26720	41	20	-	-	PUNCT
fcis-26720	41	21	db	db	NOUN
fcis-26720	41	22	,	,	PUNCT
fcis-26720	41	23	ravdess	ravdess	NOUN
fcis-26720	41	24	corpus	corpus	NOUN
fcis-26720	41	25	,	,	PUNCT
fcis-26720	41	26	and	and	CCONJ
fcis-26720	41	27	the	the	DET
fcis-26720	41	28	selfbuilt	selfbuilt	ADJ
fcis-26720	41	29	tibetan	tibetan	ADJ
fcis-26720	41	30	speech	speech	NOUN
fcis-26720	41	31	emotion	emotion	NOUN
fcis-26720	41	32	corpus	corpus	PROPN
fcis-26720	41	33	tbsec001	tbsec001	PROPN
fcis-26720	41	34	,	,	PUNCT
fcis-26720	41	35	and	and	CCONJ
fcis-26720	41	36	finally	finally	ADV
fcis-26720	41	37	achieves	achieve	VERB
fcis-26720	41	38	satisfactory	satisfactory	ADJ
fcis-26720	41	39	classification	classification	NOUN
fcis-26720	41	40	results	result	NOUN
fcis-26720	41	41	.	.	PUNCT
fcis-26720	42	1	the	the	DET
fcis-26720	42	2	main	main	ADJ
fcis-26720	42	3	research	research	NOUN
fcis-26720	42	4	contents	content	NOUN
fcis-26720	42	5	of	of	ADP
fcis-26720	42	6	this	this	DET
fcis-26720	42	7	paper	paper	NOUN
fcis-26720	42	8	are	be	AUX
fcis-26720	42	9	as	as	SCONJ
fcis-26720	42	10	follows	follow	VERB
fcis-26720	42	11	:	:	PUNCT
fcis-26720	42	12	(	(	PUNCT
fcis-26720	42	13	1	1	X
fcis-26720	42	14	)	)	PUNCT
fcis-26720	42	15	a	a	DET
fcis-26720	42	16	tibetan	tibetan	ADJ
fcis-26720	42	17	speech	speech	NOUN
fcis-26720	42	18	emotion	emotion	NOUN
fcis-26720	42	19	corpus	corpus	PROPN
fcis-26720	42	20	tbsec001	tbsec001	PUNCT
fcis-26720	42	21	is	be	AUX
fcis-26720	42	22	constructed	construct	VERB
fcis-26720	42	23	and	and	CCONJ
fcis-26720	42	24	improved	improve	VERB
fcis-26720	42	25	,	,	PUNCT
fcis-26720	42	26	which	which	PRON
fcis-26720	42	27	provides	provide	VERB
fcis-26720	42	28	data	datum	NOUN
fcis-26720	42	29	resources	resource	NOUN
fcis-26720	42	30	for	for	ADP
fcis-26720	42	31	related	related	ADJ
fcis-26720	42	32	research	research	NOUN
fcis-26720	42	33	.	.	PUNCT
fcis-26720	43	1	(	(	PUNCT
fcis-26720	43	2	2	2	X
fcis-26720	43	3	)	)	PUNCT
fcis-26720	43	4	a	a	DET
fcis-26720	43	5	lightweight	lightweight	ADJ
fcis-26720	43	6	network	network	NOUN
fcis-26720	43	7	model	model	NOUN
fcis-26720	43	8	is	be	AUX
fcis-26720	43	9	proposed	propose	VERB
fcis-26720	43	10	,	,	PUNCT
fcis-26720	43	11	which	which	PRON
fcis-26720	43	12	fuses	fuse	VERB
fcis-26720	43	13	the	the	DET
fcis-26720	43	14	spatiotemporal	spatiotemporal	ADJ
fcis-26720	43	15	features	feature	NOUN
fcis-26720	43	16	of	of	ADP
fcis-26720	43	17	mfcc	mfcc	NOUN
fcis-26720	43	18	coefficients	coefficient	NOUN
fcis-26720	43	19	and	and	CCONJ
fcis-26720	43	20	the	the	DET
fcis-26720	43	21	vectorization	vectorization	NOUN
fcis-26720	43	22	capability	capability	NOUN
fcis-26720	43	23	of	of	ADP
fcis-26720	43	24	capsule	capsule	NOUN
fcis-26720	43	25	networks	network	NOUN
fcis-26720	43	26	.	.	PUNCT
fcis-26720	44	1	the	the	DET
fcis-26720	44	2	self	self	NOUN
fcis-26720	44	3	-	-	PUNCT
fcis-26720	44	4	built	build	VERB
fcis-26720	44	5	corpus	corpus	NOUN
fcis-26720	44	6	tbsec001	tbsec001	NUM
fcis-26720	44	7	was	be	AUX
fcis-26720	44	8	used	use	VERB
fcis-26720	44	9	for	for	ADP
fcis-26720	44	10	tibetan	tibetan	ADJ
fcis-26720	44	11	speech	speech	NOUN
fcis-26720	44	12	emotion	emotion	NOUN
fcis-26720	44	13	recognition	recognition	NOUN
fcis-26720	44	14	,	,	PUNCT
fcis-26720	44	15	and	and	CCONJ
fcis-26720	44	16	the	the	DET
fcis-26720	44	17	accuracy	accuracy	NOUN
fcis-26720	44	18	was	be	AUX
fcis-26720	44	19	81.52	81.52	NUM
fcis-26720	44	20	%	%	NOUN
fcis-26720	44	21	,	,	PUNCT
fcis-26720	44	22	and	and	CCONJ
fcis-26720	44	23	the	the	DET
fcis-26720	44	24	unweighted	unweighted	ADJ
fcis-26720	44	25	average	average	ADJ
fcis-26720	44	26	accuracy	accuracy	NOUN
fcis-26720	44	27	(	(	PUNCT
fcis-26720	44	28	ua	ua	NOUN
fcis-26720	44	29	)	)	PUNCT
fcis-26720	44	30	was	be	AUX
fcis-26720	44	31	85.63	85.63	NUM
fcis-26720	44	32	%	%	NOUN
fcis-26720	44	33	and	and	CCONJ
fcis-26720	44	34	95.54	95.54	NUM
fcis-26720	44	35	%	%	NOUN
fcis-26720	44	36	on	on	ADP
fcis-26720	44	37	emo	emo	NOUN
fcis-26720	44	38	-	-	PUNCT
fcis-26720	44	39	db	db	NOUN
fcis-26720	44	40	and	and	CCONJ
fcis-26720	44	41	ravdess	ravdess	NOUN
fcis-26720	44	42	corpus	corpus	PROPN
fcis-26720	44	43	,	,	PUNCT
fcis-26720	44	44	respectively	respectively	ADV
fcis-26720	44	45	.	.	PUNCT
fcis-26720	45	1	2	2	X
fcis-26720	45	2	.	.	X
fcis-26720	45	3	model	model	NOUN
fcis-26720	45	4	design	design	VERB
fcis-26720	45	5	the	the	DET
fcis-26720	45	6	network	network	NOUN
fcis-26720	45	7	model	model	NOUN
fcis-26720	45	8	structure	structure	NOUN
fcis-26720	45	9	proposed	propose	VERB
fcis-26720	45	10	in	in	ADP
fcis-26720	45	11	this	this	DET
fcis-26720	45	12	paper	paper	NOUN
fcis-26720	45	13	is	be	AUX
fcis-26720	45	14	shown	show	VERB
fcis-26720	45	15	in	in	ADP
fcis-26720	45	16	figure	figure	NOUN
fcis-26720	45	17	1	1	NUM
fcis-26720	45	18	.	.	PUNCT
fcis-26720	46	1	it	it	PRON
fcis-26720	46	2	is	be	AUX
fcis-26720	46	3	mainly	mainly	ADV
fcis-26720	46	4	composed	compose	VERB
fcis-26720	46	5	of	of	ADP
fcis-26720	46	6	four	four	NUM
fcis-26720	46	7	parts	part	NOUN
fcis-26720	46	8	:	:	PUNCT
fcis-26720	46	9	mfcc	mfcc	NOUN
fcis-26720	46	10	(	(	PUNCT
fcis-26720	46	11	input	input	NOUN
fcis-26720	46	12	feature	feature	NOUN
fcis-26720	46	13	)	)	PUNCT
fcis-26720	46	14	,	,	PUNCT
fcis-26720	46	15	spatiotemporal	spatiotemporal	ADJ
fcis-26720	46	16	feature	feature	NOUN
fcis-26720	46	17	extraction	extraction	NOUN
fcis-26720	46	18	backbone	backbone	NOUN
fcis-26720	46	19	network	network	NOUN
fcis-26720	46	20	(	(	PUNCT
fcis-26720	46	21	stfeatcnn-4l	stfeatcnn-4l	NOUN
fcis-26720	46	22	)	)	PUNCT
fcis-26720	46	23	,	,	PUNCT
fcis-26720	46	24	capsule	capsule	NOUN
fcis-26720	46	25	network	network	NOUN
fcis-26720	46	26	layer	layer	NOUN
fcis-26720	46	27	and	and	CCONJ
fcis-26720	46	28	margin	margin	NOUN
fcis-26720	46	29	-	-	PUNCT
fcis-26720	46	30	loss	loss	NOUN
fcis-26720	46	31	(	(	PUNCT
fcis-26720	46	32	loss	loss	NOUN
fcis-26720	46	33	function	function	NOUN
fcis-26720	46	34	)	)	PUNCT
fcis-26720	46	35	.	.	PUNCT
fcis-26720	47	1	fig	fig	NOUN
fcis-26720	47	2	1	1	NUM
fcis-26720	47	3	.	.	PUNCT
fcis-26720	48	1	overall	overall	ADJ
fcis-26720	48	2	structure	structure	NOUN
fcis-26720	48	3	of	of	ADP
fcis-26720	48	4	the	the	DET
fcis-26720	48	5	model	model	NOUN
fcis-26720	48	6	first	first	ADV
fcis-26720	48	7	,	,	PUNCT
fcis-26720	48	8	the	the	DET
fcis-26720	48	9	model	model	NOUN
fcis-26720	48	10	receives	receive	VERB
fcis-26720	48	11	and	and	CCONJ
fcis-26720	48	12	processes	process	VERB
fcis-26720	48	13	the	the	DET
fcis-26720	48	14	mfcc	mfcc	NOUN
fcis-26720	48	15	as	as	ADP
fcis-26720	48	16	features	feature	NOUN
fcis-26720	48	17	,	,	PUNCT
fcis-26720	48	18	and	and	CCONJ
fcis-26720	48	19	these	these	DET
fcis-26720	48	20	features	feature	NOUN
fcis-26720	48	21	are	be	AUX
fcis-26720	48	22	sent	send	VERB
fcis-26720	48	23	to	to	ADP
fcis-26720	48	24	the	the	DET
fcis-26720	48	25	spatiotemporal	spatiotemporal	ADJ
fcis-26720	48	26	feature	feature	NOUN
fcis-26720	48	27	extraction	extraction	NOUN
fcis-26720	48	28	backbone	backbone	NOUN
fcis-26720	48	29	network	network	NOUN
fcis-26720	48	30	stfeatcnn-4l	stfeatcnn-4l	NUM
fcis-26720	48	31	to	to	PART
fcis-26720	48	32	extract	extract	VERB
fcis-26720	48	33	spatiotemporal	spatiotemporal	ADJ
fcis-26720	48	34	information	information	NOUN
fcis-26720	48	35	;	;	PUNCT
fcis-26720	48	36	then	then	ADV
fcis-26720	48	37	,	,	PUNCT
fcis-26720	48	38	the	the	DET
fcis-26720	48	39	data	data	NOUN
fcis-26720	48	40	is	be	AUX
fcis-26720	48	41	transmitted	transmit	VERB
fcis-26720	48	42	to	to	ADP
fcis-26720	48	43	the	the	DET
fcis-26720	48	44	capsule	capsule	NOUN
fcis-26720	48	45	layer	layer	NOUN
fcis-26720	48	46	.	.	PUNCT
fcis-26720	49	1	with	with	ADP
fcis-26720	49	2	its	its	PRON
fcis-26720	49	3	unique	unique	ADJ
fcis-26720	49	4	dynamic	dynamic	ADJ
fcis-26720	49	5	routing	routing	NOUN
fcis-26720	49	6	update	update	NOUN
fcis-26720	49	7	algorithm	algorithm	NOUN
fcis-26720	49	8	,	,	PUNCT
fcis-26720	49	9	the	the	DET
fcis-26720	49	10	capsule	capsule	NOUN
fcis-26720	49	11	network	network	NOUN
fcis-26720	49	12	layer	layer	NOUN
fcis-26720	49	13	efficiently	efficiently	ADV
fcis-26720	49	14	processes	process	VERB
fcis-26720	49	15	the	the	DET
fcis-26720	49	16	input	input	NOUN
fcis-26720	49	17	features	feature	NOUN
fcis-26720	49	18	and	and	CCONJ
fcis-26720	49	19	outputs	output	VERB
fcis-26720	49	20	the	the	DET
fcis-26720	49	21	prediction	prediction	NOUN
fcis-26720	49	22	results	result	VERB
fcis-26720	49	23	for	for	ADP
fcis-26720	49	24	emotion	emotion	NOUN
fcis-26720	49	25	;	;	PUNCT
fcis-26720	49	26	finally	finally	ADV
fcis-26720	49	27	,	,	PUNCT
fcis-26720	49	28	margin	margin	NOUN
fcis-26720	49	29	-	-	PUNCT
fcis-26720	49	30	loss	loss	NOUN
fcis-26720	49	31	is	be	AUX
fcis-26720	49	32	used	use	VERB
fcis-26720	49	33	as	as	ADP
fcis-26720	49	34	the	the	DET
fcis-26720	49	35	loss	loss	NOUN
fcis-26720	49	36	function	function	NOUN
fcis-26720	49	37	to	to	PART
fcis-26720	49	38	guide	guide	VERB
fcis-26720	49	39	the	the	DET
fcis-26720	49	40	model	model	NOUN
fcis-26720	49	41	to	to	PART
fcis-26720	49	42	adjust	adjust	VERB
fcis-26720	49	43	parameters	parameter	NOUN
fcis-26720	49	44	to	to	PART
fcis-26720	49	45	optimize	optimize	VERB
fcis-26720	49	46	the	the	DET
fcis-26720	49	47	performance	performance	NOUN
fcis-26720	49	48	of	of	ADP
fcis-26720	49	49	the	the	DET
fcis-26720	49	50	model	model	NOUN
fcis-26720	49	51	,	,	PUNCT
fcis-26720	49	52	so	so	SCONJ
fcis-26720	49	53	as	as	SCONJ
fcis-26720	49	54	to	to	PART
fcis-26720	49	55	complete	complete	VERB
fcis-26720	49	56	the	the	DET
fcis-26720	49	57	task	task	NOUN
fcis-26720	49	58	of	of	ADP
fcis-26720	49	59	efficient	efficient	ADJ
fcis-26720	49	60	and	and	CCONJ
fcis-26720	49	61	accurate	accurate	ADJ
fcis-26720	49	62	speech	speech	NOUN
fcis-26720	49	63	emotion	emotion	NOUN
fcis-26720	49	64	recognition	recognition	NOUN
fcis-26720	49	65	.	.	PUNCT
fcis-26720	50	1	table	table	NOUN
fcis-26720	50	2	1	1	NUM
fcis-26720	50	3	.	.	PUNCT
fcis-26720	51	1	parameter	parameter	NOUN
fcis-26720	51	2	configuration	configuration	NOUN
fcis-26720	51	3	of	of	ADP
fcis-26720	51	4	stfeatcnn-4l	stfeatcnn-4l	NOUN
fcis-26720	51	5	layer	layer	NOUN
fcis-26720	51	6	layer	layer	NOUN
fcis-26720	51	7	name	name	NOUN
fcis-26720	51	8	components	component	NOUN
fcis-26720	51	9	kernel	kernel	PROPN
fcis-26720	51	10	size	size	NOUN
fcis-26720	51	11	stride	stride	ADJ
fcis-26720	51	12	input	input	NOUN
fcis-26720	51	13	channel	channel	NOUN
fcis-26720	51	14	output	output	NOUN
fcis-26720	51	15	channel	channel	NOUN
fcis-26720	51	16	conv1	conv1	ADP
fcis-26720	51	17	convolutional	convolutional	ADJ
fcis-26720	51	18	layer	layer	NOUN
fcis-26720	51	19	3×3	3×3	NUM
fcis-26720	51	20	1×1	1×1	NUM
fcis-26720	51	21	1	1	NUM
fcis-26720	51	22	96	96	NUM
fcis-26720	51	23	bn	bn	NOUN
fcis-26720	51	24	(	(	PUNCT
fcis-26720	51	25	batch	batch	NOUN
fcis-26720	51	26	normalization	normalization	NOUN
fcis-26720	51	27	)	)	PUNCT
fcis-26720	52	1	+	+	CCONJ
fcis-26720	52	2	relu	relu	NOUN
fcis-26720	52	3	max	max	NOUN
fcis-26720	52	4	-	-	PUNCT
fcis-26720	52	5	pooling	pool	VERB
fcis-26720	52	6	2×2	2×2	NUM
fcis-26720	52	7	1×1	1×1	NUM
fcis-26720	52	8	dropout	dropout	NOUN
fcis-26720	52	9	conv2	conv2	NOUN
fcis-26720	52	10	convolutional	convolutional	ADJ
fcis-26720	52	11	layer	layer	NOUN
fcis-26720	52	12	3×3	3×3	NUM
fcis-26720	52	13	1×1	1×1	NUM
fcis-26720	52	14	96	96	NUM
fcis-26720	52	15	96	96	NUM
fcis-26720	52	16	bn	bn	NOUN
fcis-26720	52	17	+	+	NUM
fcis-26720	52	18	relu	relu	NOUN
fcis-26720	52	19	max	max	PROPN
fcis-26720	52	20	-	-	PUNCT
fcis-26720	52	21	pooling	pool	VERB
fcis-26720	52	22	2×2	2×2	NUM
fcis-26720	52	23	1×1	1×1	NUM
fcis-26720	52	24	dropout	dropout	NOUN
fcis-26720	52	25	conv3	conv3	NOUN
fcis-26720	52	26	convolutional	convolutional	ADJ
fcis-26720	52	27	layer	layer	NOUN
fcis-26720	52	28	3×3	3×3	NUM
fcis-26720	52	29	1×1	1×1	NUM
fcis-26720	52	30	96	96	NUM
fcis-26720	52	31	96	96	NUM
fcis-26720	52	32	bn	bn	NOUN
fcis-26720	53	1	+	+	NUM
fcis-26720	53	2	relu	relu	NOUN
fcis-26720	53	3	max	max	PROPN
fcis-26720	53	4	-	-	PUNCT
fcis-26720	53	5	pooling	pool	VERB
fcis-26720	53	6	2×2	2×2	NUM
fcis-26720	53	7	1×1	1×1	NUM
fcis-26720	53	8	dropout	dropout	NOUN
fcis-26720	53	9	conv4	conv4	NOUN
fcis-26720	53	10	convolutional	convolutional	ADJ
fcis-26720	53	11	layer	layer	NOUN
fcis-26720	53	12	3×3	3×3	NUM
fcis-26720	53	13	1×1	1×1	NUM
fcis-26720	53	14	96	96	NUM
fcis-26720	53	15	96	96	NUM
fcis-26720	53	16	×	×	NOUN
fcis-26720	53	17	emotion	emotion	NOUN
fcis-26720	53	18	types	type	NOUN
fcis-26720	53	19	relu	relu	NOUN
fcis-26720	53	20	(	(	PUNCT
fcis-26720	53	21	1	1	X
fcis-26720	53	22	)	)	PUNCT
fcis-26720	53	23	backbone	backbone	NOUN
fcis-26720	53	24	network	network	NOUN
fcis-26720	53	25	stfeatcnn-4l	stfeatcnn-4l	VERB
fcis-26720	53	26	in	in	ADP
fcis-26720	53	27	order	order	NOUN
fcis-26720	53	28	to	to	PART
fcis-26720	53	29	extract	extract	VERB
fcis-26720	53	30	the	the	DET
fcis-26720	53	31	spatiotemporal	spatiotemporal	ADJ
fcis-26720	53	32	features	feature	NOUN
fcis-26720	53	33	of	of	ADP
fcis-26720	53	34	mfcc	mfcc	NOUN
fcis-26720	53	35	,	,	PUNCT
fcis-26720	53	36	the	the	DET
fcis-26720	53	37	backbone	backbone	NOUN
fcis-26720	53	38	network	network	NOUN
fcis-26720	53	39	stfeatcnn-4l	stfeatcnn-4l	PROPN
fcis-26720	53	40	is	be	AUX
fcis-26720	53	41	designed	design	VERB
fcis-26720	53	42	in	in	ADP
fcis-26720	53	43	this	this	DET
fcis-26720	53	44	paper	paper	NOUN
fcis-26720	53	45	.	.	PUNCT
fcis-26720	54	1	in	in	ADP
fcis-26720	54	2	stfeatcnn-4l	stfeatcnn-4l	PROPN
fcis-26720	54	3	,	,	PUNCT
fcis-26720	54	4	the	the	DET
fcis-26720	54	5	scaling	scale	VERB
fcis-26720	54	6	invariance	invariance	NOUN
fcis-26720	54	7	of	of	ADP
fcis-26720	54	8	cnn	cnn	PROPN
fcis-26720	54	9	is	be	AUX
fcis-26720	54	10	adopted	adopt	VERB
fcis-26720	54	11	to	to	PART
fcis-26720	54	12	capture	capture	VERB
fcis-26720	54	13	and	and	CCONJ
fcis-26720	54	14	retain	retain	VERB
fcis-26720	54	15	the	the	DET
fcis-26720	54	16	key	key	ADJ
fcis-26720	54	17	features	feature	NOUN
fcis-26720	54	18	among	among	ADP
fcis-26720	54	19	the	the	DET
fcis-26720	54	20	input	input	NOUN
fcis-26720	54	21	features	feature	VERB
fcis-26720	54	22	[	[	X
fcis-26720	54	23	20	20	NUM
fcis-26720	54	24	]	]	PUNCT
fcis-26720	54	25	,	,	PUNCT
fcis-26720	54	26	while	while	SCONJ
fcis-26720	54	27	multiple	multiple	ADJ
fcis-26720	54	28	convolutional	convolutional	ADJ
fcis-26720	54	29	filters	filter	NOUN
fcis-26720	54	30	are	be	AUX
fcis-26720	54	31	used	use	VERB
fcis-26720	54	32	.	.	PUNCT
fcis-26720	55	1	87	87	NUM
fcis-26720	55	2	these	these	DET
fcis-26720	55	3	filters	filter	NOUN
fcis-26720	55	4	are	be	AUX
fcis-26720	55	5	able	able	ADJ
fcis-26720	55	6	to	to	PART
fcis-26720	55	7	learn	learn	VERB
fcis-26720	55	8	and	and	CCONJ
fcis-26720	55	9	extract	extract	VERB
fcis-26720	55	10	different	different	ADJ
fcis-26720	55	11	spatial	spatial	ADJ
fcis-26720	55	12	features	feature	NOUN
fcis-26720	55	13	in	in	ADP
fcis-26720	55	14	mfcc	mfcc	NOUN
fcis-26720	55	15	,	,	PUNCT
fcis-26720	55	16	and	and	CCONJ
fcis-26720	55	17	by	by	ADP
fcis-26720	55	18	combining	combine	VERB
fcis-26720	55	19	spatiotemporal	spatiotemporal	ADJ
fcis-26720	55	20	features	feature	NOUN
fcis-26720	55	21	,	,	PUNCT
fcis-26720	55	22	the	the	DET
fcis-26720	55	23	network	network	NOUN
fcis-26720	55	24	can	can	AUX
fcis-26720	55	25	understand	understand	VERB
fcis-26720	55	26	the	the	DET
fcis-26720	55	27	spatial	spatial	ADJ
fcis-26720	55	28	structure	structure	NOUN
fcis-26720	55	29	of	of	ADP
fcis-26720	55	30	the	the	DET
fcis-26720	55	31	speech	speech	NOUN
fcis-26720	55	32	signal	signal	NOUN
fcis-26720	55	33	more	more	ADV
fcis-26720	55	34	comprehensively	comprehensively	ADV
fcis-26720	55	35	[	[	X
fcis-26720	55	36	21	21	NUM
fcis-26720	55	37	,	,	PUNCT
fcis-26720	55	38	22	22	NUM
fcis-26720	55	39	]	]	PUNCT
fcis-26720	55	40	,	,	PUNCT
fcis-26720	55	41	thus	thus	ADV
fcis-26720	55	42	improving	improve	VERB
fcis-26720	55	43	the	the	DET
fcis-26720	55	44	recognition	recognition	NOUN
fcis-26720	55	45	performance	performance	NOUN
fcis-26720	55	46	.	.	PUNCT
fcis-26720	56	1	in	in	ADP
fcis-26720	56	2	addition	addition	NOUN
fcis-26720	56	3	,	,	PUNCT
fcis-26720	56	4	batch	batch	NOUN
fcis-26720	56	5	normalization	normalization	NOUN
fcis-26720	56	6	is	be	AUX
fcis-26720	56	7	introduced	introduce	VERB
fcis-26720	56	8	in	in	ADP
fcis-26720	56	9	each	each	DET
fcis-26720	56	10	layer	layer	NOUN
fcis-26720	56	11	of	of	ADP
fcis-26720	56	12	the	the	DET
fcis-26720	56	13	network	network	NOUN
fcis-26720	56	14	to	to	PART
fcis-26720	56	15	reduce	reduce	VERB
fcis-26720	56	16	the	the	DET
fcis-26720	56	17	internal	internal	ADJ
fcis-26720	56	18	covariate	covariate	ADJ
fcis-26720	56	19	shift	shift	NOUN
fcis-26720	56	20	and	and	CCONJ
fcis-26720	56	21	stabilize	stabilize	VERB
fcis-26720	56	22	the	the	DET
fcis-26720	56	23	training	training	NOUN
fcis-26720	56	24	process	process	NOUN
fcis-26720	56	25	;	;	PUNCT
fcis-26720	56	26	using	use	VERB
fcis-26720	56	27	relu	relu	NOUN
fcis-26720	56	28	activation	activation	NOUN
fcis-26720	56	29	function	function	VERB
fcis-26720	56	30	to	to	PART
fcis-26720	56	31	increase	increase	VERB
fcis-26720	56	32	nonlinear	nonlinear	ADJ
fcis-26720	56	33	expression	expression	NOUN
fcis-26720	56	34	ability	ability	NOUN
fcis-26720	56	35	;	;	PUNCT
fcis-26720	56	36	using	use	VERB
fcis-26720	56	37	max	max	PROPN
fcis-26720	56	38	-	-	PUNCT
fcis-26720	56	39	pooling	pooling	NOUN
fcis-26720	56	40	to	to	PART
fcis-26720	56	41	reduce	reduce	VERB
fcis-26720	56	42	dimensionality	dimensionality	NOUN
fcis-26720	56	43	and	and	CCONJ
fcis-26720	56	44	preserve	preserve	VERB
fcis-26720	56	45	important	important	ADJ
fcis-26720	56	46	features	feature	NOUN
fcis-26720	56	47	;	;	PUNCT
fcis-26720	56	48	dropout	dropout	NOUN
fcis-26720	56	49	is	be	AUX
fcis-26720	56	50	adopted	adopt	VERB
fcis-26720	56	51	to	to	PART
fcis-26720	56	52	prevent	prevent	VERB
fcis-26720	56	53	overfitting	overfitting	NOUN
fcis-26720	56	54	and	and	CCONJ
fcis-26720	56	55	improve	improve	VERB
fcis-26720	56	56	generalization	generalization	NOUN
fcis-26720	56	57	performance	performance	NOUN
fcis-26720	56	58	.	.	PUNCT
fcis-26720	57	1	regarding	regard	VERB
fcis-26720	57	2	the	the	DET
fcis-26720	57	3	specific	specific	ADJ
fcis-26720	57	4	parameter	parameter	NOUN
fcis-26720	57	5	configuration	configuration	NOUN
fcis-26720	57	6	of	of	ADP
fcis-26720	57	7	stfeatcnn-4l	stfeatcnn-4l	PROPN
fcis-26720	57	8	,	,	PUNCT
fcis-26720	57	9	table	table	NOUN
fcis-26720	57	10	1	1	NUM
fcis-26720	57	11	provides	provide	VERB
fcis-26720	57	12	an	an	DET
fcis-26720	57	13	explanation	explanation	NOUN
fcis-26720	57	14	.	.	PUNCT
fcis-26720	58	1	(	(	PUNCT
fcis-26720	58	2	2	2	X
fcis-26720	58	3	)	)	PUNCT
fcis-26720	58	4	capsule	capsule	NOUN
fcis-26720	58	5	network	network	NOUN
fcis-26720	58	6	layer	layer	NOUN
fcis-26720	58	7	the	the	DET
fcis-26720	58	8	capsule	capsule	NOUN
fcis-26720	58	9	network	network	NOUN
fcis-26720	58	10	layer	layer	NOUN
fcis-26720	58	11	consists	consist	VERB
fcis-26720	58	12	of	of	ADP
fcis-26720	58	13	the	the	DET
fcis-26720	58	14	main	main	ADJ
fcis-26720	58	15	capsule	capsule	NOUN
fcis-26720	58	16	layer	layer	NOUN
fcis-26720	58	17	(	(	PUNCT
fcis-26720	58	18	primarycaps	primarycap	NOUN
fcis-26720	58	19	)	)	PUNCT
fcis-26720	58	20	and	and	CCONJ
fcis-26720	58	21	the	the	DET
fcis-26720	58	22	label	label	NOUN
fcis-26720	58	23	capsule	capsule	NOUN
fcis-26720	58	24	layer	layer	NOUN
fcis-26720	58	25	(	(	PUNCT
fcis-26720	58	26	labelcaps	labelcaps	PROPN
fcis-26720	58	27	)	)	PUNCT
fcis-26720	58	28	.	.	PUNCT
fcis-26720	59	1	compared	compare	VERB
fcis-26720	59	2	with	with	ADP
fcis-26720	59	3	the	the	DET
fcis-26720	59	4	convolutional	convolutional	ADJ
fcis-26720	59	5	neural	neural	ADJ
fcis-26720	59	6	network	network	NOUN
fcis-26720	59	7	,	,	PUNCT
fcis-26720	59	8	its	its	PRON
fcis-26720	59	9	remarkable	remarkable	ADJ
fcis-26720	59	10	feature	feature	NOUN
fcis-26720	59	11	is	be	AUX
fcis-26720	59	12	that	that	PRON
fcis-26720	59	13	can	can	AUX
fcis-26720	59	14	convert	convert	VERB
fcis-26720	59	15	the	the	DET
fcis-26720	59	16	calculated	calculated	ADJ
fcis-26720	59	17	neurons	neuron	NOUN
fcis-26720	59	18	from	from	ADP
fcis-26720	59	19	a	a	DET
fcis-26720	59	20	single	single	ADJ
fcis-26720	59	21	value	value	NOUN
fcis-26720	59	22	into	into	ADP
fcis-26720	59	23	a	a	DET
fcis-26720	59	24	vector	vector	NOUN
fcis-26720	59	25	[	[	X
fcis-26720	59	26	23	23	NUM
fcis-26720	59	27	]	]	PUNCT
fcis-26720	59	28	.	.	PUNCT
fcis-26720	60	1	this	this	DET
fcis-26720	60	2	vector	vector	NOUN
fcis-26720	60	3	not	not	PART
fcis-26720	60	4	only	only	ADV
fcis-26720	60	5	reflects	reflect	VERB
fcis-26720	60	6	the	the	DET
fcis-26720	60	7	existence	existence	NOUN
fcis-26720	60	8	of	of	ADP
fcis-26720	60	9	neurons	neuron	NOUN
fcis-26720	60	10	,	,	PUNCT
fcis-26720	60	11	but	but	CCONJ
fcis-26720	60	12	also	also	ADV
fcis-26720	60	13	contains	contain	VERB
fcis-26720	60	14	their	their	PRON
fcis-26720	60	15	directionality	directionality	NOUN
fcis-26720	60	16	.	.	PUNCT
fcis-26720	61	1	in	in	ADP
fcis-26720	61	2	order	order	NOUN
fcis-26720	61	3	to	to	PART
fcis-26720	61	4	make	make	VERB
fcis-26720	61	5	full	full	ADJ
fcis-26720	61	6	use	use	NOUN
fcis-26720	61	7	of	of	ADP
fcis-26720	61	8	this	this	DET
fcis-26720	61	9	feature	feature	NOUN
fcis-26720	61	10	and	and	CCONJ
fcis-26720	61	11	obtain	obtain	VERB
fcis-26720	61	12	spatiotemporal	spatiotemporal	ADJ
fcis-26720	61	13	features	feature	NOUN
fcis-26720	61	14	,	,	PUNCT
fcis-26720	61	15	this	this	DET
fcis-26720	61	16	paper	paper	NOUN
fcis-26720	61	17	will	will	AUX
fcis-26720	61	18	convolve	convolve	VERB
fcis-26720	61	19	and	and	CCONJ
fcis-26720	61	20	reshape	reshape	VERB
fcis-26720	61	21	the	the	DET
fcis-26720	61	22	features	feature	NOUN
fcis-26720	61	23	after	after	ADP
fcis-26720	61	24	stfeatcnn-4l	stfeatcnn-4l	PROPN
fcis-26720	61	25	,	,	PUNCT
fcis-26720	61	26	and	and	CCONJ
fcis-26720	61	27	then	then	ADV
fcis-26720	61	28	use	use	VERB
fcis-26720	61	29	dimensional	dimensional	ADJ
fcis-26720	61	30	transformation	transformation	NOUN
fcis-26720	61	31	to	to	PART
fcis-26720	61	32	turn	turn	VERB
fcis-26720	61	33	them	they	PRON
fcis-26720	61	34	into	into	ADP
fcis-26720	61	35	multiple	multiple	ADJ
fcis-26720	61	36	capsule	capsule	NOUN
fcis-26720	61	37	vectors	vector	NOUN
fcis-26720	61	38	,	,	PUNCT
fcis-26720	61	39	finally	finally	ADV
fcis-26720	61	40	forming	form	VERB
fcis-26720	61	41	primarycaps	primarycap	NOUN
fcis-26720	61	42	.	.	PUNCT
fcis-26720	62	1	the	the	DET
fcis-26720	62	2	capsules	capsule	NOUN
fcis-26720	62	3	in	in	ADP
fcis-26720	62	4	primarycaps	primarycap	NOUN
fcis-26720	62	5	are	be	AUX
fcis-26720	62	6	output	output	NOUN
fcis-26720	62	7	to	to	ADP
fcis-26720	62	8	labelscaps	labelscap	NOUN
fcis-26720	62	9	after	after	ADP
fcis-26720	62	10	updating	update	VERB
fcis-26720	62	11	the	the	DET
fcis-26720	62	12	algorithm	algorithm	NOUN
fcis-26720	62	13	through	through	ADP
fcis-26720	62	14	dynamic	dynamic	ADJ
fcis-26720	62	15	routing	routing	NOUN
fcis-26720	62	16	.	.	PUNCT
fcis-26720	63	1	table	table	NOUN
fcis-26720	63	2	2	2	NUM
fcis-26720	63	3	shows	show	VERB
fcis-26720	63	4	the	the	DET
fcis-26720	63	5	changes	change	NOUN
fcis-26720	63	6	of	of	ADP
fcis-26720	63	7	features	feature	NOUN
fcis-26720	63	8	before	before	ADV
fcis-26720	63	9	and	and	CCONJ
fcis-26720	63	10	after	after	ADP
fcis-26720	63	11	shaping	shape	VERB
fcis-26720	63	12	and	and	CCONJ
fcis-26720	63	13	dimension	dimension	NOUN
fcis-26720	63	14	transformation	transformation	NOUN
fcis-26720	63	15	,	,	PUNCT
fcis-26720	63	16	where	where	SCONJ
fcis-26720	63	17	l(length	l(length	NOUN
fcis-26720	63	18	)	)	PUNCT
fcis-26720	63	19	and	and	CCONJ
fcis-26720	63	20	w(width	w(width	ADJ
fcis-26720	63	21	)	)	PUNCT
fcis-26720	63	22	represent	represent	VERB
fcis-26720	63	23	the	the	DET
fcis-26720	63	24	dimensions	dimension	NOUN
fcis-26720	63	25	of	of	ADP
fcis-26720	63	26	features	feature	NOUN
fcis-26720	63	27	after	after	ADP
fcis-26720	63	28	stfeatcnn-4	stfeatcnn-4	PROPN
fcis-26720	63	29	l.	l.	PROPN
fcis-26720	63	30	table	table	PROPN
fcis-26720	63	31	2	2	NUM
fcis-26720	63	32	.	.	PUNCT
fcis-26720	63	33	changes	change	NOUN
fcis-26720	63	34	in	in	ADP
fcis-26720	63	35	characteristic	characteristic	ADJ
fcis-26720	63	36	dimensions	dimension	NOUN
fcis-26720	63	37	of	of	ADP
fcis-26720	63	38	capsule	capsule	NOUN
fcis-26720	63	39	network	network	NOUN
fcis-26720	63	40	operation	operation	NOUN
fcis-26720	63	41	input	input	NOUN
fcis-26720	63	42	feature	feature	NOUN
fcis-26720	63	43	shape	shape	NOUN
fcis-26720	63	44	output	output	NOUN
fcis-26720	63	45	feature	feature	NOUN
fcis-26720	63	46	shape	shape	NOUN
fcis-26720	63	47	original	original	ADJ
fcis-26720	63	48	features	feature	NOUN
fcis-26720	63	49	[	[	X
fcis-26720	63	50	batch	batch	NOUN
fcis-26720	63	51	,	,	PUNCT
fcis-26720	63	52	96	96	NUM
fcis-26720	63	53	×	×	NOUN
fcis-26720	63	54	emotion	emotion	NOUN
fcis-26720	63	55	types	type	NOUN
fcis-26720	63	56	,	,	PUNCT
fcis-26720	63	57	l	l	NOUN
fcis-26720	63	58	,	,	PUNCT
fcis-26720	63	59	w	w	NOUN
fcis-26720	63	60	]	]	X
fcis-26720	63	61	reshape	reshape	NOUN
fcis-26720	63	62	[	[	PUNCT
fcis-26720	63	63	batch	batch	NOUN
fcis-26720	63	64	,	,	PUNCT
fcis-26720	63	65	96	96	NUM
fcis-26720	63	66	×	×	NOUN
fcis-26720	63	67	emotion	emotion	NOUN
fcis-26720	63	68	types	type	NOUN
fcis-26720	63	69	,	,	PUNCT
fcis-26720	63	70	l	l	NOUN
fcis-26720	63	71	,	,	PUNCT
fcis-26720	63	72	w	w	PROPN
fcis-26720	63	73	]	]	X
fcis-26720	63	74	[	[	X
fcis-26720	63	75	batch	batch	NOUN
fcis-26720	63	76	,	,	PUNCT
fcis-26720	63	77	96	96	NUM
fcis-26720	63	78	,	,	PUNCT
fcis-26720	63	79	emotion	emotion	NOUN
fcis-26720	63	80	types	type	NOUN
fcis-26720	63	81	,	,	PUNCT
fcis-26720	63	82	l	l	PROPN
fcis-26720	63	83	×	×	PROPN
fcis-26720	63	84	w	w	PROPN
fcis-26720	63	85	]	]	X
fcis-26720	63	86	dimension	dimension	NOUN
fcis-26720	63	87	permutation	permutation	NOUN
fcis-26720	64	1	[	[	X
fcis-26720	64	2	batch	batch	NOUN
fcis-26720	64	3	,	,	PUNCT
fcis-26720	64	4	96	96	NUM
fcis-26720	64	5	,	,	PUNCT
fcis-26720	64	6	emotion	emotion	NOUN
fcis-26720	64	7	types	type	NOUN
fcis-26720	64	8	,	,	PUNCT
fcis-26720	64	9	l	l	PROPN
fcis-26720	64	10	×	×	PROPN
fcis-26720	64	11	w	w	X
fcis-26720	64	12	]	]	X
fcis-26720	64	13	[	[	X
fcis-26720	64	14	batch	batch	NOUN
fcis-26720	64	15	,	,	PUNCT
fcis-26720	64	16	emotion	emotion	NOUN
fcis-26720	64	17	types	type	NOUN
fcis-26720	64	18	,	,	PUNCT
fcis-26720	64	19	l	l	PROPN
fcis-26720	64	20	×	×	PROPN
fcis-26720	64	21	w	w	PROPN
fcis-26720	64	22	,	,	PUNCT
fcis-26720	64	23	96	96	NUM
fcis-26720	64	24	]	]	PUNCT
fcis-26720	64	25	the	the	DET
fcis-26720	64	26	core	core	NOUN
fcis-26720	64	27	component	component	NOUN
fcis-26720	64	28	of	of	ADP
fcis-26720	64	29	capsule	capsule	NOUN
fcis-26720	64	30	network	network	NOUN
fcis-26720	64	31	is	be	AUX
fcis-26720	64	32	dynamic	dynamic	ADJ
fcis-26720	64	33	routing	routing	NOUN
fcis-26720	64	34	update	update	NOUN
fcis-26720	64	35	algorithm	algorithm	NOUN
fcis-26720	64	36	.	.	PUNCT
fcis-26720	65	1	assuming	assume	VERB
fcis-26720	65	2	that	that	SCONJ
fcis-26720	65	3	the	the	DET
fcis-26720	65	4	number	number	NOUN
fcis-26720	65	5	of	of	ADP
fcis-26720	65	6	iterations	iteration	NOUN
fcis-26720	65	7	is	be	AUX
fcis-26720	65	8	,	,	PUNCT
fcis-26720	65	9	each	each	DET
fcis-26720	65	10	capsule	capsule	NOUN
fcis-26720	65	11	is	be	AUX
fcis-26720	65	12	represented	represent	VERB
fcis-26720	65	13	by	by	ADP
fcis-26720	65	14	,	,	PUNCT
fcis-26720	65	15	and	and	CCONJ
fcis-26720	65	16	initialize	initialize	VERB
fcis-26720	65	17	routing	routing	NOUN
fcis-26720	65	18	parameter	parameter	NOUN
fcis-26720	65	19	0	0	NUM
fcis-26720	65	20	.	.	PUNCT
fcis-26720	66	1	in	in	ADP
fcis-26720	66	2	each	each	DET
fcis-26720	66	3	iteration	iteration	NOUN
fcis-26720	66	4	1,2	1,2	NUM
fcis-26720	66	5	,	,	PUNCT
fcis-26720	66	6	…	…	PUNCT
fcis-26720	66	7	,	,	PUNCT
fcis-26720	66	8	,	,	PUNCT
fcis-26720	66	9	the	the	DET
fcis-26720	66	10	dot	dot	NOUN
fcis-26720	66	11	-	-	PUNCT
fcis-26720	66	12	multiplication	multiplication	NOUN
fcis-26720	66	13	operation	operation	NOUN
fcis-26720	66	14	with	with	ADP
fcis-26720	66	15	the	the	DET
fcis-26720	66	16	weight	weight	NOUN
fcis-26720	66	17	matrix	matrix	NOUN
fcis-26720	66	18	and	and	CCONJ
fcis-26720	66	19	needs	need	VERB
fcis-26720	66	20	to	to	PART
fcis-26720	66	21	be	be	AUX
fcis-26720	66	22	performed	perform	VERB
fcis-26720	66	23	,	,	PUNCT
fcis-26720	66	24	and	and	CCONJ
fcis-26720	66	25	the	the	DET
fcis-26720	66	26	value	value	NOUN
fcis-26720	66	27	obtained	obtain	VERB
fcis-26720	66	28	by	by	ADP
fcis-26720	66	29	dotmultiplication	dotmultiplication	NOUN
fcis-26720	66	30	is	be	AUX
fcis-26720	66	31	multiplied	multiply	VERB
fcis-26720	66	32	by	by	ADP
fcis-26720	66	33	the	the	DET
fcis-26720	66	34	coefficient	coefficient	NOUN
fcis-26720	66	35	and	and	CCONJ
fcis-26720	66	36	accumulated	accumulate	VERB
fcis-26720	66	37	to	to	PART
fcis-26720	66	38	obtain	obtain	VERB
fcis-26720	66	39	.	.	PUNCT
fcis-26720	67	1	at	at	ADP
fcis-26720	67	2	the	the	DET
fcis-26720	67	3	same	same	ADJ
fcis-26720	67	4	time	time	NOUN
fcis-26720	67	5	,	,	PUNCT
fcis-26720	67	6	in	in	ADP
fcis-26720	67	7	order	order	NOUN
fcis-26720	67	8	to	to	PART
fcis-26720	67	9	represent	represent	VERB
fcis-26720	67	10	whether	whether	SCONJ
fcis-26720	67	11	the	the	DET
fcis-26720	67	12	vector	vector	NOUN
fcis-26720	67	13	exists	exist	VERB
fcis-26720	67	14	or	or	CCONJ
fcis-26720	67	15	not	not	PART
fcis-26720	67	16	,	,	PUNCT
fcis-26720	67	17	it	it	PRON
fcis-26720	67	18	is	be	AUX
fcis-26720	67	19	squash	squash	NOUN
fcis-26720	67	20	normalized	normalize	VERB
fcis-26720	67	21	to	to	PART
fcis-26720	67	22	obtain	obtain	VERB
fcis-26720	67	23	jv	jv	NOUN
fcis-26720	67	24	.	.	PUNCT
fcis-26720	68	1	the	the	DET
fcis-26720	68	2	specific	specific	ADJ
fcis-26720	68	3	formula	formula	NOUN
fcis-26720	68	4	is	be	AUX
fcis-26720	68	5	as	as	SCONJ
fcis-26720	68	6	follows	follow	VERB
fcis-26720	68	7	:	:	PUNCT
fcis-26720	68	8	|ˆ	|ˆ	NOUN
fcis-26720	68	9	ꞏi	ꞏi	VERB
fcis-26720	68	10	j	j	PROPN
fcis-26720	68	11	ij	ij	NOUN
fcis-26720	68	12	iwu	iwu	NOUN
fcis-26720	68	13	u	u	X
fcis-26720	68	14	(	(	PUNCT
fcis-26720	68	15	1	1	NUM
fcis-26720	68	16	)	)	PUNCT
fcis-26720	68	17	exp	exp	NOUN
fcis-26720	68	18	(	(	PUNCT
fcis-26720	68	19	)	)	PUNCT
fcis-26720	68	20	(	(	PUNCT
fcis-26720	68	21	)	)	PUNCT
fcis-26720	68	22	exp	exp	NOUN
fcis-26720	68	23	(	(	PUNCT
fcis-26720	68	24	)	)	PUNCT
fcis-26720	69	1	ij	ij	INTJ
fcis-26720	69	2	ij	ij	INTJ
fcis-26720	70	1	ij	ij	INTJ
fcis-26720	71	1	k	k	PROPN
fcis-26720	71	2	ik	ik	PROPN
fcis-26720	71	3	b	b	PROPN
fcis-26720	71	4	c	c	PROPN
fcis-26720	71	5	softmax	softmax	PROPN
fcis-26720	71	6	b	b	PROPN
fcis-26720	71	7	b	b	PROPN
fcis-26720	71	8			NOUN
fcis-26720	72	1			NOUN
fcis-26720	72	2			X
fcis-26720	72	3	(	(	PUNCT
fcis-26720	72	4	2	2	NUM
fcis-26720	72	5	)	)	PUNCT
fcis-26720	72	6	|ˆj	|ˆj	NUM
fcis-26720	72	7	i	i	PRON
fcis-26720	73	1	ij	ij	INTJ
fcis-26720	73	2	j	j	PROPN
fcis-26720	73	3	ic	ic	PROPN
fcis-26720	73	4	us	us	PROPN
fcis-26720	73	5			PROPN
fcis-26720	73	6	(	(	PUNCT
fcis-26720	73	7	3	3	NUM
fcis-26720	73	8	)	)	SYM
fcis-26720	73	9	2	2	NUM
fcis-26720	73	10	2	2	NUM
fcis-26720	73	11	|	|	ADV
fcis-26720	73	12	||	||	NOUN
fcis-26720	74	1	||	||	PROPN
fcis-26720	75	1	(	(	PUNCT
fcis-26720	75	2	)	)	PUNCT
fcis-26720	75	3	1	1	NUM
fcis-26720	75	4	||	||	NOUN
fcis-26720	75	5	||	||	NOUN
fcis-26720	76	1	||	||	PUNCT
fcis-26720	77	1	|	|	ADV
fcis-26720	78	1	j	j	PROPN
fcis-26720	79	1	j	j	PROPN
fcis-26720	79	2	j	j	PROPN
fcis-26720	80	1	j	j	PROPN
fcis-26720	80	2	j	j	PROPN
fcis-26720	80	3	j	j	PROPN
fcis-26720	80	4	v	v	ADP
fcis-26720	81	1	s	s	X
fcis-26720	81	2	s	s	NOUN
fcis-26720	81	3	s	s	NOUN
fcis-26720	81	4	squa	squa	NOUN
fcis-26720	81	5	h	h	NOUN
fcis-26720	81	6	s	s	PART
fcis-26720	81	7	s	s	NOUN
fcis-26720	81	8	s	s	X
fcis-26720	81	9			PROPN
fcis-26720	81	10			NUM
fcis-26720	81	11			X
fcis-26720	81	12	(	(	PUNCT
fcis-26720	81	13	4	4	NUM
fcis-26720	81	14	)	)	PUNCT
fcis-26720	81	15	then	then	ADV
fcis-26720	81	16	update	update	VERB
fcis-26720	81	17	the	the	DET
fcis-26720	81	18	routing	routing	NOUN
fcis-26720	81	19	coefficients	coefficient	NOUN
fcis-26720	81	20	ijb	ijb	NOUN
fcis-26720	81	21	：	：	PUNCT
fcis-26720	81	22	|ˆij	|ˆij	PROPN
fcis-26720	81	23	ij	ij	INTJ
fcis-26720	81	24	j	j	PROPN
fcis-26720	82	1	i	i	PRON
fcis-26720	82	2	jub	jub	PROPN
fcis-26720	82	3	b	b	PROPN
fcis-26720	82	4	v	v	ADV
fcis-26720	82	5			VERB
fcis-26720	83	1			PROPN
fcis-26720	83	2	(	(	PUNCT
fcis-26720	83	3	5	5	NUM
fcis-26720	83	4	)	)	PUNCT
fcis-26720	83	5	repeat	repeat	VERB
fcis-26720	83	6	the	the	DET
fcis-26720	83	7	above	above	ADJ
fcis-26720	83	8	steps	step	NOUN
fcis-26720	83	9	until	until	SCONJ
fcis-26720	83	10	iterations	iteration	NOUN
fcis-26720	83	11	are	be	AUX
fcis-26720	83	12	completed	complete	VERB
fcis-26720	83	13	,	,	PUNCT
fcis-26720	83	14	the	the	DET
fcis-26720	83	15	updated	update	VERB
fcis-26720	83	16	is	be	AUX
fcis-26720	83	17	sent	send	VERB
fcis-26720	83	18	to	to	ADP
fcis-26720	83	19	the	the	DET
fcis-26720	83	20	labelscaps	labelscaps	ADJ
fcis-26720	83	21	layer	layer	NOUN
fcis-26720	83	22	.	.	PUNCT
fcis-26720	84	1	in	in	ADP
fcis-26720	84	2	this	this	DET
fcis-26720	84	3	layer	layer	NOUN
fcis-26720	84	4	,	,	PUNCT
fcis-26720	84	5	it	it	PRON
fcis-26720	84	6	is	be	AUX
fcis-26720	84	7	necessary	necessary	ADJ
fcis-26720	84	8	to	to	PART
fcis-26720	84	9	modulate	modulate	VERB
fcis-26720	84	10	each	each	DET
fcis-26720	84	11	emotion	emotion	NOUN
fcis-26720	84	12	and	and	CCONJ
fcis-26720	84	13	use	use	VERB
fcis-26720	84	14	a	a	DET
fcis-26720	84	15	softmax	softmax	NOUN
fcis-26720	84	16	function	function	NOUN
fcis-26720	84	17	to	to	PART
fcis-26720	84	18	predict	predict	VERB
fcis-26720	84	19	the	the	DET
fcis-26720	84	20	possibility	possibility	NOUN
fcis-26720	84	21	of	of	ADP
fcis-26720	84	22	each	each	DET
fcis-26720	84	23	emotion	emotion	NOUN
fcis-26720	84	24	,	,	PUNCT
fcis-26720	84	25	and	and	CCONJ
fcis-26720	84	26	finally	finally	ADV
fcis-26720	84	27	realize	realize	VERB
fcis-26720	84	28	the	the	DET
fcis-26720	84	29	task	task	NOUN
fcis-26720	84	30	of	of	ADP
fcis-26720	84	31	speech	speech	NOUN
fcis-26720	84	32	emotion	emotion	NOUN
fcis-26720	84	33	recognition	recognition	NOUN
fcis-26720	84	34	.	.	PUNCT
fcis-26720	85	1	according	accord	VERB
fcis-26720	85	2	to	to	ADP
fcis-26720	85	3	the	the	DET
fcis-26720	85	4	core	core	NOUN
fcis-26720	85	5	mechanism	mechanism	NOUN
fcis-26720	85	6	of	of	ADP
fcis-26720	85	7	the	the	DET
fcis-26720	85	8	dynamic	dynamic	ADJ
fcis-26720	85	9	routing	routing	NOUN
fcis-26720	85	10	update	update	NOUN
fcis-26720	85	11	algorithm	algorithm	NOUN
fcis-26720	85	12	,	,	PUNCT
fcis-26720	85	13	the	the	DET
fcis-26720	85	14	capsule	capsule	NOUN
fcis-26720	85	15	network	network	NOUN
fcis-26720	85	16	will	will	AUX
fcis-26720	85	17	adjust	adjust	VERB
fcis-26720	85	18	the	the	DET
fcis-26720	85	19	length	length	NOUN
fcis-26720	85	20	of	of	ADP
fcis-26720	85	21	the	the	DET
fcis-26720	85	22	"	"	PUNCT
fcis-26720	85	23	sum	sum	NOUN
fcis-26720	85	24	vector	vector	NOUN
fcis-26720	85	25	"	"	PUNCT
fcis-26720	85	26	and	and	CCONJ
fcis-26720	85	27	each	each	DET
fcis-26720	85	28	prediction	prediction	NOUN
fcis-26720	85	29	vector	vector	NOUN
fcis-26720	85	30	according	accord	VERB
fcis-26720	85	31	to	to	ADP
fcis-26720	85	32	the	the	DET
fcis-26720	85	33	similarity	similarity	NOUN
fcis-26720	85	34	between	between	ADP
fcis-26720	85	35	them	they	PRON
fcis-26720	85	36	during	during	ADP
fcis-26720	85	37	the	the	DET
fcis-26720	85	38	route	route	NOUN
fcis-26720	85	39	update	update	NOUN
fcis-26720	85	40	process	process	NOUN
fcis-26720	85	41	,	,	PUNCT
fcis-26720	85	42	that	that	ADV
fcis-26720	85	43	is	is	ADV
fcis-26720	85	44	,	,	PUNCT
fcis-26720	85	45	the	the	DET
fcis-26720	85	46	vector	vector	NOUN
fcis-26720	85	47	with	with	ADP
fcis-26720	85	48	high	high	ADJ
fcis-26720	85	49	similarity	similarity	NOUN
fcis-26720	85	50	will	will	AUX
fcis-26720	85	51	be	be	AUX
fcis-26720	85	52	elongated	elongate	VERB
fcis-26720	85	53	,	,	PUNCT
fcis-26720	85	54	while	while	SCONJ
fcis-26720	85	55	the	the	DET
fcis-26720	85	56	vector	vector	NOUN
fcis-26720	85	57	with	with	ADP
fcis-26720	85	58	low	low	ADJ
fcis-26720	85	59	similarity	similarity	NOUN
fcis-26720	85	60	will	will	AUX
fcis-26720	85	61	become	become	VERB
fcis-26720	85	62	shorter	short	ADJ
fcis-26720	85	63	.	.	PUNCT
fcis-26720	86	1	this	this	DET
fcis-26720	86	2	mechanism	mechanism	NOUN
fcis-26720	86	3	ensures	ensure	VERB
fcis-26720	86	4	that	that	SCONJ
fcis-26720	86	5	vectors	vector	NOUN
fcis-26720	86	6	that	that	PRON
fcis-26720	86	7	are	be	AUX
fcis-26720	86	8	more	more	ADV
fcis-26720	86	9	similar	similar	ADJ
fcis-26720	86	10	to	to	ADP
fcis-26720	86	11	"	"	PUNCT
fcis-26720	86	12	sum	sum	NOUN
fcis-26720	86	13	vectors	vector	NOUN
fcis-26720	86	14	"	"	PUNCT
fcis-26720	86	15	can	can	AUX
fcis-26720	86	16	be	be	AUX
fcis-26720	86	17	sent	send	VERB
fcis-26720	86	18	to	to	ADP
fcis-26720	86	19	labelcaps	labelcaps	PROPN
fcis-26720	86	20	more	more	ADV
fcis-26720	86	21	,	,	PUNCT
fcis-26720	86	22	thus	thus	ADV
fcis-26720	86	23	effectively	effectively	ADV
fcis-26720	86	24	achieving	achieve	VERB
fcis-26720	86	25	classification	classification	NOUN
fcis-26720	86	26	effects	effect	NOUN
fcis-26720	86	27	.	.	PUNCT
fcis-26720	87	1	according	accord	VERB
fcis-26720	87	2	to	to	ADP
fcis-26720	87	3	the	the	DET
fcis-26720	87	4	characteristics	characteristic	NOUN
fcis-26720	87	5	of	of	ADP
fcis-26720	87	6	capsule	capsule	NOUN
fcis-26720	87	7	network	network	NOUN
fcis-26720	87	8	,	,	PUNCT
fcis-26720	87	9	this	this	DET
fcis-26720	87	10	paper	paper	NOUN
fcis-26720	87	11	deeply	deeply	ADV
fcis-26720	87	12	studies	study	VERB
fcis-26720	87	13	the	the	DET
fcis-26720	87	14	parameter	parameter	NOUN
fcis-26720	87	15	of	of	ADP
fcis-26720	87	16	weight	weight	NOUN
fcis-26720	87	17	matrix	matrix	NOUN
fcis-26720	87	18	in	in	ADP
fcis-26720	87	19	dynamic	dynamic	ADJ
fcis-26720	87	20	routing	routing	NOUN
fcis-26720	87	21	algorithm	algorithm	NOUN
fcis-26720	87	22	.	.	PUNCT
fcis-26720	88	1	when	when	SCONJ
fcis-26720	88	2	the	the	DET
fcis-26720	88	3	weight	weight	NOUN
fcis-26720	88	4	matrix	matrix	NOUN
fcis-26720	88	5	is	be	AUX
fcis-26720	88	6	assigned	assign	VERB
fcis-26720	88	7	to	to	ADP
fcis-26720	88	8	each	each	DET
fcis-26720	88	9	capsule	capsule	NOUN
fcis-26720	88	10	separately	separately	ADV
fcis-26720	88	11	for	for	ADP
fcis-26720	88	12	training	training	NOUN
fcis-26720	88	13	,	,	PUNCT
fcis-26720	88	14	the	the	DET
fcis-26720	88	15	model	model	NOUN
fcis-26720	88	16	parameters	parameter	NOUN
fcis-26720	88	17	will	will	AUX
fcis-26720	88	18	increase	increase	VERB
fcis-26720	88	19	dramatically	dramatically	ADV
fcis-26720	88	20	,	,	PUNCT
fcis-26720	88	21	and	and	CCONJ
fcis-26720	88	22	problems	problem	NOUN
fcis-26720	88	23	such	such	ADJ
fcis-26720	88	24	as	as	ADP
fcis-26720	88	25	overfitting	overfitte	VERB
fcis-26720	88	26	,	,	PUNCT
fcis-26720	88	27	gradient	gradient	ADJ
fcis-26720	88	28	explosion	explosion	NOUN
fcis-26720	88	29	and	and	CCONJ
fcis-26720	88	30	gradient	gradient	NOUN
fcis-26720	88	31	disappearance	disappearance	NOUN
fcis-26720	88	32	will	will	AUX
fcis-26720	88	33	also	also	ADV
fcis-26720	88	34	occur	occur	VERB
fcis-26720	88	35	.	.	PUNCT
fcis-26720	89	1	therefore	therefore	ADV
fcis-26720	89	2	,	,	PUNCT
fcis-26720	89	3	this	this	DET
fcis-26720	89	4	paper	paper	NOUN
fcis-26720	89	5	adopts	adopt	VERB
fcis-26720	89	6	the	the	DET
fcis-26720	89	7	method	method	NOUN
fcis-26720	89	8	of	of	ADP
fcis-26720	89	9	sharing	share	VERB
fcis-26720	89	10	weights	weight	NOUN
fcis-26720	89	11	,	,	PUNCT
fcis-26720	89	12	which	which	PRON
fcis-26720	89	13	reduces	reduce	VERB
fcis-26720	89	14	the	the	DET
fcis-26720	89	15	complexity	complexity	NOUN
fcis-26720	89	16	of	of	ADP
fcis-26720	89	17	the	the	DET
fcis-26720	89	18	model	model	NOUN
fcis-26720	89	19	and	and	CCONJ
fcis-26720	89	20	allows	allow	VERB
fcis-26720	89	21	multiple	multiple	ADJ
fcis-26720	89	22	capsules	capsule	NOUN
fcis-26720	89	23	to	to	PART
fcis-26720	89	24	share	share	VERB
fcis-26720	89	25	the	the	DET
fcis-26720	89	26	same	same	ADJ
fcis-26720	89	27	set	set	NOUN
fcis-26720	89	28	of	of	ADP
fcis-26720	89	29	weights	weight	NOUN
fcis-26720	89	30	,	,	PUNCT
fcis-26720	89	31	so	so	SCONJ
fcis-26720	89	32	that	that	SCONJ
fcis-26720	89	33	the	the	DET
fcis-26720	89	34	model	model	NOUN
fcis-26720	89	35	can	can	AUX
fcis-26720	89	36	more	more	ADV
fcis-26720	89	37	accurately	accurately	ADV
fcis-26720	89	38	grasp	grasp	VERB
fcis-26720	89	39	the	the	DET
fcis-26720	89	40	internal	internal	ADJ
fcis-26720	89	41	relationship	relationship	NOUN
fcis-26720	89	42	between	between	ADP
fcis-26720	89	43	features	feature	NOUN
fcis-26720	89	44	and	and	CCONJ
fcis-26720	89	45	improve	improve	VERB
fcis-26720	89	46	its	its	PRON
fcis-26720	89	47	generalization	generalization	NOUN
fcis-26720	89	48	ability	ability	NOUN
fcis-26720	89	49	.	.	PUNCT
fcis-26720	90	1	(	(	PUNCT
fcis-26720	90	2	3	3	X
fcis-26720	90	3	)	)	PUNCT
fcis-26720	90	4	loss	loss	NOUN
fcis-26720	90	5	function	function	NOUN
fcis-26720	90	6	the	the	DET
fcis-26720	90	7	loss	loss	NOUN
fcis-26720	90	8	function	function	NOUN
fcis-26720	90	9	used	use	VERB
fcis-26720	90	10	in	in	ADP
fcis-26720	90	11	this	this	DET
fcis-26720	90	12	paper	paper	NOUN
fcis-26720	90	13	is	be	AUX
fcis-26720	90	14	the	the	DET
fcis-26720	90	15	margin	margin	NOUN
fcis-26720	90	16	-	-	PUNCT
fcis-26720	90	17	loss	loss	NOUN
fcis-26720	90	18	function	function	NOUN
fcis-26720	90	19	.	.	PUNCT
fcis-26720	91	1	compared	compare	VERB
fcis-26720	91	2	with	with	ADP
fcis-26720	91	3	traditional	traditional	ADJ
fcis-26720	91	4	loss	loss	NOUN
fcis-26720	91	5	functions	function	NOUN
fcis-26720	91	6	such	such	ADJ
fcis-26720	91	7	as	as	ADP
fcis-26720	91	8	cross	cross	NOUN
fcis-26720	91	9	entropy	entropy	NOUN
fcis-26720	91	10	and	and	CCONJ
fcis-26720	91	11	mean	mean	VERB
fcis-26720	91	12	square	square	ADJ
fcis-26720	91	13	error	error	NOUN
fcis-26720	91	14	,	,	PUNCT
fcis-26720	91	15	margin	margin	NOUN
fcis-26720	91	16	-	-	PUNCT
fcis-26720	91	17	loss	loss	NOUN
fcis-26720	91	18	focuses	focus	VERB
fcis-26720	91	19	more	more	ADJ
fcis-26720	91	20	on	on	ADP
fcis-26720	91	21	the	the	DET
fcis-26720	91	22	relative	relative	ADJ
fcis-26720	91	23	distance	distance	NOUN
fcis-26720	91	24	and	and	CCONJ
fcis-26720	91	25	interval	interval	NOUN
fcis-26720	91	26	between	between	ADP
fcis-26720	91	27	samples	sample	NOUN
fcis-26720	91	28	,	,	PUNCT
fcis-26720	91	29	it	it	PRON
fcis-26720	91	30	can	can	AUX
fcis-26720	91	31	maximize	maximize	VERB
fcis-26720	91	32	the	the	DET
fcis-26720	91	33	interval	interval	NOUN
fcis-26720	91	34	of	of	ADP
fcis-26720	91	35	different	different	ADJ
fcis-26720	91	36	types	type	NOUN
fcis-26720	91	37	of	of	ADP
fcis-26720	91	38	samples	sample	NOUN
fcis-26720	91	39	to	to	PART
fcis-26720	91	40	optimize	optimize	VERB
fcis-26720	91	41	the	the	DET
fcis-26720	91	42	model	model	NOUN
fcis-26720	91	43	[	[	X
fcis-26720	91	44	24	24	NUM
fcis-26720	91	45	]	]	PUNCT
fcis-26720	91	46	.	.	PUNCT
fcis-26720	92	1	margin	margin	NOUN
fcis-26720	92	2	-	-	PUNCT
fcis-26720	92	3	loss	loss	NOUN
fcis-26720	92	4	is	be	AUX
fcis-26720	92	5	calculated	calculate	VERB
fcis-26720	92	6	as	as	SCONJ
fcis-26720	92	7	follows	follow	VERB
fcis-26720	92	8	:	:	PUNCT
fcis-26720	92	9	2	2	NUM
fcis-26720	92	10	2(0	2(0	NUM
fcis-26720	92	11	,	,	PUNCT
fcis-26720	92	12	||	||	NOUN
fcis-26720	92	13	||	||	NOUN
fcis-26720	92	14	)	)	PUNCT
fcis-26720	93	1	(	(	PUNCT
fcis-26720	93	2	1	1	X
fcis-26720	93	3	)	)	PUNCT
fcis-26720	93	4	(	(	PUNCT
fcis-26720	93	5	0,||	0,||	NUM
fcis-26720	93	6	||	||	NOUN
fcis-26720	93	7	)	)	PUNCT
fcis-26720	94	1	c	c	NOUN
fcis-26720	94	2	c	c	NOUN
fcis-26720	94	3	c	c	NOUN
fcis-26720	94	4	c	c	X
fcis-26720	94	5	cl	cl	NOUN
fcis-26720	95	1	t	t	PROPN
fcis-26720	95	2	max	max	PROPN
fcis-26720	95	3	m	m	PROPN
fcis-26720	95	4	v	v	PROPN
fcis-26720	95	5	t	t	PROPN
fcis-26720	95	6	max	max	PROPN
fcis-26720	95	7	v	v	ADP
fcis-26720	95	8	m	m	PROPN
fcis-26720	95	9			ADJ
fcis-26720	95	10			PROPN
fcis-26720	95	11			PROPN
fcis-26720	95	12			VERB
fcis-26720	95	13			NOUN
fcis-26720	95	14			PROPN
fcis-26720	95	15			NOUN
fcis-26720	95	16	(	(	PUNCT
fcis-26720	95	17	6	6	NUM
fcis-26720	95	18	)	)	PUNCT
fcis-26720	95	19	among	among	ADP
fcis-26720	95	20	them	they	PRON
fcis-26720	95	21	,	,	PUNCT
fcis-26720	95	22	is	be	AUX
fcis-26720	95	23	the	the	DET
fcis-26720	95	24	vector	vector	NOUN
fcis-26720	95	25	obtained	obtain	VERB
fcis-26720	95	26	by	by	ADP
fcis-26720	95	27	one	one	NUM
fcis-26720	95	28	-	-	PUNCT
fcis-26720	95	29	hot	hot	ADJ
fcis-26720	95	30	encoding	encoding	NOUN
fcis-26720	95	31	the	the	DET
fcis-26720	95	32	emotion	emotion	NOUN
fcis-26720	95	33	label	label	NOUN
fcis-26720	95	34	,	,	PUNCT
fcis-26720	95	35	represents	represent	VERB
fcis-26720	95	36	the	the	DET
fcis-26720	95	37	modulo	modulo	NOUN
fcis-26720	95	38	of	of	ADP
fcis-26720	95	39	each	each	DET
fcis-26720	95	40	emotion	emotion	NOUN
fcis-26720	95	41	entity	entity	NOUN
fcis-26720	95	42	,	,	PUNCT
fcis-26720	95	43	∈	∈	PROPN
fcis-26720	95	44	0,1	0,1	NUM
fcis-26720	95	45	.	.	PUNCT
fcis-26720	96	1	3	3	X
fcis-26720	96	2	.	.	X
fcis-26720	96	3	experimental	experimental	ADJ
fcis-26720	96	4	analysis	analysis	NOUN
fcis-26720	96	5	(	(	PUNCT
fcis-26720	96	6	1	1	X
fcis-26720	96	7	)	)	PUNCT
fcis-26720	96	8	corpus	corpus	NOUN
fcis-26720	96	9	to	to	PART
fcis-26720	96	10	realize	realize	VERB
fcis-26720	96	11	tibetan	tibetan	ADJ
fcis-26720	96	12	speech	speech	NOUN
fcis-26720	96	13	emotion	emotion	NOUN
fcis-26720	96	14	recognition	recognition	NOUN
fcis-26720	96	15	,	,	PUNCT
fcis-26720	96	16	this	this	DET
fcis-26720	96	17	paper	paper	NOUN
fcis-26720	96	18	constructs	construct	VERB
fcis-26720	96	19	a	a	DET
fcis-26720	96	20	tibetan	tibetan	ADJ
fcis-26720	96	21	speech	speech	NOUN
fcis-26720	96	22	emotion	emotion	NOUN
fcis-26720	96	23	corpus	corpus	NOUN
fcis-26720	96	24	:	:	PUNCT
fcis-26720	96	25	tbsec001	tbsec001	X
fcis-26720	96	26	.	.	PUNCT
fcis-26720	97	1	at	at	ADP
fcis-26720	97	2	the	the	DET
fcis-26720	97	3	same	same	ADJ
fcis-26720	97	4	time	time	NOUN
fcis-26720	97	5	,	,	PUNCT
fcis-26720	97	6	in	in	ADP
fcis-26720	97	7	order	order	NOUN
fcis-26720	97	8	to	to	PART
fcis-26720	97	9	enhance	enhance	VERB
fcis-26720	97	10	the	the	DET
fcis-26720	97	11	comprehensiveness	comprehensiveness	NOUN
fcis-26720	97	12	and	and	CCONJ
fcis-26720	97	13	accuracy	accuracy	NOUN
fcis-26720	97	14	of	of	ADP
fcis-26720	97	15	the	the	DET
fcis-26720	97	16	research	research	NOUN
fcis-26720	97	17	,	,	PUNCT
fcis-26720	97	18	this	this	DET
fcis-26720	97	19	paper	paper	NOUN
fcis-26720	97	20	also	also	ADV
fcis-26720	97	21	selects	select	VERB
fcis-26720	97	22	two	two	NUM
fcis-26720	97	23	widely	widely	ADV
fcis-26720	97	24	recognized	recognize	VERB
fcis-26720	97	25	public	public	ADJ
fcis-26720	97	26	corpora	corpora	NOUN
fcis-26720	97	27	as	as	ADP
fcis-26720	97	28	supplements	supplement	NOUN
fcis-26720	97	29	:	:	PUNCT
fcis-26720	97	30	emo	emo	NOUN
fcis-26720	97	31	-	-	PUNCT
fcis-26720	97	32	db	db	NOUN
fcis-26720	97	33	and	and	CCONJ
fcis-26720	97	34	ravdess	ravdess	NOUN
fcis-26720	97	35	[	[	X
fcis-26720	97	36	25	25	NUM
fcis-26720	97	37	]	]	PUNCT
fcis-26720	97	38	.	.	PUNCT
fcis-26720	98	1	the	the	DET
fcis-26720	98	2	tbsec001	tbsec001	PROPN
fcis-26720	98	3	corpus	corpus	NOUN
fcis-26720	98	4	consists	consist	VERB
fcis-26720	98	5	of	of	ADP
fcis-26720	98	6	speeches	speech	NOUN
fcis-26720	98	7	by	by	ADP
fcis-26720	98	8	12	12	NUM
fcis-26720	98	9	native	native	ADJ
fcis-26720	98	10	tibetan	tibetan	PROPN
fcis-26720	98	11	speakers	speaker	NOUN
fcis-26720	98	12	,	,	PUNCT
fcis-26720	98	13	including	include	VERB
fcis-26720	98	14	6	6	NUM
fcis-26720	98	15	women	woman	NOUN
fcis-26720	98	16	and	and	CCONJ
fcis-26720	98	17	6	6	NUM
fcis-26720	98	18	men	man	NOUN
fcis-26720	98	19	.	.	PUNCT
fcis-26720	99	1	all	all	DET
fcis-26720	99	2	the	the	DET
fcis-26720	99	3	participants	participant	NOUN
fcis-26720	99	4	in	in	ADP
fcis-26720	99	5	the	the	DET
fcis-26720	99	6	recording	recording	NOUN
fcis-26720	99	7	are	be	AUX
fcis-26720	99	8	from	from	ADP
fcis-26720	99	9	tibet	tibet	PROPN
fcis-26720	99	10	university	university	PROPN
fcis-26720	99	11	,	,	PUNCT
fcis-26720	99	12	and	and	CCONJ
fcis-26720	99	13	their	their	PRON
fcis-26720	99	14	mother	mother	NOUN
fcis-26720	99	15	tongues	tongue	NOUN
fcis-26720	99	16	are	be	AUX
fcis-26720	99	17	‘	'	PUNCT
fcis-26720	99	18	weizang	weizang	NOUN
fcis-26720	99	19	’	'	PUNCT
fcis-26720	99	20	dialect	dialect	NOUN
fcis-26720	99	21	.	.	PUNCT
fcis-26720	100	1	the	the	DET
fcis-26720	100	2	database	database	NOUN
fcis-26720	100	3	contains	contain	VERB
fcis-26720	100	4	five	five	NUM
fcis-26720	100	5	emotional	emotional	ADJ
fcis-26720	100	6	categories	category	NOUN
fcis-26720	100	7	:	:	PUNCT
fcis-26720	100	8	angry	angry	ADJ
fcis-26720	100	9	,	,	PUNCT
fcis-26720	100	10	fear	fear	NOUN
fcis-26720	100	11	,	,	PUNCT
fcis-26720	100	12	happy	happy	ADJ
fcis-26720	100	13	,	,	PUNCT
fcis-26720	100	14	neutral	neutral	ADJ
fcis-26720	100	15	and	and	CCONJ
fcis-26720	100	16	sad	sad	ADJ
fcis-26720	100	17	.	.	PUNCT
fcis-26720	101	1	during	during	ADP
fcis-26720	101	2	the	the	DET
fcis-26720	101	3	recording	recording	NOUN
fcis-26720	101	4	process	process	NOUN
fcis-26720	101	5	,	,	PUNCT
fcis-26720	101	6	12	12	NUM
fcis-26720	101	7	recording	recording	NOUN
fcis-26720	101	8	personnel	personnel	NOUN
fcis-26720	101	9	not	not	PART
fcis-26720	101	10	only	only	ADV
fcis-26720	101	11	participated	participate	VERB
fcis-26720	101	12	in	in	ADP
fcis-26720	101	13	the	the	DET
fcis-26720	101	14	recording	recording	NOUN
fcis-26720	101	15	work	work	NOUN
fcis-26720	101	16	,	,	PUNCT
fcis-26720	101	17	but	but	CCONJ
fcis-26720	101	18	also	also	ADV
fcis-26720	101	19	checked	check	VERB
fcis-26720	101	20	each	each	DET
fcis-26720	101	21	others	other	NOUN
fcis-26720	101	22	emotional	emotional	ADJ
fcis-26720	101	23	expression	expression	NOUN
fcis-26720	101	24	of	of	ADP
fcis-26720	101	25	the	the	DET
fcis-26720	101	26	audio	audio	NOUN
fcis-26720	101	27	to	to	PART
fcis-26720	101	28	ensure	ensure	VERB
fcis-26720	101	29	that	that	SCONJ
fcis-26720	101	30	each	each	DET
fcis-26720	101	31	audio	audio	NOUN
fcis-26720	101	32	can	can	AUX
fcis-26720	101	33	accurately	accurately	ADV
fcis-26720	101	34	convey	convey	VERB
fcis-26720	101	35	the	the	DET
fcis-26720	101	36	preset	preset	ADJ
fcis-26720	101	37	emotion	emotion	NOUN
fcis-26720	101	38	.	.	PUNCT
fcis-26720	102	1	the	the	DET
fcis-26720	102	2	audio	audio	NOUN
fcis-26720	102	3	that	that	PRON
fcis-26720	102	4	does	do	AUX
fcis-26720	102	5	not	not	PART
fcis-26720	102	6	meet	meet	VERB
fcis-26720	102	7	the	the	DET
fcis-26720	102	8	requirements	requirement	NOUN
fcis-26720	102	9	is	be	AUX
fcis-26720	102	10	culled	cull	VERB
fcis-26720	102	11	to	to	PART
fcis-26720	102	12	ensure	ensure	VERB
fcis-26720	102	13	the	the	DET
fcis-26720	102	14	overall	overall	ADJ
fcis-26720	102	15	quality	quality	NOUN
fcis-26720	102	16	of	of	ADP
fcis-26720	102	17	the	the	DET
fcis-26720	102	18	corpus	corpus	NOUN
fcis-26720	102	19	.	.	PUNCT
fcis-26720	103	1	through	through	ADP
fcis-26720	103	2	repeated	repeat	VERB
fcis-26720	103	3	inspection	inspection	NOUN
fcis-26720	103	4	88	88	NUM
fcis-26720	103	5	and	and	CCONJ
fcis-26720	103	6	recording	recording	NOUN
fcis-26720	103	7	,	,	PUNCT
fcis-26720	103	8	the	the	DET
fcis-26720	103	9	tbsec001	tbsec001	NOUN
fcis-26720	103	10	corpus	corpus	NOUN
fcis-26720	103	11	finally	finally	ADV
fcis-26720	103	12	included	include	VERB
fcis-26720	103	13	6,000	6,000	NUM
fcis-26720	103	14	emotional	emotional	ADJ
fcis-26720	103	15	and	and	CCONJ
fcis-26720	103	16	high	high	ADJ
fcis-26720	103	17	-	-	PUNCT
fcis-26720	103	18	quality	quality	NOUN
fcis-26720	103	19	tibetan	tibetan	ADJ
fcis-26720	103	20	speeches	speech	NOUN
fcis-26720	103	21	.	.	PUNCT
fcis-26720	104	1	these	these	DET
fcis-26720	104	2	speech	speech	NOUN
fcis-26720	104	3	samples	sample	NOUN
fcis-26720	104	4	can	can	AUX
fcis-26720	104	5	provide	provide	VERB
fcis-26720	104	6	solid	solid	ADJ
fcis-26720	104	7	data	datum	NOUN
fcis-26720	104	8	support	support	NOUN
fcis-26720	104	9	for	for	ADP
fcis-26720	104	10	the	the	DET
fcis-26720	104	11	research	research	NOUN
fcis-26720	104	12	of	of	ADP
fcis-26720	104	13	this	this	DET
fcis-26720	104	14	paper	paper	NOUN
fcis-26720	104	15	.	.	PUNCT
fcis-26720	105	1	table	table	NOUN
fcis-26720	105	2	3	3	NUM
fcis-26720	105	3	lists	list	VERB
fcis-26720	105	4	the	the	DET
fcis-26720	105	5	specific	specific	ADJ
fcis-26720	105	6	information	information	NOUN
fcis-26720	105	7	of	of	ADP
fcis-26720	105	8	tbsec001	tbsec001	NOUN
fcis-26720	105	9	sentiment	sentiment	NOUN
fcis-26720	105	10	corpus	corpus	NOUN
fcis-26720	105	11	in	in	ADP
fcis-26720	105	12	detail	detail	NOUN
fcis-26720	105	13	.	.	PUNCT
fcis-26720	106	1	table	table	NOUN
fcis-26720	106	2	3	3	NUM
fcis-26720	106	3	.	.	PUNCT
fcis-26720	107	1	introduction	introduction	NOUN
fcis-26720	107	2	to	to	ADP
fcis-26720	107	3	tbsec001	tbsec001	NOUN
fcis-26720	107	4	speech	speech	NOUN
fcis-26720	107	5	emotion	emotion	NOUN
fcis-26720	107	6	corpus	corpus	NOUN
fcis-26720	107	7	corpus	corpus	PROPN
fcis-26720	107	8	tbsec001	tbsec001	PUNCT
fcis-26720	107	9	language	language	NOUN
fcis-26720	107	10	tibetan	tibetan	ADJ
fcis-26720	107	11	dialect	dialect	PROPN
fcis-26720	107	12	weizang	weizang	PROPN
fcis-26720	107	13	dialect	dialect	PROPN
fcis-26720	107	14	type	type	NOUN
fcis-26720	107	15	performance	performance	NOUN
fcis-26720	107	16	type	type	NOUN
fcis-26720	107	17	types	type	NOUN
fcis-26720	107	18	of	of	ADP
fcis-26720	107	19	emotions	emotion	NOUN
fcis-26720	107	20	5	5	NUM
fcis-26720	107	21	kinds	kind	NOUN
fcis-26720	107	22	number	number	NOUN
fcis-26720	107	23	of	of	ADP
fcis-26720	107	24	recordings	recording	NOUN
fcis-26720	107	25	6	6	NUM
fcis-26720	107	26	males	male	NOUN
fcis-26720	107	27	6	6	NUM
fcis-26720	107	28	females	female	NOUN
fcis-26720	107	29	tone	tone	NOUN
fcis-26720	107	30	4	4	NUM
fcis-26720	107	31	characteristics	characteristic	NOUN
fcis-26720	107	32	of	of	ADP
fcis-26720	107	33	consonant	consonant	ADJ
fcis-26720	107	34	pronunciation	pronunciation	NOUN
fcis-26720	107	35	voiced	voice	VERB
fcis-26720	107	36	consonants	consonant	NOUN
fcis-26720	107	37	are	be	AUX
fcis-26720	107	38	cleared	clear	VERB
fcis-26720	107	39	and	and	CCONJ
fcis-26720	107	40	complex	complex	ADJ
fcis-26720	107	41	consonants	consonant	NOUN
fcis-26720	107	42	are	be	AUX
fcis-26720	107	43	simplified	simplify	VERB
fcis-26720	107	44	[	[	X
fcis-26720	107	45	26	26	NUM
fcis-26720	107	46	]	]	PUNCT
fcis-26720	107	47	characteristics	characteristic	NOUN
fcis-26720	107	48	of	of	ADP
fcis-26720	107	49	vowel	vowel	NOUN
fcis-26720	107	50	pronunciation	pronunciation	NOUN
fcis-26720	107	51	nasalized	nasalized	ADJ
fcis-26720	107	52	&	&	CCONJ
fcis-26720	107	53	non	non	ADJ
fcis-26720	107	54	-	-	ADJ
fcis-26720	107	55	nasalized	nasalized	ADJ
fcis-26720	107	56	vowels	vowel	NOUN
fcis-26720	107	57	[	[	X
fcis-26720	107	58	27	27	NUM
fcis-26720	107	59	]	]	PUNCT
fcis-26720	107	60	duration	duration	NOUN
fcis-26720	107	61	0.8	0.8	NUM
fcis-26720	107	62	~	~	SYM
fcis-26720	107	63	4	4	NUM
fcis-26720	107	64	seconds	second	NOUN
fcis-26720	107	65	number	number	NOUN
fcis-26720	107	66	of	of	ADP
fcis-26720	107	67	corpus	corpus	NOUN
fcis-26720	107	68	1200	1200	NUM
fcis-26720	107	69	items	item	NOUN
fcis-26720	107	70	for	for	ADP
fcis-26720	107	71	each	each	PRON
fcis-26720	107	72	of	of	ADP
fcis-26720	107	73	5	5	NUM
fcis-26720	107	74	emotions	emotion	NOUN
fcis-26720	107	75	,	,	PUNCT
fcis-26720	107	76	6000	6000	NUM
fcis-26720	107	77	items	item	NOUN
fcis-26720	107	78	in	in	ADP
fcis-26720	107	79	total	total	NOUN
fcis-26720	107	80	in	in	ADP
fcis-26720	107	81	addition	addition	NOUN
fcis-26720	107	82	to	to	ADP
fcis-26720	107	83	some	some	DET
fcis-26720	107	84	basic	basic	ADJ
fcis-26720	107	85	information	information	NOUN
fcis-26720	107	86	about	about	ADP
fcis-26720	107	87	tbsec001	tbsec001	PROPN
fcis-26720	107	88	,	,	PUNCT
fcis-26720	107	89	table	table	NOUN
fcis-26720	107	90	3	3	NUM
fcis-26720	107	91	also	also	ADV
fcis-26720	107	92	illustrates	illustrate	VERB
fcis-26720	107	93	some	some	DET
fcis-26720	107	94	unique	unique	ADJ
fcis-26720	107	95	features	feature	NOUN
fcis-26720	107	96	of	of	ADP
fcis-26720	107	97	tibetan	tibetan	ADJ
fcis-26720	107	98	pronunciation	pronunciation	NOUN
fcis-26720	107	99	:	:	PUNCT
fcis-26720	107	100	literature	literature	NOUN
fcis-26720	108	1	[	[	X
fcis-26720	108	2	26	26	NUM
fcis-26720	108	3	]	]	PUNCT
fcis-26720	108	4	shows	show	VERB
fcis-26720	108	5	that	that	SCONJ
fcis-26720	108	6	tibetan	tibetan	PROPN
fcis-26720	108	7	has	have	VERB
fcis-26720	108	8	a	a	DET
fcis-26720	108	9	tendency	tendency	NOUN
fcis-26720	108	10	to	to	PART
fcis-26720	108	11	clear	clear	ADJ
fcis-26720	108	12	voiced	voice	VERB
fcis-26720	108	13	consonants	consonant	NOUN
fcis-26720	108	14	and	and	CCONJ
fcis-26720	108	15	simplify	simplify	ADJ
fcis-26720	108	16	complex	complex	ADJ
fcis-26720	108	17	consonants	consonant	NOUN
fcis-26720	108	18	;	;	PUNCT
fcis-26720	108	19	in	in	ADP
fcis-26720	108	20	terms	term	NOUN
fcis-26720	108	21	of	of	ADP
fcis-26720	108	22	vowels	vowel	NOUN
fcis-26720	108	23	,	,	PUNCT
fcis-26720	108	24	the	the	DET
fcis-26720	108	25	vowels	vowel	NOUN
fcis-26720	108	26	of	of	ADP
fcis-26720	108	27	unit	unit	NOUN
fcis-26720	108	28	sounds	sound	VERB
fcis-26720	108	29	,	,	PUNCT
fcis-26720	108	30	especially	especially	ADV
fcis-26720	108	31	the	the	DET
fcis-26720	108	32	vowels	vowel	NOUN
fcis-26720	108	33	of	of	ADP
fcis-26720	108	34	nasalized	nasalized	ADJ
fcis-26720	108	35	vowels	vowel	NOUN
fcis-26720	108	36	,	,	PUNCT
fcis-26720	108	37	have	have	AUX
fcis-26720	108	38	increased	increase	VERB
fcis-26720	108	39	,	,	PUNCT
fcis-26720	108	40	forming	form	VERB
fcis-26720	108	41	two	two	NUM
fcis-26720	108	42	kinds	kind	NOUN
fcis-26720	108	43	of	of	ADP
fcis-26720	108	44	complex	complex	ADJ
fcis-26720	108	45	vowel	vowel	NOUN
fcis-26720	108	46	vowels	vowel	NOUN
fcis-26720	108	47	:	:	PUNCT
fcis-26720	108	48	nasalized	nasalized	ADJ
fcis-26720	108	49	vowels	vowel	NOUN
fcis-26720	108	50	and	and	CCONJ
fcis-26720	108	51	non	non	ADJ
fcis-26720	108	52	-	-	ADJ
fcis-26720	108	53	nasalized	nasalized	ADJ
fcis-26720	108	54	vowels	vowel	NOUN
fcis-26720	108	55	.	.	PUNCT
fcis-26720	109	1	in	in	ADP
fcis-26720	109	2	addition	addition	NOUN
fcis-26720	109	3	,	,	PUNCT
fcis-26720	109	4	tibetan	tibetan	PROPN
fcis-26720	109	5	has	have	VERB
fcis-26720	109	6	four	four	NUM
fcis-26720	109	7	tones	tone	NOUN
fcis-26720	109	8	,	,	PUNCT
fcis-26720	109	9	and	and	CCONJ
fcis-26720	109	10	the	the	DET
fcis-26720	109	11	cadence	cadence	NOUN
fcis-26720	109	12	of	of	ADP
fcis-26720	109	13	tones	tone	NOUN
fcis-26720	109	14	can	can	AUX
fcis-26720	109	15	help	help	VERB
fcis-26720	109	16	it	it	PRON
fcis-26720	109	17	better	well	ADV
fcis-26720	109	18	express	express	VERB
fcis-26720	109	19	a	a	DET
fcis-26720	109	20	person	person	NOUN
fcis-26720	109	21	’s	’s	PART
fcis-26720	109	22	emotions	emotion	NOUN
fcis-26720	109	23	.	.	PUNCT
fcis-26720	110	1	overall	overall	ADV
fcis-26720	110	2	,	,	PUNCT
fcis-26720	110	3	it	it	PRON
fcis-26720	110	4	is	be	AUX
fcis-26720	110	5	feasible	feasible	ADJ
fcis-26720	110	6	to	to	PART
fcis-26720	110	7	use	use	VERB
fcis-26720	110	8	tbsec001	tbsec001	NOUN
fcis-26720	110	9	for	for	ADP
fcis-26720	110	10	tibetan	tibetan	ADJ
fcis-26720	110	11	language	language	NOUN
fcis-26720	110	12	speech	speech	NOUN
fcis-26720	110	13	emotion	emotion	NOUN
fcis-26720	110	14	recognition	recognition	NOUN
fcis-26720	110	15	.	.	PUNCT
fcis-26720	111	1	emo	emo	NOUN
fcis-26720	111	2	-	-	PUNCT
fcis-26720	111	3	db	db	NOUN
fcis-26720	111	4	corpus	corpus	NOUN
fcis-26720	111	5	,	,	PUNCT
fcis-26720	111	6	also	also	ADV
fcis-26720	111	7	known	know	VERB
fcis-26720	111	8	as	as	ADP
fcis-26720	111	9	berlin	berlin	PROPN
fcis-26720	111	10	emotional	emotional	ADJ
fcis-26720	111	11	speech	speech	NOUN
fcis-26720	111	12	database	database	NOUN
fcis-26720	111	13	,	,	PUNCT
fcis-26720	111	14	is	be	AUX
fcis-26720	111	15	a	a	DET
fcis-26720	111	16	performance	performance	NOUN
fcis-26720	111	17	corpus	corpus	NOUN
fcis-26720	111	18	recorded	record	VERB
fcis-26720	111	19	by	by	ADP
fcis-26720	111	20	the	the	DET
fcis-26720	111	21	technical	technical	PROPN
fcis-26720	111	22	university	university	PROPN
fcis-26720	111	23	of	of	ADP
fcis-26720	111	24	berlin	berlin	PROPN
fcis-26720	111	25	,	,	PUNCT
fcis-26720	111	26	germany	germany	PROPN
fcis-26720	111	27	,	,	PUNCT
fcis-26720	111	28	containing	contain	VERB
fcis-26720	111	29	5	5	NUM
fcis-26720	111	30	men	man	NOUN
fcis-26720	111	31	and	and	CCONJ
fcis-26720	111	32	5	5	NUM
fcis-26720	111	33	women	woman	NOUN
fcis-26720	111	34	,	,	PUNCT
fcis-26720	111	35	cover	cover	VERB
fcis-26720	111	36	seven	seven	NUM
fcis-26720	111	37	different	different	ADJ
fcis-26720	111	38	emotional	emotional	ADJ
fcis-26720	111	39	types	type	NOUN
fcis-26720	111	40	:	:	PUNCT
fcis-26720	111	41	angry	angry	ADJ
fcis-26720	111	42	,	,	PUNCT
fcis-26720	111	43	boredom	boredom	NOUN
fcis-26720	111	44	,	,	PUNCT
fcis-26720	111	45	disgust	disgust	NOUN
fcis-26720	111	46	,	,	PUNCT
fcis-26720	111	47	fear	fear	NOUN
fcis-26720	111	48	,	,	PUNCT
fcis-26720	111	49	happy	happy	ADJ
fcis-26720	111	50	,	,	PUNCT
fcis-26720	111	51	neutral	neutral	ADJ
fcis-26720	111	52	and	and	CCONJ
fcis-26720	111	53	sad	sad	ADJ
fcis-26720	111	54	.	.	PUNCT
fcis-26720	112	1	in	in	ADP
fcis-26720	112	2	this	this	DET
fcis-26720	112	3	paper	paper	NOUN
fcis-26720	112	4	,	,	PUNCT
fcis-26720	112	5	five	five	NUM
fcis-26720	112	6	emotions	emotion	NOUN
fcis-26720	112	7	:	:	PUNCT
fcis-26720	112	8	angry	angry	ADJ
fcis-26720	112	9	,	,	PUNCT
fcis-26720	112	10	fear	fear	NOUN
fcis-26720	112	11	,	,	PUNCT
fcis-26720	112	12	happy	happy	ADJ
fcis-26720	112	13	,	,	PUNCT
fcis-26720	112	14	neutral	neutral	ADJ
fcis-26720	112	15	and	and	CCONJ
fcis-26720	112	16	sad	sad	ADJ
fcis-26720	112	17	,	,	PUNCT
fcis-26720	112	18	are	be	AUX
fcis-26720	112	19	selected	select	VERB
fcis-26720	112	20	to	to	PART
fcis-26720	112	21	evaluate	evaluate	VERB
fcis-26720	112	22	the	the	DET
fcis-26720	112	23	performance	performance	NOUN
fcis-26720	112	24	of	of	ADP
fcis-26720	112	25	the	the	DET
fcis-26720	112	26	model	model	NOUN
fcis-26720	112	27	.	.	PUNCT
fcis-26720	113	1	the	the	DET
fcis-26720	113	2	full	full	ADJ
fcis-26720	113	3	name	name	NOUN
fcis-26720	113	4	of	of	ADP
fcis-26720	113	5	ravdess	ravdess	NOUN
fcis-26720	113	6	corpus	corpus	NOUN
fcis-26720	113	7	is	be	AUX
fcis-26720	113	8	ryerson	ryerson	NOUN
fcis-26720	113	9	audiovisual	audiovisual	ADJ
fcis-26720	113	10	database	database	NOUN
fcis-26720	113	11	of	of	ADP
fcis-26720	113	12	emotional	emotional	ADJ
fcis-26720	113	13	speech	speech	NOUN
fcis-26720	113	14	and	and	CCONJ
fcis-26720	113	15	song	song	NOUN
fcis-26720	113	16	.	.	PUNCT
fcis-26720	114	1	it	it	PRON
fcis-26720	114	2	is	be	AUX
fcis-26720	114	3	a	a	DET
fcis-26720	114	4	widely	widely	ADV
fcis-26720	114	5	used	use	VERB
fcis-26720	114	6	speech	speech	NOUN
fcis-26720	114	7	and	and	CCONJ
fcis-26720	114	8	emotional	emotional	ADJ
fcis-26720	114	9	corpus	corpus	NOUN
fcis-26720	114	10	recorded	record	VERB
fcis-26720	114	11	by	by	ADP
fcis-26720	114	12	a	a	DET
fcis-26720	114	13	total	total	NOUN
fcis-26720	114	14	of	of	ADP
fcis-26720	114	15	24	24	NUM
fcis-26720	114	16	actors	actor	NOUN
fcis-26720	114	17	including	include	VERB
fcis-26720	114	18	12	12	NUM
fcis-26720	114	19	men	man	NOUN
fcis-26720	114	20	and	and	CCONJ
fcis-26720	114	21	12	12	NUM
fcis-26720	114	22	women	woman	NOUN
fcis-26720	114	23	.	.	PUNCT
fcis-26720	115	1	the	the	DET
fcis-26720	115	2	corpus	corpus	NOUN
fcis-26720	115	3	is	be	AUX
fcis-26720	115	4	divided	divide	VERB
fcis-26720	115	5	into	into	ADP
fcis-26720	115	6	speech	speech	NOUN
fcis-26720	115	7	part	part	NOUN
fcis-26720	115	8	and	and	CCONJ
fcis-26720	115	9	song	song	ADJ
fcis-26720	115	10	part	part	NOUN
fcis-26720	115	11	.	.	PUNCT
fcis-26720	116	1	this	this	DET
fcis-26720	116	2	paper	paper	NOUN
fcis-26720	116	3	uses	use	VERB
fcis-26720	116	4	the	the	DET
fcis-26720	116	5	corpus	corpus	NOUN
fcis-26720	116	6	of	of	ADP
fcis-26720	116	7	songs	song	NOUN
fcis-26720	116	8	,	,	PUNCT
fcis-26720	116	9	and	and	CCONJ
fcis-26720	116	10	there	there	PRON
fcis-26720	116	11	are	be	VERB
fcis-26720	116	12	six	six	NUM
fcis-26720	116	13	emotions	emotion	NOUN
fcis-26720	116	14	:	:	PUNCT
fcis-26720	116	15	angry	angry	ADJ
fcis-26720	116	16	,	,	PUNCT
fcis-26720	116	17	calm	calm	ADJ
fcis-26720	116	18	,	,	PUNCT
fcis-26720	116	19	fear	fear	NOUN
fcis-26720	116	20	,	,	PUNCT
fcis-26720	116	21	happy	happy	ADJ
fcis-26720	116	22	,	,	PUNCT
fcis-26720	116	23	neutral	neutral	ADJ
fcis-26720	116	24	and	and	CCONJ
fcis-26720	116	25	sad	sad	ADJ
fcis-26720	116	26	.	.	PUNCT
fcis-26720	117	1	(	(	PUNCT
fcis-26720	117	2	2	2	X
fcis-26720	117	3	)	)	PUNCT
fcis-26720	117	4	experimental	experimental	ADJ
fcis-26720	117	5	preparation	preparation	NOUN
fcis-26720	117	6	and	and	CCONJ
fcis-26720	117	7	evaluation	evaluation	NOUN
fcis-26720	117	8	methods	method	NOUN
fcis-26720	117	9	in	in	ADP
fcis-26720	117	10	this	this	DET
fcis-26720	117	11	paper	paper	NOUN
fcis-26720	117	12	,	,	PUNCT
fcis-26720	117	13	nvidia	nvidia	PROPN
fcis-26720	117	14	tesla	tesla	PROPN
fcis-26720	117	15	p40	p40	NOUN
fcis-26720	117	16	is	be	AUX
fcis-26720	117	17	used	use	VERB
fcis-26720	117	18	as	as	ADP
fcis-26720	117	19	the	the	DET
fcis-26720	117	20	hardware	hardware	NOUN
fcis-26720	117	21	platform	platform	NOUN
fcis-26720	117	22	,	,	PUNCT
fcis-26720	117	23	and	and	CCONJ
fcis-26720	117	24	tensorflow	tensorflow	NOUN
fcis-26720	117	25	is	be	AUX
fcis-26720	117	26	selected	select	VERB
fcis-26720	117	27	as	as	ADP
fcis-26720	117	28	the	the	DET
fcis-26720	117	29	deep	deep	ADJ
fcis-26720	117	30	learning	learning	NOUN
fcis-26720	117	31	framework	framework	NOUN
fcis-26720	117	32	.	.	PUNCT
fcis-26720	118	1	all	all	DET
fcis-26720	118	2	corpus	corpus	NOUN
fcis-26720	118	3	files	file	NOUN
fcis-26720	118	4	are	be	AUX
fcis-26720	118	5	read	read	VERB
fcis-26720	118	6	at	at	ADP
fcis-26720	118	7	a	a	DET
fcis-26720	118	8	sampling	sample	VERB
fcis-26720	118	9	rate	rate	NOUN
fcis-26720	118	10	of	of	ADP
fcis-26720	118	11	16k	16k	NUM
fcis-26720	118	12	,	,	PUNCT
fcis-26720	118	13	and	and	CCONJ
fcis-26720	118	14	the	the	DET
fcis-26720	118	15	mfcc	mfcc	NOUN
fcis-26720	118	16	of	of	ADP
fcis-26720	118	17	audio	audio	ADJ
fcis-26720	118	18	segments	segment	NOUN
fcis-26720	118	19	are	be	AUX
fcis-26720	118	20	extracted	extract	VERB
fcis-26720	118	21	by	by	ADP
fcis-26720	118	22	using	use	VERB
fcis-26720	118	23	librosa	librosa	ADJ
fcis-26720	118	24	library	library	NOUN
fcis-26720	118	25	.	.	PUNCT
fcis-26720	119	1	in	in	ADP
fcis-26720	119	2	addition	addition	NOUN
fcis-26720	119	3	,	,	PUNCT
fcis-26720	119	4	the	the	DET
fcis-26720	119	5	data	datum	NOUN
fcis-26720	119	6	set	set	VERB
fcis-26720	119	7	is	be	AUX
fcis-26720	119	8	divided	divide	VERB
fcis-26720	119	9	into	into	ADP
fcis-26720	119	10	training	training	NOUN
fcis-26720	119	11	set	set	NOUN
fcis-26720	119	12	and	and	CCONJ
fcis-26720	119	13	verification	verification	NOUN
fcis-26720	119	14	set	set	NOUN
fcis-26720	119	15	according	accord	VERB
fcis-26720	119	16	to	to	ADP
fcis-26720	119	17	the	the	DET
fcis-26720	119	18	ratio	ratio	NOUN
fcis-26720	119	19	of	of	ADP
fcis-26720	119	20	4	4	NUM
fcis-26720	119	21	:	:	SYM
fcis-26720	119	22	1	1	NUM
fcis-26720	119	23	.	.	X
fcis-26720	120	1	the	the	DET
fcis-26720	120	2	entire	entire	ADJ
fcis-26720	120	3	training	training	NOUN
fcis-26720	120	4	process	process	NOUN
fcis-26720	120	5	is	be	AUX
fcis-26720	120	6	set	set	VERB
fcis-26720	120	7	to	to	ADP
fcis-26720	120	8	300	300	NUM
fcis-26720	120	9	rounds	round	NOUN
fcis-26720	120	10	,	,	PUNCT
fcis-26720	120	11	and	and	CCONJ
fcis-26720	120	12	the	the	DET
fcis-26720	120	13	adam	adam	PROPN
fcis-26720	120	14	optimizer	optimizer	NOUN
fcis-26720	120	15	is	be	AUX
fcis-26720	120	16	used	use	VERB
fcis-26720	120	17	to	to	PART
fcis-26720	120	18	optimize	optimize	VERB
fcis-26720	120	19	the	the	DET
fcis-26720	120	20	model	model	NOUN
fcis-26720	120	21	.	.	PUNCT
fcis-26720	121	1	in	in	ADP
fcis-26720	121	2	order	order	NOUN
fcis-26720	121	3	to	to	PART
fcis-26720	121	4	effectively	effectively	ADV
fcis-26720	121	5	deal	deal	VERB
fcis-26720	121	6	with	with	ADP
fcis-26720	121	7	the	the	DET
fcis-26720	121	8	gradient	gradient	NOUN
fcis-26720	121	9	instability	instability	NOUN
fcis-26720	121	10	in	in	ADP
fcis-26720	121	11	the	the	DET
fcis-26720	121	12	training	training	NOUN
fcis-26720	121	13	process	process	NOUN
fcis-26720	121	14	,	,	PUNCT
fcis-26720	121	15	this	this	DET
fcis-26720	121	16	paper	paper	NOUN
fcis-26720	121	17	adopts	adopt	VERB
fcis-26720	121	18	a	a	DET
fcis-26720	121	19	strategy	strategy	NOUN
fcis-26720	121	20	of	of	ADP
fcis-26720	121	21	dynamically	dynamically	ADV
fcis-26720	121	22	adjusting	adjust	VERB
fcis-26720	121	23	the	the	DET
fcis-26720	121	24	learning	learning	NOUN
fcis-26720	121	25	rate	rate	NOUN
fcis-26720	121	26	:	:	PUNCT
fcis-26720	121	27	every	every	DET
fcis-26720	121	28	100	100	NUM
fcis-26720	121	29	training	training	NOUN
fcis-26720	121	30	rounds	round	NOUN
fcis-26720	121	31	,	,	PUNCT
fcis-26720	121	32	the	the	DET
fcis-26720	121	33	learning	learning	NOUN
fcis-26720	121	34	rate	rate	NOUN
fcis-26720	121	35	is	be	AUX
fcis-26720	121	36	halved	halve	VERB
fcis-26720	121	37	.	.	PUNCT
fcis-26720	122	1	this	this	DET
fcis-26720	122	2	strategy	strategy	NOUN
fcis-26720	122	3	aims	aim	VERB
fcis-26720	122	4	to	to	PART
fcis-26720	122	5	maintain	maintain	VERB
fcis-26720	122	6	the	the	DET
fcis-26720	122	7	convergence	convergence	NOUN
fcis-26720	122	8	speed	speed	NOUN
fcis-26720	122	9	of	of	ADP
fcis-26720	122	10	the	the	DET
fcis-26720	122	11	model	model	NOUN
fcis-26720	122	12	while	while	SCONJ
fcis-26720	122	13	ensuring	ensure	VERB
fcis-26720	122	14	the	the	DET
fcis-26720	122	15	stability	stability	NOUN
fcis-26720	122	16	and	and	CCONJ
fcis-26720	122	17	generalization	generalization	NOUN
fcis-26720	122	18	ability	ability	NOUN
fcis-26720	122	19	of	of	ADP
fcis-26720	122	20	training	training	NOUN
fcis-26720	122	21	.	.	PUNCT
fcis-26720	123	1	according	accord	VERB
fcis-26720	123	2	to	to	ADP
fcis-26720	123	3	the	the	DET
fcis-26720	123	4	characteristics	characteristic	NOUN
fcis-26720	123	5	of	of	ADP
fcis-26720	123	6	each	each	DET
fcis-26720	123	7	corpus	corpus	NOUN
fcis-26720	123	8	,	,	PUNCT
fcis-26720	123	9	different	different	ADJ
fcis-26720	123	10	initial	initial	ADJ
fcis-26720	123	11	learning	learning	NOUN
fcis-26720	123	12	rates	rate	NOUN
fcis-26720	123	13	are	be	AUX
fcis-26720	123	14	applied	apply	VERB
fcis-26720	123	15	.	.	PUNCT
fcis-26720	124	1	tables	table	NOUN
fcis-26720	124	2	4	4	NUM
fcis-26720	124	3	and	and	CCONJ
fcis-26720	124	4	5	5	NUM
fcis-26720	124	5	detail	detail	NOUN
fcis-26720	124	6	the	the	DET
fcis-26720	124	7	usage	usage	NOUN
fcis-26720	124	8	details	detail	NOUN
fcis-26720	124	9	of	of	ADP
fcis-26720	124	10	the	the	DET
fcis-26720	124	11	corpus	corpus	NOUN
fcis-26720	124	12	,	,	PUNCT
fcis-26720	124	13	as	as	ADV
fcis-26720	124	14	well	well	ADV
fcis-26720	124	15	as	as	ADP
fcis-26720	124	16	the	the	DET
fcis-26720	124	17	dimensions	dimension	NOUN
fcis-26720	124	18	and	and	CCONJ
fcis-26720	124	19	related	related	ADJ
fcis-26720	124	20	parameters	parameter	NOUN
fcis-26720	124	21	of	of	ADP
fcis-26720	124	22	the	the	DET
fcis-26720	124	23	extracted	extract	VERB
fcis-26720	124	24	mfcc	mfcc	NOUN
fcis-26720	124	25	.	.	PUNCT
fcis-26720	125	1	table	table	NOUN
fcis-26720	125	2	4	4	NUM
fcis-26720	125	3	.	.	PUNCT
fcis-26720	126	1	corpus	corpus	NOUN
fcis-26720	126	2	emotion	emotion	NOUN
fcis-26720	126	3	and	and	CCONJ
fcis-26720	126	4	quantity	quantity	NOUN
fcis-26720	126	5	statistics	statistic	NOUN
fcis-26720	126	6	table	table	NOUN
fcis-26720	126	7	corpus	corpus	PROPN
fcis-26720	126	8	angry	angry	ADJ
fcis-26720	126	9	calm	calm	NOUN
fcis-26720	126	10	fear	fear	VERB
fcis-26720	126	11	happy	happy	ADJ
fcis-26720	126	12	neutral	neutral	ADJ
fcis-26720	126	13	sad	sad	ADJ
fcis-26720	126	14	number	number	NOUN
fcis-26720	126	15	of	of	ADP
fcis-26720	126	16	training	training	NOUN
fcis-26720	126	17	number	number	NOUN
fcis-26720	126	18	of	of	ADP
fcis-26720	126	19	testing	test	VERB
fcis-26720	126	20	total	total	ADJ
fcis-26720	126	21	tbsec001	tbsec001	PROPN
fcis-26720	126	22	1200	1200	NUM
fcis-26720	126	23	1200	1200	NUM
fcis-26720	126	24	1200	1200	NUM
fcis-26720	126	25	1200	1200	NUM
fcis-26720	126	26	1200	1200	NUM
fcis-26720	126	27	4800	4800	NUM
fcis-26720	126	28	1200	1200	NUM
fcis-26720	126	29	6000	6000	NUM
fcis-26720	126	30	emo	emo	NOUN
fcis-26720	126	31	-	-	PUNCT
fcis-26720	126	32	db	db	PROPN
fcis-26720	126	33	129	129	NUM
fcis-26720	126	34	69	69	NUM
fcis-26720	126	35	72	72	NUM
fcis-26720	126	36	99	99	NUM
fcis-26720	126	37	135	135	NUM
fcis-26720	126	38	404	404	NUM
fcis-26720	126	39	100	100	NUM
fcis-26720	126	40	504	504	NUM
fcis-26720	126	41	ravdess	ravdess	NOUN
fcis-26720	126	42	184	184	NUM
fcis-26720	126	43	184	184	NUM
fcis-26720	126	44	184	184	NUM
fcis-26720	126	45	184	184	NUM
fcis-26720	126	46	92	92	NUM
fcis-26720	126	47	184	184	NUM
fcis-26720	126	48	663	663	NUM
fcis-26720	126	49	165	165	NUM
fcis-26720	126	50	828	828	NUM
fcis-26720	126	51	table	table	NOUN
fcis-26720	126	52	5	5	NUM
fcis-26720	126	53	.	.	PUNCT
fcis-26720	126	54	mfcc	mfcc	NOUN
fcis-26720	126	55	dimensions	dimension	NOUN
fcis-26720	126	56	and	and	CCONJ
fcis-26720	126	57	parameter	parameter	NOUN
fcis-26720	126	58	configuration	configuration	NOUN
fcis-26720	126	59	extracted	extract	VERB
fcis-26720	126	60	from	from	ADP
fcis-26720	126	61	each	each	DET
fcis-26720	126	62	corpus	corpus	NOUN
fcis-26720	126	63	corpus	corpus	PROPN
fcis-26720	126	64	mfcc	mfcc	NOUN
fcis-26720	126	65	dimension	dimension	PROPN
fcis-26720	126	66	fft	fft	PROPN
fcis-26720	126	67	window	window	NOUN
fcis-26720	126	68	frame	frame	NOUN
fcis-26720	126	69	shift	shift	NOUN
fcis-26720	126	70	mel	mel	PROPN
fcis-26720	126	71	filter	filter	NOUN
fcis-26720	126	72	learning	learn	VERB
fcis-26720	126	73	rate	rate	NOUN
fcis-26720	126	74	tbsec001	tbsec001	NUM
fcis-26720	126	75	45	45	NUM
fcis-26720	126	76	2048	2048	NUM
fcis-26720	126	77	512	512	NUM
fcis-26720	126	78	45	45	NUM
fcis-26720	126	79	0.005	0.005	NUM
fcis-26720	126	80	emo	emo	NOUN
fcis-26720	126	81	-	-	PUNCT
fcis-26720	126	82	db	db	NOUN
fcis-26720	126	83	42	42	NUM
fcis-26720	126	84	800	800	NUM
fcis-26720	126	85	200	200	NUM
fcis-26720	126	86	42	42	NUM
fcis-26720	126	87	0.0001	0.0001	NUM
fcis-26720	126	88	ravdess	ravdess	NOUN
fcis-26720	126	89	45	45	NUM
fcis-26720	126	90	800	800	NUM
fcis-26720	126	91	200	200	NUM
fcis-26720	126	92	45	45	NUM
fcis-26720	126	93	0.0001	0.0001	NUM
fcis-26720	126	94	the	the	DET
fcis-26720	126	95	indexes	index	NOUN
fcis-26720	126	96	used	use	VERB
fcis-26720	126	97	in	in	ADP
fcis-26720	126	98	the	the	DET
fcis-26720	126	99	model	model	NOUN
fcis-26720	126	100	evaluation	evaluation	NOUN
fcis-26720	126	101	are	be	AUX
fcis-26720	126	102	unweighted	unweighted	ADJ
fcis-26720	126	103	accuracy	accuracy	NOUN
fcis-26720	126	104	(	(	PUNCT
fcis-26720	126	105	ua	ua	NOUN
fcis-26720	126	106	)	)	PUNCT
fcis-26720	126	107	and	and	CCONJ
fcis-26720	126	108	weighted	weight	VERB
fcis-26720	126	109	accuracy	accuracy	NOUN
fcis-26720	126	110	(	(	PUNCT
fcis-26720	126	111	wa	wa	NOUN
fcis-26720	126	112	)	)	PUNCT
fcis-26720	126	113	.	.	PUNCT
fcis-26720	127	1	the	the	DET
fcis-26720	127	2	formula	formula	NOUN
fcis-26720	127	3	is	be	AUX
fcis-26720	127	4	as	as	SCONJ
fcis-26720	127	5	follows	follow	VERB
fcis-26720	127	6	:	:	PUNCT
fcis-26720	127	7	ua	ua	PROPN
fcis-26720	127	8	t	t	PROPN
fcis-26720	127	9	n	n	ADV
fcis-26720	127	10			PROPN
fcis-26720	127	11	(	(	PUNCT
fcis-26720	127	12	7	7	X
fcis-26720	127	13	)	)	PUNCT
fcis-26720	127	14	i	i	PRON
fcis-26720	127	15	itwa	itwa	VERB
fcis-26720	127	16	n	n	CCONJ
fcis-26720	127	17			X
fcis-26720	127	18			NUM
fcis-26720	127	19			X
fcis-26720	127	20	(	(	PUNCT
fcis-26720	127	21	8)	8)	NUM
fcis-26720	127	22	among	among	ADP
fcis-26720	127	23	them	they	PRON
fcis-26720	127	24	,	,	PUNCT
fcis-26720	127	25	represents	represent	VERB
fcis-26720	127	26	the	the	DET
fcis-26720	127	27	total	total	ADJ
fcis-26720	127	28	number	number	NOUN
fcis-26720	127	29	of	of	ADP
fcis-26720	127	30	samples	sample	NOUN
fcis-26720	127	31	,	,	PUNCT
fcis-26720	127	32	represents	represent	VERB
fcis-26720	127	33	the	the	DET
fcis-26720	127	34	total	total	ADJ
fcis-26720	127	35	number	number	NOUN
fcis-26720	127	36	of	of	ADP
fcis-26720	127	37	samples	sample	NOUN
fcis-26720	127	38	with	with	ADP
fcis-26720	127	39	correct	correct	ADJ
fcis-26720	127	40	prediction	prediction	NOUN
fcis-26720	127	41	,	,	PUNCT
fcis-26720	127	42	represents	represent	VERB
fcis-26720	127	43	the	the	DET
fcis-26720	127	44	proportion	proportion	NOUN
fcis-26720	127	45	of	of	ADP
fcis-26720	127	46	type	type	NOUN
fcis-26720	127	47	of	of	ADP
fcis-26720	127	48	emotion	emotion	NOUN
fcis-26720	127	49	in	in	ADP
fcis-26720	127	50	the	the	DET
fcis-26720	127	51	total	total	ADJ
fcis-26720	127	52	samples	sample	NOUN
fcis-26720	127	53	,	,	PUNCT
fcis-26720	127	54	and	and	CCONJ
fcis-26720	127	55	represents	represent	VERB
fcis-26720	127	56	the	the	DET
fcis-26720	127	57	number	number	NOUN
fcis-26720	127	58	of	of	ADP
fcis-26720	127	59	samples	sample	NOUN
fcis-26720	127	60	with	with	ADP
fcis-26720	127	61	correct	correct	ADJ
fcis-26720	127	62	prediction	prediction	NOUN
fcis-26720	127	63	for	for	ADP
fcis-26720	127	64	type	type	NOUN
fcis-26720	127	65	of	of	ADP
fcis-26720	127	66	emotion	emotion	NOUN
fcis-26720	127	67	.	.	PUNCT
fcis-26720	128	1	(	(	PUNCT
fcis-26720	128	2	3	3	X
fcis-26720	128	3	)	)	PUNCT
fcis-26720	128	4	experimental	experimental	ADJ
fcis-26720	128	5	results	result	NOUN
fcis-26720	128	6	and	and	CCONJ
fcis-26720	128	7	analysis	analysis	NOUN
fcis-26720	128	8	in	in	ADP
fcis-26720	128	9	the	the	DET
fcis-26720	128	10	ser	ser	NOUN
fcis-26720	128	11	task	task	PROPN
fcis-26720	128	12	experiment	experiment	NOUN
fcis-26720	128	13	for	for	ADP
fcis-26720	128	14	tbsec001	tbsec001	NOUN
fcis-26720	128	15	corpus	corpus	NOUN
fcis-26720	128	16	,	,	PUNCT
fcis-26720	128	17	the	the	DET
fcis-26720	128	18	training	training	NOUN
fcis-26720	128	19	process	process	NOUN
fcis-26720	128	20	of	of	ADP
fcis-26720	128	21	the	the	DET
fcis-26720	128	22	model	model	NOUN
fcis-26720	128	23	is	be	AUX
fcis-26720	128	24	shown	show	VERB
fcis-26720	128	25	in	in	ADP
fcis-26720	128	26	figure	figure	NOUN
fcis-26720	128	27	2	2	NUM
fcis-26720	128	28	.	.	PUNCT
fcis-26720	129	1	within	within	ADP
fcis-26720	129	2	the	the	DET
fcis-26720	129	3	first	first	ADJ
fcis-26720	129	4	100	100	NUM
fcis-26720	129	5	rounds	round	NOUN
fcis-26720	129	6	of	of	ADP
fcis-26720	129	7	training	training	NOUN
fcis-26720	129	8	,	,	PUNCT
fcis-26720	129	9	the	the	DET
fcis-26720	129	10	model	model	NOUN
fcis-26720	129	11	effectively	effectively	ADV
fcis-26720	129	12	learns	learn	VERB
fcis-26720	129	13	the	the	DET
fcis-26720	129	14	emotional	emotional	ADJ
fcis-26720	129	15	features	feature	NOUN
fcis-26720	129	16	in	in	ADP
fcis-26720	129	17	the	the	DET
fcis-26720	129	18	data	datum	NOUN
fcis-26720	129	19	,	,	PUNCT
fcis-26720	129	20	and	and	CCONJ
fcis-26720	129	21	the	the	DET
fcis-26720	129	22	test	test	NOUN
fcis-26720	129	23	curve	curve	NOUN
fcis-26720	129	24	converges	converge	VERB
fcis-26720	129	25	rapidly	rapidly	ADV
fcis-26720	129	26	.	.	PUNCT
fcis-26720	130	1	between	between	ADP
fcis-26720	130	2	100	100	NUM
fcis-26720	130	3	-	-	SYM
fcis-26720	130	4	300	300	NUM
fcis-26720	130	5	rounds	round	NOUN
fcis-26720	130	6	,	,	PUNCT
fcis-26720	130	7	although	although	SCONJ
fcis-26720	130	8	there	there	PRON
fcis-26720	130	9	were	be	VERB
fcis-26720	130	10	some	some	DET
fcis-26720	130	11	fluctuations	fluctuation	NOUN
fcis-26720	130	12	in	in	ADP
fcis-26720	130	13	the	the	DET
fcis-26720	130	14	test	test	NOUN
fcis-26720	130	15	part	part	NOUN
fcis-26720	130	16	,	,	PUNCT
fcis-26720	130	17	the	the	DET
fcis-26720	130	18	overall	overall	ADJ
fcis-26720	130	19	performance	performance	NOUN
fcis-26720	130	20	was	be	AUX
fcis-26720	130	21	still	still	ADV
fcis-26720	130	22	good	good	ADJ
fcis-26720	130	23	.	.	PUNCT
fcis-26720	131	1	the	the	DET
fcis-26720	131	2	final	final	ADJ
fcis-26720	131	3	model	model	NOUN
fcis-26720	131	4	achieved	achieve	VERB
fcis-26720	131	5	an	an	DET
fcis-26720	131	6	accuracy	accuracy	NOUN
fcis-26720	131	7	rate	rate	NOUN
fcis-26720	131	8	of	of	ADP
fcis-26720	131	9	81.52	81.52	NUM
fcis-26720	131	10	%	%	NOUN
fcis-26720	131	11	,	,	PUNCT
fcis-26720	131	12	and	and	CCONJ
fcis-26720	131	13	the	the	DET
fcis-26720	131	14	loss	loss	NOUN
fcis-26720	131	15	value	value	NOUN
fcis-26720	131	16	was	be	AUX
fcis-26720	131	17	stable	stable	ADJ
fcis-26720	131	18	around	around	ADP
fcis-26720	131	19	0.16	0.16	NUM
fcis-26720	131	20	.	.	PUNCT
fcis-26720	132	1	89	89	NUM
fcis-26720	133	1	(	(	PUNCT
fcis-26720	133	2	a	a	X
fcis-26720	133	3	)	)	PUNCT
fcis-26720	133	4	tbsec001	tbsec001	NOUN
fcis-26720	133	5	corpus	corpus	NOUN
fcis-26720	133	6	training	training	NOUN
fcis-26720	133	7	and	and	CCONJ
fcis-26720	133	8	testing	testing	NOUN
fcis-26720	133	9	accuracy	accuracy	NOUN
fcis-26720	133	10	curve	curve	NOUN
fcis-26720	133	11	(	(	PUNCT
fcis-26720	133	12	b	b	NOUN
fcis-26720	133	13	)	)	PUNCT
fcis-26720	133	14	tbsec001	tbsec001	NOUN
fcis-26720	133	15	corpus	corpus	NOUN
fcis-26720	133	16	training	training	NOUN
fcis-26720	133	17	and	and	CCONJ
fcis-26720	133	18	validation	validation	NOUN
fcis-26720	133	19	loss	loss	NOUN
fcis-26720	133	20	curves	curve	NOUN
fcis-26720	133	21	fig	fig	NOUN
fcis-26720	133	22	2	2	NUM
fcis-26720	133	23	.	.	PUNCT
fcis-26720	133	24	accuracy	accuracy	NOUN
fcis-26720	133	25	and	and	CCONJ
fcis-26720	133	26	loss	loss	NOUN
fcis-26720	133	27	curve	curve	NOUN
fcis-26720	133	28	of	of	ADP
fcis-26720	133	29	tbsec001	tbsec001	NOUN
fcis-26720	133	30	corpus	corpus	NOUN
fcis-26720	133	31	in	in	ADP
fcis-26720	133	32	order	order	NOUN
fcis-26720	133	33	to	to	PART
fcis-26720	133	34	explain	explain	VERB
fcis-26720	133	35	the	the	DET
fcis-26720	133	36	prediction	prediction	NOUN
fcis-26720	133	37	of	of	ADP
fcis-26720	133	38	each	each	DET
fcis-26720	133	39	emotion	emotion	NOUN
fcis-26720	133	40	in	in	ADP
fcis-26720	133	41	more	more	ADJ
fcis-26720	133	42	detail	detail	NOUN
fcis-26720	133	43	,	,	PUNCT
fcis-26720	133	44	the	the	DET
fcis-26720	133	45	confusion	confusion	NOUN
fcis-26720	133	46	matrix	matrix	NOUN
fcis-26720	133	47	is	be	AUX
fcis-26720	133	48	drawn	draw	VERB
fcis-26720	133	49	in	in	ADP
fcis-26720	133	50	this	this	DET
fcis-26720	133	51	paper	paper	NOUN
fcis-26720	133	52	,	,	PUNCT
fcis-26720	133	53	as	as	SCONJ
fcis-26720	133	54	shown	show	VERB
fcis-26720	133	55	in	in	ADP
fcis-26720	133	56	figure	figure	NOUN
fcis-26720	133	57	3	3	NUM
fcis-26720	133	58	.	.	PUNCT
fcis-26720	134	1	among	among	ADP
fcis-26720	134	2	them	they	PRON
fcis-26720	134	3	,	,	PUNCT
fcis-26720	134	4	the	the	DET
fcis-26720	134	5	classification	classification	NOUN
fcis-26720	134	6	effect	effect	NOUN
fcis-26720	134	7	of	of	ADP
fcis-26720	134	8	angry	angry	ADJ
fcis-26720	134	9	emotions	emotion	NOUN
fcis-26720	134	10	is	be	AUX
fcis-26720	134	11	the	the	DET
fcis-26720	134	12	best	good	ADJ
fcis-26720	134	13	,	,	PUNCT
fcis-26720	134	14	and	and	CCONJ
fcis-26720	134	15	a	a	DET
fcis-26720	134	16	total	total	NOUN
fcis-26720	134	17	of	of	ADP
fcis-26720	134	18	83.92	83.92	NUM
fcis-26720	134	19	%	%	NOUN
fcis-26720	134	20	of	of	ADP
fcis-26720	134	21	angry	angry	ADJ
fcis-26720	134	22	emotions	emotion	NOUN
fcis-26720	134	23	are	be	AUX
fcis-26720	134	24	successfully	successfully	ADV
fcis-26720	134	25	identified	identify	VERB
fcis-26720	134	26	.	.	PUNCT
fcis-26720	135	1	11.61	11.61	NUM
fcis-26720	135	2	%	%	NOUN
fcis-26720	135	3	of	of	ADP
fcis-26720	135	4	fear	fear	NOUN
fcis-26720	135	5	was	be	AUX
fcis-26720	135	6	identified	identify	VERB
fcis-26720	135	7	as	as	ADP
fcis-26720	135	8	sad	sad	ADJ
fcis-26720	135	9	.	.	PUNCT
fcis-26720	136	1	overall	overall	ADJ
fcis-26720	136	2	,	,	PUNCT
fcis-26720	136	3	tbsec001	tbsec001	NOUN
fcis-26720	136	4	performs	perform	VERB
fcis-26720	136	5	well	well	ADV
fcis-26720	136	6	on	on	ADP
fcis-26720	136	7	the	the	DET
fcis-26720	136	8	model	model	NOUN
fcis-26720	136	9	in	in	ADP
fcis-26720	136	10	this	this	DET
fcis-26720	136	11	paper	paper	NOUN
fcis-26720	136	12	.	.	PUNCT
fcis-26720	137	1	fig	fig	NOUN
fcis-26720	137	2	3	3	NUM
fcis-26720	137	3	.	.	PUNCT
fcis-26720	137	4	confusion	confusion	NOUN
fcis-26720	137	5	matrix	matrix	NOUN
fcis-26720	137	6	of	of	ADP
fcis-26720	137	7	this	this	DET
fcis-26720	137	8	model	model	NOUN
fcis-26720	137	9	on	on	ADP
fcis-26720	137	10	tbsec001	tbsec001	PUNCT
fcis-26720	137	11	corpus	corpus	NOUN
fcis-26720	137	12	the	the	DET
fcis-26720	137	13	training	training	NOUN
fcis-26720	137	14	process	process	NOUN
fcis-26720	137	15	of	of	ADP
fcis-26720	137	16	the	the	DET
fcis-26720	137	17	emo	emo	NOUN
fcis-26720	137	18	-	-	PUNCT
fcis-26720	137	19	db	db	NOUN
fcis-26720	137	20	corpus	corpus	NOUN
fcis-26720	137	21	is	be	AUX
fcis-26720	137	22	shown	show	VERB
fcis-26720	137	23	in	in	ADP
fcis-26720	137	24	figure	figure	NOUN
fcis-26720	137	25	4	4	NUM
fcis-26720	137	26	.	.	PUNCT
fcis-26720	138	1	it	it	PRON
fcis-26720	138	2	can	can	AUX
fcis-26720	138	3	be	be	AUX
fcis-26720	138	4	seen	see	VERB
fcis-26720	138	5	from	from	ADP
fcis-26720	138	6	the	the	DET
fcis-26720	138	7	figure	figure	NOUN
fcis-26720	138	8	that	that	SCONJ
fcis-26720	138	9	with	with	ADP
fcis-26720	138	10	the	the	DET
fcis-26720	138	11	increase	increase	NOUN
fcis-26720	138	12	of	of	ADP
fcis-26720	138	13	training	training	NOUN
fcis-26720	138	14	rounds	round	NOUN
fcis-26720	138	15	,	,	PUNCT
fcis-26720	138	16	the	the	DET
fcis-26720	138	17	ua	ua	PROPN
fcis-26720	138	18	gradually	gradually	ADV
fcis-26720	138	19	increases	increase	VERB
fcis-26720	138	20	,	,	PUNCT
fcis-26720	138	21	and	and	CCONJ
fcis-26720	138	22	the	the	DET
fcis-26720	138	23	maximum	maximum	ADJ
fcis-26720	138	24	ua	ua	PROPN
fcis-26720	138	25	value	value	NOUN
fcis-26720	138	26	of	of	ADP
fcis-26720	138	27	85.63	85.63	NUM
fcis-26720	138	28	%	%	NOUN
fcis-26720	138	29	is	be	AUX
fcis-26720	138	30	obtained	obtain	VERB
fcis-26720	138	31	when	when	SCONJ
fcis-26720	138	32	the	the	DET
fcis-26720	138	33	training	training	NOUN
fcis-26720	138	34	rounds	round	NOUN
fcis-26720	138	35	reach	reach	VERB
fcis-26720	138	36	about	about	ADV
fcis-26720	138	37	270	270	NUM
fcis-26720	138	38	.	.	PUNCT
fcis-26720	139	1	compared	compare	VERB
fcis-26720	139	2	with	with	ADP
fcis-26720	139	3	the	the	DET
fcis-26720	139	4	accuracy	accuracy	NOUN
fcis-26720	139	5	curve	curve	NOUN
fcis-26720	139	6	,	,	PUNCT
fcis-26720	139	7	the	the	DET
fcis-26720	139	8	loss	loss	NOUN
fcis-26720	139	9	value	value	NOUN
fcis-26720	139	10	of	of	ADP
fcis-26720	139	11	the	the	DET
fcis-26720	139	12	loss	loss	NOUN
fcis-26720	139	13	curve	curve	NOUN
fcis-26720	139	14	fluctuates	fluctuate	NOUN
fcis-26720	139	15	greatly	greatly	ADV
fcis-26720	139	16	,	,	PUNCT
fcis-26720	139	17	and	and	CCONJ
fcis-26720	139	18	the	the	DET
fcis-26720	139	19	loss	loss	NOUN
fcis-26720	139	20	value	value	NOUN
fcis-26720	139	21	fluctuates	fluctuate	VERB
fcis-26720	139	22	between	between	ADP
fcis-26720	139	23	0.15	0.15	NUM
fcis-26720	139	24	and	and	CCONJ
fcis-26720	139	25	0.19	0.19	NUM
fcis-26720	139	26	.	.	PUNCT
fcis-26720	140	1	(	(	PUNCT
fcis-26720	140	2	a	a	X
fcis-26720	140	3	)	)	PUNCT
fcis-26720	140	4	training	training	NOUN
fcis-26720	140	5	and	and	CCONJ
fcis-26720	140	6	validation	validation	NOUN
fcis-26720	140	7	accuracy	accuracy	NOUN
fcis-26720	140	8	curves	curve	NOUN
fcis-26720	140	9	of	of	ADP
fcis-26720	140	10	emo	emo	NOUN
fcis-26720	140	11	-	-	PUNCT
fcis-26720	140	12	db	db	NOUN
fcis-26720	140	13	corpus	corpus	NOUN
fcis-26720	140	14	(	(	PUNCT
fcis-26720	140	15	b	b	NOUN
fcis-26720	140	16	)	)	PUNCT
fcis-26720	140	17	training	training	NOUN
fcis-26720	140	18	and	and	CCONJ
fcis-26720	140	19	validation	validation	NOUN
fcis-26720	140	20	loss	loss	NOUN
fcis-26720	140	21	value	value	NOUN
fcis-26720	140	22	curves	curve	NOUN
fcis-26720	140	23	of	of	ADP
fcis-26720	140	24	emo	emo	NOUN
fcis-26720	140	25	-	-	PUNCT
fcis-26720	140	26	db	db	NOUN
fcis-26720	140	27	corpus	corpus	NOUN
fcis-26720	140	28	fig	fig	NOUN
fcis-26720	140	29	4	4	NUM
fcis-26720	140	30	.	.	PUNCT
fcis-26720	140	31	accuracy	accuracy	NOUN
fcis-26720	140	32	and	and	CCONJ
fcis-26720	140	33	loss	loss	NOUN
fcis-26720	140	34	curve	curve	NOUN
fcis-26720	140	35	of	of	ADP
fcis-26720	140	36	emo	emo	NOUN
fcis-26720	140	37	-	-	PUNCT
fcis-26720	140	38	db	db	NOUN
fcis-26720	140	39	corpus	corpus	NOUN
fcis-26720	140	40	the	the	DET
fcis-26720	140	41	confusion	confusion	NOUN
fcis-26720	140	42	matrix	matrix	NOUN
fcis-26720	140	43	in	in	ADP
fcis-26720	140	44	fig	fig	NOUN
fcis-26720	140	45	.	.	PUNCT
fcis-26720	141	1	5	5	NUM
fcis-26720	141	2	describes	describe	VERB
fcis-26720	141	3	the	the	DET
fcis-26720	141	4	classification	classification	NOUN
fcis-26720	141	5	of	of	ADP
fcis-26720	141	6	emo	emo	NOUN
fcis-26720	141	7	-	-	PUNCT
fcis-26720	141	8	db	db	NOUN
fcis-26720	141	9	corpus	corpus	NOUN
fcis-26720	141	10	by	by	ADP
fcis-26720	141	11	the	the	DET
fcis-26720	141	12	model	model	NOUN
fcis-26720	141	13	in	in	ADP
fcis-26720	141	14	this	this	DET
fcis-26720	141	15	paper	paper	NOUN
fcis-26720	141	16	,	,	PUNCT
fcis-26720	141	17	in	in	ADP
fcis-26720	141	18	which	which	PRON
fcis-26720	141	19	the	the	DET
fcis-26720	141	20	recognition	recognition	NOUN
fcis-26720	141	21	effect	effect	NOUN
fcis-26720	141	22	of	of	ADP
fcis-26720	141	23	neutral	neutral	ADJ
fcis-26720	141	24	and	and	CCONJ
fcis-26720	141	25	sad	sad	ADJ
fcis-26720	141	26	emotions	emotion	NOUN
fcis-26720	141	27	is	be	AUX
fcis-26720	141	28	better	well	ADJ
fcis-26720	141	29	,	,	PUNCT
fcis-26720	141	30	reaching	reach	VERB
fcis-26720	141	31	the	the	DET
fcis-26720	141	32	accuracy	accuracy	NOUN
fcis-26720	141	33	rates	rate	NOUN
fcis-26720	141	34	of	of	ADP
fcis-26720	141	35	95.00	95.00	NUM
fcis-26720	141	36	%	%	NOUN
fcis-26720	141	37	and	and	CCONJ
fcis-26720	141	38	96.88	96.88	NUM
fcis-26720	141	39	%	%	NOUN
fcis-26720	141	40	respectively	respectively	ADV
fcis-26720	141	41	;	;	PUNCT
fcis-26720	141	42	however	however	ADV
fcis-26720	141	43	,	,	PUNCT
fcis-26720	141	44	the	the	DET
fcis-26720	141	45	recognition	recognition	NOUN
fcis-26720	141	46	ability	ability	NOUN
fcis-26720	141	47	of	of	ADP
fcis-26720	141	48	fear	fear	NOUN
fcis-26720	141	49	and	and	CCONJ
fcis-26720	141	50	happy	happy	ADJ
fcis-26720	141	51	emotions	emotion	NOUN
fcis-26720	141	52	is	be	AUX
fcis-26720	141	53	low	low	ADJ
fcis-26720	141	54	,	,	PUNCT
fcis-26720	141	55	which	which	PRON
fcis-26720	141	56	may	may	AUX
fcis-26720	141	57	be	be	AUX
fcis-26720	141	58	due	due	ADJ
fcis-26720	141	59	to	to	ADP
fcis-26720	141	60	their	their	PRON
fcis-26720	141	61	fewer	few	ADJ
fcis-26720	141	62	corpus	corpus	NOUN
fcis-26720	141	63	files	file	NOUN
fcis-26720	141	64	,	,	PUNCT
fcis-26720	141	65	which	which	PRON
fcis-26720	141	66	leads	lead	VERB
fcis-26720	141	67	to	to	ADP
fcis-26720	141	68	the	the	DET
fcis-26720	141	69	better	well	ADJ
fcis-26720	141	70	classification	classification	NOUN
fcis-26720	141	71	effect	effect	NOUN
fcis-26720	141	72	of	of	ADP
fcis-26720	141	73	other	other	ADJ
fcis-26720	141	74	emotions	emotion	NOUN
fcis-26720	141	75	.	.	PUNCT
fcis-26720	142	1	fig	fig	NOUN
fcis-26720	142	2	5	5	NUM
fcis-26720	142	3	.	.	PUNCT
fcis-26720	142	4	confusion	confusion	NOUN
fcis-26720	142	5	matrix	matrix	NOUN
fcis-26720	142	6	of	of	ADP
fcis-26720	142	7	this	this	DET
fcis-26720	142	8	model	model	NOUN
fcis-26720	142	9	on	on	ADP
fcis-26720	142	10	emo	emo	NOUN
fcis-26720	142	11	-	-	PUNCT
fcis-26720	142	12	db	db	NOUN
fcis-26720	142	13	corpus	corpus	NOUN
fcis-26720	142	14	figure	figure	NOUN
fcis-26720	142	15	6	6	NUM
fcis-26720	142	16	depicts	depict	VERB
fcis-26720	142	17	the	the	DET
fcis-26720	142	18	training	training	NOUN
fcis-26720	142	19	of	of	ADP
fcis-26720	142	20	the	the	DET
fcis-26720	142	21	model	model	NOUN
fcis-26720	142	22	in	in	ADP
fcis-26720	142	23	this	this	DET
fcis-26720	142	24	paper	paper	NOUN
fcis-26720	142	25	on	on	ADP
fcis-26720	142	26	the	the	DET
fcis-26720	142	27	ravdess	ravdess	NOUN
fcis-26720	142	28	corpus	corpus	NOUN
fcis-26720	142	29	.	.	PUNCT
fcis-26720	143	1	it	it	PRON
fcis-26720	143	2	can	can	AUX
fcis-26720	143	3	be	be	AUX
fcis-26720	143	4	seen	see	VERB
fcis-26720	143	5	from	from	ADP
fcis-26720	143	6	the	the	DET
fcis-26720	143	7	figure	figure	NOUN
fcis-26720	143	8	that	that	SCONJ
fcis-26720	143	9	the	the	DET
fcis-26720	143	10	accuracy	accuracy	NOUN
fcis-26720	143	11	of	of	ADP
fcis-26720	143	12	the	the	DET
fcis-26720	143	13	model	model	NOUN
fcis-26720	143	14	increases	increase	VERB
fcis-26720	143	15	with	with	ADP
fcis-26720	143	16	the	the	DET
fcis-26720	143	17	increase	increase	NOUN
fcis-26720	143	18	of	of	ADP
fcis-26720	143	19	training	training	NOUN
fcis-26720	143	20	rounds	round	NOUN
fcis-26720	143	21	,	,	PUNCT
fcis-26720	143	22	and	and	CCONJ
fcis-26720	143	23	the	the	DET
fcis-26720	143	24	maximum	maximum	PROPN
fcis-26720	143	25	ua	ua	PROPN
fcis-26720	143	26	value	value	NOUN
fcis-26720	143	27	is	be	AUX
fcis-26720	143	28	95.54	95.54	NUM
fcis-26720	143	29	%	%	NOUN
fcis-26720	143	30	around	around	ADP
fcis-26720	143	31	the	the	DET
fcis-26720	143	32	290th	290th	ADJ
fcis-26720	143	33	round	round	NOUN
fcis-26720	143	34	;	;	PUNCT
fcis-26720	143	35	loss	loss	NOUN
fcis-26720	143	36	values	value	NOUN
fcis-26720	143	37	fluctuate	fluctuate	NOUN
fcis-26720	143	38	between	between	ADP
fcis-26720	143	39	0.05	0.05	NUM
fcis-26720	143	40	and	and	CCONJ
fcis-26720	143	41	0.1	0.1	NUM
fcis-26720	143	42	.	.	PUNCT
fcis-26720	144	1	90	90	NUM
fcis-26720	144	2	(	(	PUNCT
fcis-26720	144	3	a	a	NOUN
fcis-26720	144	4	)	)	PUNCT
fcis-26720	144	5	training	training	NOUN
fcis-26720	144	6	and	and	CCONJ
fcis-26720	144	7	validation	validation	NOUN
fcis-26720	144	8	accuracy	accuracy	NOUN
fcis-26720	144	9	curves	curve	NOUN
fcis-26720	144	10	of	of	ADP
fcis-26720	144	11	ravdess	ravdess	NOUN
fcis-26720	144	12	corpus	corpus	NOUN
fcis-26720	144	13	(	(	PUNCT
fcis-26720	144	14	b	b	NOUN
fcis-26720	144	15	)	)	PUNCT
fcis-26720	144	16	training	training	NOUN
fcis-26720	144	17	and	and	CCONJ
fcis-26720	144	18	validation	validation	NOUN
fcis-26720	144	19	loss	loss	NOUN
fcis-26720	144	20	value	value	NOUN
fcis-26720	144	21	curves	curve	NOUN
fcis-26720	144	22	of	of	ADP
fcis-26720	144	23	ravdess	ravdess	NOUN
fcis-26720	144	24	corpus	corpus	PROPN
fcis-26720	144	25	fig	fig	NOUN
fcis-26720	144	26	6	6	NUM
fcis-26720	144	27	.	.	PUNCT
fcis-26720	144	28	accuracy	accuracy	NOUN
fcis-26720	144	29	and	and	CCONJ
fcis-26720	144	30	loss	loss	NOUN
fcis-26720	144	31	curve	curve	NOUN
fcis-26720	144	32	of	of	ADP
fcis-26720	144	33	tbsec001	tbsec001	NOUN
fcis-26720	144	34	corpus	corpus	NOUN
fcis-26720	144	35	in	in	ADP
fcis-26720	144	36	the	the	DET
fcis-26720	144	37	confusion	confusion	NOUN
fcis-26720	144	38	matrix	matrix	NOUN
fcis-26720	144	39	of	of	ADP
fcis-26720	144	40	figure	figure	NOUN
fcis-26720	144	41	7	7	NUM
fcis-26720	144	42	,	,	PUNCT
fcis-26720	144	43	calm	calm	ADJ
fcis-26720	144	44	,	,	PUNCT
fcis-26720	144	45	happy	happy	ADJ
fcis-26720	144	46	and	and	CCONJ
fcis-26720	144	47	neutral	neutral	ADJ
fcis-26720	144	48	emotions	emotion	NOUN
fcis-26720	144	49	perform	perform	VERB
fcis-26720	144	50	best	good	ADJ
fcis-26720	144	51	,	,	PUNCT
fcis-26720	144	52	and	and	CCONJ
fcis-26720	144	53	their	their	PRON
fcis-26720	144	54	accuracy	accuracy	NOUN
fcis-26720	144	55	rates	rate	NOUN
fcis-26720	144	56	all	all	PRON
fcis-26720	144	57	reach	reach	VERB
fcis-26720	144	58	100.00	100.00	NUM
fcis-26720	144	59	%	%	NOUN
fcis-26720	144	60	.	.	PUNCT
fcis-26720	145	1	fear	fear	NOUN
fcis-26720	145	2	was	be	AUX
fcis-26720	145	3	the	the	DET
fcis-26720	145	4	worst	bad	ADJ
fcis-26720	145	5	,	,	PUNCT
fcis-26720	145	6	and	and	CCONJ
fcis-26720	145	7	6.98	6.98	NUM
fcis-26720	145	8	%	%	NOUN
fcis-26720	145	9	of	of	ADP
fcis-26720	145	10	the	the	DET
fcis-26720	145	11	corpus	corpus	NOUN
fcis-26720	145	12	were	be	AUX
fcis-26720	145	13	predicted	predict	VERB
fcis-26720	145	14	to	to	PART
fcis-26720	145	15	be	be	AUX
fcis-26720	145	16	angry	angry	ADJ
fcis-26720	145	17	and	and	CCONJ
fcis-26720	145	18	sad	sad	ADJ
fcis-26720	145	19	.	.	PUNCT
fcis-26720	146	1	fig	fig	PROPN
fcis-26720	146	2	7	7	NUM
fcis-26720	146	3	.	.	PUNCT
fcis-26720	146	4	confusion	confusion	NOUN
fcis-26720	146	5	matrix	matrix	NOUN
fcis-26720	146	6	of	of	ADP
fcis-26720	146	7	this	this	DET
fcis-26720	146	8	model	model	NOUN
fcis-26720	146	9	on	on	ADP
fcis-26720	146	10	ravdess	ravdess	NOUN
fcis-26720	146	11	corpus	corpus	NOUN
fcis-26720	146	12	(	(	PUNCT
fcis-26720	146	13	3	3	NUM
fcis-26720	146	14	)	)	PUNCT
fcis-26720	146	15	comparative	comparative	ADJ
fcis-26720	146	16	experiment	experiment	NOUN
fcis-26720	146	17	in	in	ADP
fcis-26720	146	18	this	this	DET
fcis-26720	146	19	paper	paper	NOUN
fcis-26720	146	20	,	,	PUNCT
fcis-26720	146	21	the	the	DET
fcis-26720	146	22	proposed	propose	VERB
fcis-26720	146	23	model	model	NOUN
fcis-26720	146	24	is	be	AUX
fcis-26720	146	25	compared	compare	VERB
fcis-26720	146	26	with	with	ADP
fcis-26720	146	27	the	the	DET
fcis-26720	146	28	speech	speech	NOUN
fcis-26720	146	29	emotion	emotion	NOUN
fcis-26720	146	30	recognition	recognition	NOUN
fcis-26720	146	31	models	model	NOUN
fcis-26720	146	32	published	publish	VERB
fcis-26720	146	33	in	in	ADP
fcis-26720	146	34	the	the	DET
fcis-26720	146	35	past	past	ADJ
fcis-26720	146	36	three	three	NUM
fcis-26720	146	37	years	year	NOUN
fcis-26720	146	38	,	,	PUNCT
fcis-26720	146	39	and	and	CCONJ
fcis-26720	146	40	comparative	comparative	ADJ
fcis-26720	146	41	experiments	experiment	NOUN
fcis-26720	146	42	are	be	AUX
fcis-26720	146	43	carried	carry	VERB
fcis-26720	146	44	out	out	ADP
fcis-26720	146	45	on	on	ADP
fcis-26720	146	46	tbsec001	tbsec001	PROPN
fcis-26720	146	47	,	,	PUNCT
fcis-26720	146	48	emo	emo	NOUN
fcis-26720	146	49	-	-	PUNCT
fcis-26720	146	50	db	db	NOUN
fcis-26720	146	51	and	and	CCONJ
fcis-26720	146	52	ravdess	ravdess	NOUN
fcis-26720	146	53	corpus	corpus	PROPN
fcis-26720	146	54	.	.	PUNCT
fcis-26720	147	1	see	see	VERB
fcis-26720	147	2	table	table	NOUN
fcis-26720	147	3	6	6	NUM
fcis-26720	147	4	for	for	ADP
fcis-26720	147	5	details	detail	NOUN
fcis-26720	147	6	.	.	PUNCT
fcis-26720	148	1	it	it	PRON
fcis-26720	148	2	can	can	AUX
fcis-26720	148	3	be	be	AUX
fcis-26720	148	4	seen	see	VERB
fcis-26720	148	5	from	from	ADP
fcis-26720	148	6	table	table	NOUN
fcis-26720	148	7	8	8	NUM
fcis-26720	148	8	that	that	SCONJ
fcis-26720	148	9	although	although	SCONJ
fcis-26720	148	10	the	the	DET
fcis-26720	148	11	model	model	NOUN
fcis-26720	148	12	in	in	ADP
fcis-26720	148	13	this	this	DET
fcis-26720	148	14	paper	paper	NOUN
fcis-26720	148	15	does	do	AUX
fcis-26720	148	16	not	not	PART
fcis-26720	148	17	perform	perform	VERB
fcis-26720	148	18	well	well	ADV
fcis-26720	148	19	enough	enough	ADV
fcis-26720	148	20	on	on	ADP
fcis-26720	148	21	tbsec001	tbsec001	NOUN
fcis-26720	148	22	and	and	CCONJ
fcis-26720	148	23	emo	emo	NOUN
fcis-26720	148	24	-	-	PUNCT
fcis-26720	148	25	db	db	NOUN
fcis-26720	148	26	corpus	corpus	NOUN
fcis-26720	148	27	,	,	PUNCT
fcis-26720	148	28	only	only	ADV
fcis-26720	148	29	the	the	DET
fcis-26720	148	30	single	single	ADJ
fcis-26720	148	31	feature	feature	NOUN
fcis-26720	148	32	of	of	ADP
fcis-26720	148	33	mfcc	mfcc	NOUN
fcis-26720	148	34	is	be	AUX
fcis-26720	148	35	used	use	VERB
fcis-26720	148	36	as	as	ADP
fcis-26720	148	37	input	input	NOUN
fcis-26720	148	38	when	when	SCONJ
fcis-26720	148	39	dealing	deal	VERB
fcis-26720	148	40	with	with	ADP
fcis-26720	148	41	tbsec001	tbsec001	NOUN
fcis-26720	148	42	corpus	corpus	NOUN
fcis-26720	148	43	.	.	PUNCT
fcis-26720	149	1	compared	compare	VERB
fcis-26720	149	2	with	with	ADP
fcis-26720	149	3	the	the	DET
fcis-26720	149	4	complex	complex	ADJ
fcis-26720	149	5	feature	feature	NOUN
fcis-26720	149	6	set	set	NOUN
fcis-26720	149	7	of	of	ADP
fcis-26720	149	8	multiple	multiple	ADJ
fcis-26720	149	9	llds	lld	NOUN
fcis-26720	149	10	and	and	CCONJ
fcis-26720	149	11	high	high	ADJ
fcis-26720	149	12	-	-	PUNCT
fcis-26720	149	13	level	level	NOUN
fcis-26720	149	14	statistical	statistical	ADJ
fcis-26720	149	15	functions	function	NOUN
fcis-26720	149	16	(	(	PUNCT
fcis-26720	149	17	hsfs	hsfs	ADJ
fcis-26720	149	18	)	)	PUNCT
fcis-26720	149	19	in	in	ADP
fcis-26720	149	20	literature	literature	NOUN
fcis-26720	149	21	[	[	X
fcis-26720	149	22	19	19	NUM
fcis-26720	149	23	]	]	X
fcis-26720	149	24	,	,	PUNCT
fcis-26720	149	25	its	its	PRON
fcis-26720	149	26	feature	feature	NOUN
fcis-26720	149	27	dimensions	dimension	NOUN
fcis-26720	149	28	are	be	AUX
fcis-26720	149	29	greatly	greatly	ADV
fcis-26720	149	30	simplified	simplify	VERB
fcis-26720	149	31	.	.	PUNCT
fcis-26720	150	1	on	on	ADP
fcis-26720	150	2	the	the	DET
fcis-26720	150	3	ravdess	ravdess	NOUN
fcis-26720	150	4	corpus	corpus	NOUN
fcis-26720	150	5	,	,	PUNCT
fcis-26720	150	6	this	this	DET
fcis-26720	150	7	model	model	NOUN
fcis-26720	150	8	has	have	AUX
fcis-26720	150	9	achieved	achieve	VERB
fcis-26720	150	10	the	the	DET
fcis-26720	150	11	best	good	ADJ
fcis-26720	150	12	results	result	NOUN
fcis-26720	150	13	.	.	PUNCT
fcis-26720	151	1	table	table	NOUN
fcis-26720	151	2	6	6	NUM
fcis-26720	151	3	.	.	PUNCT
fcis-26720	151	4	comparison	comparison	NOUN
fcis-26720	151	5	between	between	ADP
fcis-26720	151	6	the	the	DET
fcis-26720	151	7	model	model	NOUN
fcis-26720	151	8	in	in	ADP
fcis-26720	151	9	this	this	DET
fcis-26720	151	10	paper	paper	NOUN
fcis-26720	151	11	and	and	CCONJ
fcis-26720	151	12	the	the	DET
fcis-26720	151	13	existing	exist	VERB
fcis-26720	151	14	model	model	NOUN
fcis-26720	151	15	model	model	NOUN
fcis-26720	151	16	trait	trait	NOUN
fcis-26720	151	17	tbsec001	tbsec001	PROPN
fcis-26720	151	18	(	(	PUNCT
fcis-26720	151	19	accuracy	accuracy	NOUN
fcis-26720	151	20	)	)	PUNCT
fcis-26720	151	21	emo	emo	NOUN
fcis-26720	151	22	-	-	PUNCT
fcis-26720	151	23	db	db	PROPN
fcis-26720	151	24	(	(	PUNCT
fcis-26720	151	25	ua	ua	NOUN
fcis-26720	151	26	)	)	PUNCT
fcis-26720	151	27	ravdess	ravdess	NOUN
fcis-26720	151	28	(	(	PUNCT
fcis-26720	151	29	ua	ua	NOUN
fcis-26720	151	30	)	)	PUNCT
fcis-26720	151	31	literature	literature	NOUN
fcis-26720	151	32	[	[	X
fcis-26720	151	33	19	19	NUM
fcis-26720	151	34	]	]	X
fcis-26720	151	35	llds	lld	NOUN
fcis-26720	151	36	&	&	CCONJ
fcis-26720	151	37	hsfs	hsfs	ADJ
fcis-26720	151	38	88.40	88.40	NUM
fcis-26720	151	39	%	%	NOUN
fcis-26720	151	40	84.10	84.10	NUM
fcis-26720	151	41	%	%	NOUN
fcis-26720	151	42	74.30	74.30	NUM
fcis-26720	151	43	%	%	NOUN
fcis-26720	151	44	literature	literature	NOUN
fcis-26720	152	1	[	[	X
fcis-26720	152	2	28	28	NUM
fcis-26720	152	3	]	]	PUNCT
fcis-26720	152	4	mfcc	mfcc	NOUN
fcis-26720	152	5	95.17	95.17	NUM
fcis-26720	152	6	%	%	NOUN
fcis-26720	152	7	91.93	91.93	NUM
fcis-26720	152	8	%	%	NOUN
fcis-26720	152	9	literature	literature	NOUN
fcis-26720	152	10	[	[	X
fcis-26720	152	11	29	29	NUM
fcis-26720	152	12	]	]	X
fcis-26720	152	13	mel	mel	PROPN
fcis-26720	152	14	spectrogram	spectrogram	PROPN
fcis-26720	152	15	90.20	90.20	NUM
fcis-26720	152	16	%	%	NOUN
fcis-26720	152	17	85.80	85.80	NUM
fcis-26720	152	18	%	%	NOUN
fcis-26720	152	19	the	the	DET
fcis-26720	152	20	model	model	NOUN
fcis-26720	152	21	of	of	ADP
fcis-26720	152	22	this	this	DET
fcis-26720	152	23	paper	paper	NOUN
fcis-26720	152	24	mfcc	mfcc	NOUN
fcis-26720	152	25	81.52	81.52	NUM
fcis-26720	152	26	%	%	NOUN
fcis-26720	152	27	85.63	85.63	NUM
fcis-26720	152	28	%	%	NOUN
fcis-26720	152	29	95.54	95.54	NUM
fcis-26720	152	30	%	%	NOUN
fcis-26720	152	31	4	4	NUM
fcis-26720	152	32	.	.	NOUN
fcis-26720	152	33	conclusion	conclusion	NOUN
fcis-26720	152	34	this	this	DET
fcis-26720	152	35	paper	paper	NOUN
fcis-26720	152	36	focuses	focus	VERB
fcis-26720	152	37	on	on	ADP
fcis-26720	152	38	the	the	DET
fcis-26720	152	39	field	field	NOUN
fcis-26720	152	40	of	of	ADP
fcis-26720	152	41	tibetan	tibetan	ADJ
fcis-26720	152	42	speech	speech	NOUN
fcis-26720	152	43	emotion	emotion	NOUN
fcis-26720	152	44	recognition	recognition	NOUN
fcis-26720	152	45	,	,	PUNCT
fcis-26720	152	46	and	and	CCONJ
fcis-26720	152	47	constructs	construct	VERB
fcis-26720	152	48	a	a	DET
fcis-26720	152	49	lightweight	lightweight	ADJ
fcis-26720	152	50	speech	speech	NOUN
fcis-26720	152	51	emotion	emotion	NOUN
fcis-26720	152	52	recognition	recognition	NOUN
fcis-26720	152	53	model	model	NOUN
fcis-26720	152	54	for	for	ADP
fcis-26720	152	55	the	the	DET
fcis-26720	152	56	problems	problem	NOUN
fcis-26720	152	57	of	of	ADP
fcis-26720	152	58	relatively	relatively	ADV
fcis-26720	152	59	few	few	ADJ
fcis-26720	152	60	research	research	NOUN
fcis-26720	152	61	and	and	CCONJ
fcis-26720	152	62	complex	complex	ADJ
fcis-26720	152	63	feature	feature	NOUN
fcis-26720	152	64	extraction	extraction	NOUN
fcis-26720	152	65	methods	method	NOUN
fcis-26720	152	66	.	.	PUNCT
fcis-26720	153	1	the	the	DET
fcis-26720	153	2	model	model	NOUN
fcis-26720	153	3	takes	take	VERB
fcis-26720	153	4	mfcc	mfcc	NOUN
fcis-26720	153	5	as	as	ADP
fcis-26720	153	6	input	input	NOUN
fcis-26720	153	7	features	feature	NOUN
fcis-26720	153	8	.	.	PUNCT
fcis-26720	154	1	by	by	ADP
fcis-26720	154	2	extracting	extract	VERB
fcis-26720	154	3	the	the	DET
fcis-26720	154	4	spatiotemporal	spatiotemporal	ADJ
fcis-26720	154	5	features	feature	NOUN
fcis-26720	154	6	of	of	ADP
fcis-26720	154	7	mfcc	mfcc	NOUN
fcis-26720	154	8	and	and	CCONJ
fcis-26720	154	9	combining	combine	VERB
fcis-26720	154	10	the	the	DET
fcis-26720	154	11	dynamic	dynamic	ADJ
fcis-26720	154	12	routing	routing	NOUN
fcis-26720	154	13	update	update	NOUN
fcis-26720	154	14	algorithm	algorithm	NOUN
fcis-26720	154	15	of	of	ADP
fcis-26720	154	16	capsule	capsule	NOUN
fcis-26720	154	17	network	network	NOUN
fcis-26720	154	18	,	,	PUNCT
fcis-26720	154	19	margin	margin	NOUN
fcis-26720	154	20	-	-	PUNCT
fcis-26720	154	21	loss	loss	NOUN
fcis-26720	154	22	is	be	AUX
fcis-26720	154	23	used	use	VERB
fcis-26720	154	24	to	to	PART
fcis-26720	154	25	guide	guide	VERB
fcis-26720	154	26	the	the	DET
fcis-26720	154	27	model	model	NOUN
fcis-26720	154	28	to	to	PART
fcis-26720	154	29	update	update	VERB
fcis-26720	154	30	parameters	parameter	NOUN
fcis-26720	154	31	.	.	PUNCT
fcis-26720	155	1	satisfactory	satisfactory	ADJ
fcis-26720	155	2	recognition	recognition	NOUN
fcis-26720	155	3	results	result	NOUN
fcis-26720	155	4	are	be	AUX
fcis-26720	155	5	achieved	achieve	VERB
fcis-26720	155	6	in	in	ADP
fcis-26720	155	7	the	the	DET
fcis-26720	155	8	self	self	NOUN
fcis-26720	155	9	-	-	PUNCT
fcis-26720	155	10	built	build	VERB
fcis-26720	155	11	corpus	corpus	NOUN
fcis-26720	155	12	tbsec001	tbsec001	PROPN
fcis-26720	155	13	and	and	CCONJ
fcis-26720	155	14	public	public	ADJ
fcis-26720	155	15	corpora	corpora	PROPN
fcis-26720	155	16	emo	emo	NOUN
fcis-26720	155	17	-	-	PUNCT
fcis-26720	155	18	db	db	NOUN
fcis-26720	155	19	and	and	CCONJ
fcis-26720	155	20	ravdess	ravdess	NOUN
fcis-26720	155	21	.	.	PUNCT
fcis-26720	156	1	in	in	ADP
fcis-26720	156	2	addition	addition	NOUN
fcis-26720	156	3	,	,	PUNCT
fcis-26720	156	4	compared	compare	VERB
fcis-26720	156	5	with	with	ADP
fcis-26720	156	6	the	the	DET
fcis-26720	156	7	existing	exist	VERB
fcis-26720	156	8	excellent	excellent	ADJ
fcis-26720	156	9	models	model	NOUN
fcis-26720	156	10	,	,	PUNCT
fcis-26720	156	11	the	the	DET
fcis-26720	156	12	experiment	experiment	NOUN
fcis-26720	156	13	has	have	VERB
fcis-26720	156	14	a	a	DET
fcis-26720	156	15	good	good	ADJ
fcis-26720	156	16	performance	performance	NOUN
fcis-26720	156	17	,	,	PUNCT
fcis-26720	156	18	and	and	CCONJ
fcis-26720	156	19	also	also	ADV
fcis-26720	156	20	shows	show	VERB
fcis-26720	156	21	strong	strong	ADJ
fcis-26720	156	22	competitiveness	competitiveness	NOUN
fcis-26720	156	23	and	and	CCONJ
fcis-26720	156	24	generalization	generalization	NOUN
fcis-26720	156	25	ability	ability	NOUN
fcis-26720	156	26	.	.	PUNCT
fcis-26720	157	1	in	in	ADP
fcis-26720	157	2	future	future	ADJ
fcis-26720	157	3	work	work	NOUN
fcis-26720	157	4	,	,	PUNCT
fcis-26720	157	5	based	base	VERB
fcis-26720	157	6	on	on	ADP
fcis-26720	157	7	this	this	DET
fcis-26720	157	8	model	model	NOUN
fcis-26720	157	9	,	,	PUNCT
fcis-26720	157	10	we	we	PRON
fcis-26720	157	11	will	will	AUX
fcis-26720	157	12	combine	combine	VERB
fcis-26720	157	13	the	the	DET
fcis-26720	157	14	knowledge	knowledge	NOUN
fcis-26720	157	15	of	of	ADP
fcis-26720	157	16	deep	deep	ADJ
fcis-26720	157	17	learning	learning	NOUN
fcis-26720	157	18	and	and	CCONJ
fcis-26720	157	19	machine	machine	NOUN
fcis-26720	157	20	learning	learning	NOUN
fcis-26720	157	21	,	,	PUNCT
fcis-26720	157	22	new	new	ADJ
fcis-26720	157	23	algorithms	algorithm	NOUN
fcis-26720	157	24	and	and	CCONJ
fcis-26720	157	25	models	model	NOUN
fcis-26720	157	26	will	will	AUX
fcis-26720	157	27	be	be	AUX
fcis-26720	157	28	proposed	propose	VERB
fcis-26720	157	29	to	to	PART
fcis-26720	157	30	better	well	ADV
fcis-26720	157	31	handle	handle	VERB
fcis-26720	157	32	speech	speech	NOUN
fcis-26720	157	33	emotion	emotion	NOUN
fcis-26720	157	34	recognition	recognition	NOUN
fcis-26720	157	35	tasks	task	NOUN
fcis-26720	157	36	.	.	PUNCT
fcis-26720	158	1	references	reference	NOUN
fcis-26720	158	2	[	[	X
fcis-26720	158	3	1	1	NUM
fcis-26720	158	4	]	]	X
fcis-26720	158	5	mencattini	mencattini	NOUN
fcis-26720	158	6	a	a	NOUN
fcis-26720	158	7	,	,	PUNCT
fcis-26720	158	8	martinelli	martinelli	PROPN
fcis-26720	158	9	e	e	PROPN
fcis-26720	158	10	,	,	PUNCT
fcis-26720	158	11	ringeval	ringeval	PROPN
fcis-26720	158	12	f	f	PROPN
fcis-26720	158	13	,	,	PUNCT
fcis-26720	158	14	et	et	PROPN
fcis-26720	158	15	al	al	PROPN
fcis-26720	158	16	.	.	PROPN
fcis-26720	158	17	continuous	continuous	ADJ
fcis-26720	158	18	estimation	estimation	NOUN
fcis-26720	158	19	of	of	ADP
fcis-26720	158	20	emotions	emotion	NOUN
fcis-26720	158	21	in	in	ADP
fcis-26720	158	22	speech	speech	NOUN
fcis-26720	158	23	by	by	ADP
fcis-26720	158	24	dynamic	dynamic	ADJ
fcis-26720	158	25	cooperative	cooperative	ADJ
fcis-26720	158	26	speaker	speaker	NOUN
fcis-26720	158	27	models	model	NOUN
fcis-26720	159	1	[	[	X
fcis-26720	159	2	j	j	X
fcis-26720	159	3	]	]	X
fcis-26720	159	4	.	.	PUNCT
fcis-26720	160	1	ieee	ieee	NOUN
fcis-26720	160	2	transactions	transaction	NOUN
fcis-26720	160	3	on	on	ADP
fcis-26720	160	4	affective	affective	ADJ
fcis-26720	160	5	computing	computing	NOUN
fcis-26720	160	6	,	,	PUNCT
fcis-26720	160	7	2016	2016	NUM
fcis-26720	160	8	,	,	PUNCT
fcis-26720	160	9	8	8	NUM
fcis-26720	160	10	(	(	PUNCT
fcis-26720	160	11	3	3	NUM
fcis-26720	160	12	):	):	PUNCT
fcis-26720	160	13	314	314	NUM
fcis-26720	160	14	-	-	SYM
fcis-26720	160	15	327	327	NUM
fcis-26720	160	16	.	.	PUNCT
fcis-26720	161	1	[	[	X
fcis-26720	161	2	2	2	X
fcis-26720	161	3	]	]	PUNCT
fcis-26720	161	4	hashem	hashem	PROPN
fcis-26720	161	5	a	a	PROPN
fcis-26720	161	6	,	,	PUNCT
fcis-26720	161	7	arif	arif	PROPN
fcis-26720	161	8	m	m	PROPN
fcis-26720	161	9	,	,	PUNCT
fcis-26720	161	10	alghamdi	alghamdi	NOUN
fcis-26720	161	11	m.	m.	NOUN
fcis-26720	161	12	speech	speech	NOUN
fcis-26720	161	13	emotion	emotion	NOUN
fcis-26720	161	14	recognition	recognition	NOUN
fcis-26720	161	15	approaches	approach	VERB
fcis-26720	161	16	:	:	PUNCT
fcis-26720	161	17	a	a	DET
fcis-26720	161	18	systematic	systematic	ADJ
fcis-26720	161	19	review	review	NOUN
fcis-26720	162	1	[	[	X
fcis-26720	162	2	j	j	X
fcis-26720	162	3	]	]	X
fcis-26720	162	4	.	.	PUNCT
fcis-26720	163	1	speech	speech	NOUN
fcis-26720	163	2	communication	communication	NOUN
fcis-26720	163	3	,	,	PUNCT
fcis-26720	163	4	2023	2023	NUM
fcis-26720	163	5	:	:	PUNCT
fcis-26720	163	6	102974	102974	NUM
fcis-26720	163	7	.	.	PUNCT
fcis-26720	164	1	[	[	X
fcis-26720	164	2	3	3	NUM
fcis-26720	164	3	]	]	X
fcis-26720	164	4	al	al	PROPN
fcis-26720	164	5	-	-	PUNCT
fcis-26720	164	6	dujaili	dujaili	PROPN
fcis-26720	164	7	m	m	PROPN
fcis-26720	164	8	j	j	PROPN
fcis-26720	164	9	,	,	PUNCT
fcis-26720	164	10	ebrahimi	ebrahimi	PROPN
fcis-26720	164	11	-	-	PUNCT
fcis-26720	164	12	moghadam	moghadam	PROPN
fcis-26720	164	13	a.	a.	NOUN
fcis-26720	164	14	speech	speech	PROPN
fcis-26720	164	15	emotion	emotion	NOUN
fcis-26720	164	16	recognition	recognition	NOUN
fcis-26720	164	17	:	:	PUNCT
fcis-26720	164	18	a	a	DET
fcis-26720	164	19	comprehensive	comprehensive	ADJ
fcis-26720	164	20	survey	survey	NOUN
fcis-26720	164	21	[	[	X
fcis-26720	164	22	j	j	X
fcis-26720	164	23	]	]	X
fcis-26720	164	24	.	.	PUNCT
fcis-26720	165	1	wireless	wireless	ADJ
fcis-26720	165	2	personal	personal	ADJ
fcis-26720	165	3	communications	communication	NOUN
fcis-26720	165	4	,	,	PUNCT
fcis-26720	165	5	2023	2023	NUM
fcis-26720	165	6	,	,	PUNCT
fcis-26720	165	7	129	129	NUM
fcis-26720	165	8	(	(	PUNCT
fcis-26720	165	9	4	4	NUM
fcis-26720	165	10	):	):	PUNCT
fcis-26720	165	11	2525	2525	NUM
fcis-26720	165	12	-	-	SYM
fcis-26720	165	13	2561	2561	NUM
fcis-26720	165	14	.	.	PUNCT
fcis-26720	166	1	[	[	X
fcis-26720	166	2	4	4	NUM
fcis-26720	166	3	]	]	SYM
fcis-26720	166	4	atmaja	atmaja	PROPN
fcis-26720	166	5	b	b	PROPN
fcis-26720	166	6	t	t	PROPN
fcis-26720	166	7	,	,	PUNCT
fcis-26720	166	8	akagi	akagi	PROPN
fcis-26720	166	9	m.	m.	PROPN
fcis-26720	166	10	the	the	DET
fcis-26720	166	11	effect	effect	NOUN
fcis-26720	166	12	of	of	ADP
fcis-26720	166	13	silence	silence	NOUN
fcis-26720	166	14	feature	feature	NOUN
fcis-26720	166	15	in	in	ADP
fcis-26720	166	16	dimensional	dimensional	ADJ
fcis-26720	166	17	speech	speech	NOUN
fcis-26720	166	18	emotion	emotion	NOUN
fcis-26720	166	19	recognition	recognition	NOUN
fcis-26720	167	1	[	[	X
fcis-26720	167	2	j	j	X
fcis-26720	167	3	]	]	X
fcis-26720	167	4	.	.	PUNCT
fcis-26720	168	1	arxiv	arxiv	PROPN
fcis-26720	168	2	preprint	preprint	PROPN
fcis-26720	168	3	arxiv	arxiv	PROPN
fcis-26720	168	4	:	:	PUNCT
fcis-26720	168	5	2003.01277	2003.01277	NUM
fcis-26720	168	6	,	,	PUNCT
fcis-26720	168	7	2020	2020	NUM
fcis-26720	168	8	.	.	PUNCT
fcis-26720	169	1	[	[	X
fcis-26720	169	2	5	5	NUM
fcis-26720	169	3	]	]	PUNCT
fcis-26720	169	4	akçay	akçay	NOUN
fcis-26720	169	5	m	m	PROPN
fcis-26720	169	6	b	b	PROPN
fcis-26720	169	7	,	,	PUNCT
fcis-26720	169	8	oğuz	oğuz	PROPN
fcis-26720	169	9	k.	k.	PROPN
fcis-26720	169	10	speech	speech	PROPN
fcis-26720	169	11	emotion	emotion	NOUN
fcis-26720	169	12	recognition	recognition	NOUN
fcis-26720	169	13	:	:	PUNCT
fcis-26720	169	14	emotional	emotional	ADJ
fcis-26720	169	15	models	model	NOUN
fcis-26720	169	16	,	,	PUNCT
fcis-26720	169	17	databases	database	NOUN
fcis-26720	169	18	,	,	PUNCT
fcis-26720	169	19	features	feature	NOUN
fcis-26720	169	20	,	,	PUNCT
fcis-26720	169	21	preprocessing	preprocesse	VERB
fcis-26720	169	22	methods	method	NOUN
fcis-26720	169	23	,	,	PUNCT
fcis-26720	169	24	supporting	support	VERB
fcis-26720	169	25	modalities	modality	NOUN
fcis-26720	169	26	,	,	PUNCT
fcis-26720	169	27	and	and	CCONJ
fcis-26720	169	28	classifiers	classifier	NOUN
fcis-26720	169	29	[	[	X
fcis-26720	169	30	j	j	X
fcis-26720	169	31	]	]	X
fcis-26720	169	32	.	.	PUNCT
fcis-26720	170	1	speech	speech	NOUN
fcis-26720	170	2	communication	communication	NOUN
fcis-26720	170	3	,	,	PUNCT
fcis-26720	170	4	2020	2020	NUM
fcis-26720	170	5	,	,	PUNCT
fcis-26720	170	6	116	116	NUM
fcis-26720	170	7	:	:	PUNCT
fcis-26720	170	8	56	56	NUM
fcis-26720	170	9	-	-	SYM
fcis-26720	170	10	76	76	NUM
fcis-26720	170	11	.	.	PUNCT
fcis-26720	171	1	[	[	X
fcis-26720	171	2	6	6	NUM
fcis-26720	171	3	]	]	AUX
fcis-26720	171	4	fahad	fahad	VERB
fcis-26720	171	5	m	m	PROPN
fcis-26720	171	6	s	s	PROPN
fcis-26720	171	7	,	,	PUNCT
fcis-26720	171	8	ranjan	ranjan	NOUN
fcis-26720	171	9	a	a	PROPN
fcis-26720	171	10	,	,	PUNCT
fcis-26720	171	11	yadav	yadav	PROPN
fcis-26720	171	12	j	j	PROPN
fcis-26720	171	13	,	,	PUNCT
fcis-26720	171	14	et	et	PROPN
fcis-26720	171	15	al	al	PROPN
fcis-26720	171	16	.	.	PUNCT
fcis-26720	172	1	a	a	DET
fcis-26720	172	2	survey	survey	NOUN
fcis-26720	172	3	of	of	ADP
fcis-26720	172	4	speech	speech	NOUN
fcis-26720	172	5	emotion	emotion	NOUN
fcis-26720	172	6	recognition	recognition	NOUN
fcis-26720	172	7	in	in	ADP
fcis-26720	172	8	natural	natural	ADJ
fcis-26720	172	9	environment	environment	NOUN
fcis-26720	173	1	[	[	X
fcis-26720	173	2	j	j	X
fcis-26720	173	3	]	]	X
fcis-26720	173	4	.	.	PUNCT
fcis-26720	174	1	digital	digital	PROPN
fcis-26720	174	2	signal	signal	PROPN
fcis-26720	174	3	processing	processing	NOUN
fcis-26720	174	4	,	,	PUNCT
fcis-26720	174	5	2021	2021	NUM
fcis-26720	174	6	,	,	PUNCT
fcis-26720	174	7	110	110	NUM
fcis-26720	174	8	:	:	SYM
fcis-26720	174	9	102951	102951	NUM
fcis-26720	174	10	.	.	PUNCT
fcis-26720	175	1	91	91	NUM
fcis-26720	176	1	[	[	X
fcis-26720	176	2	7	7	X
fcis-26720	176	3	]	]	X
fcis-26720	176	4	jain	jain	PROPN
fcis-26720	176	5	m	m	PROPN
fcis-26720	176	6	,	,	PUNCT
fcis-26720	176	7	narayan	narayan	PROPN
fcis-26720	176	8	s	s	PROPN
fcis-26720	176	9	,	,	PUNCT
fcis-26720	176	10	balaji	balaji	PROPN
fcis-26720	176	11	p	p	PROPN
fcis-26720	176	12	,	,	PUNCT
fcis-26720	176	13	et	et	PROPN
fcis-26720	176	14	al	al	PROPN
fcis-26720	176	15	.	.	PROPN
fcis-26720	176	16	speech	speech	NOUN
fcis-26720	176	17	emotion	emotion	NOUN
fcis-26720	176	18	recognition	recognition	NOUN
fcis-26720	176	19	using	use	VERB
fcis-26720	176	20	support	support	NOUN
fcis-26720	176	21	vector	vector	NOUN
fcis-26720	176	22	machine	machine	NOUN
fcis-26720	177	1	[	[	X
fcis-26720	177	2	j	j	X
fcis-26720	177	3	]	]	X
fcis-26720	177	4	.	.	PUNCT
fcis-26720	178	1	arxiv	arxiv	PROPN
fcis-26720	178	2	preprint	preprint	PROPN
fcis-26720	178	3	arxiv	arxiv	PROPN
fcis-26720	178	4	:	:	PUNCT
fcis-26720	178	5	2002	2002	NUM
fcis-26720	178	6	.	.	PUNCT
fcis-26720	178	7	07590	07590	NUM
fcis-26720	178	8	,	,	PUNCT
fcis-26720	178	9	2020	2020	NUM
fcis-26720	178	10	.	.	PUNCT
fcis-26720	179	1	[	[	X
fcis-26720	179	2	8	8	NUM
fcis-26720	179	3	]	]	X
fcis-26720	179	4	aouani	aouani	NOUN
fcis-26720	179	5	h	h	NOUN
fcis-26720	179	6	,	,	PUNCT
fcis-26720	179	7	ayed	aye	VERB
fcis-26720	179	8	y	y	PROPN
fcis-26720	179	9	b.	b.	PROPN
fcis-26720	179	10	speech	speech	PROPN
fcis-26720	179	11	emotion	emotion	NOUN
fcis-26720	179	12	recognition	recognition	NOUN
fcis-26720	179	13	with	with	ADP
fcis-26720	179	14	deep	deep	ADJ
fcis-26720	179	15	learning	learning	NOUN
fcis-26720	179	16	[	[	X
fcis-26720	179	17	j	j	X
fcis-26720	179	18	]	]	X
fcis-26720	179	19	.	.	PUNCT
fcis-26720	180	1	procedia	procedia	PROPN
fcis-26720	180	2	computer	computer	NOUN
fcis-26720	180	3	science	science	NOUN
fcis-26720	180	4	,	,	PUNCT
fcis-26720	180	5	2020	2020	NUM
fcis-26720	180	6	,	,	PUNCT
fcis-26720	180	7	176	176	NUM
fcis-26720	180	8	:	:	PUNCT
fcis-26720	180	9	251	251	NUM
fcis-26720	180	10	-	-	SYM
fcis-26720	180	11	260	260	NUM
fcis-26720	180	12	.	.	PUNCT
fcis-26720	181	1	[	[	X
fcis-26720	181	2	9	9	NUM
fcis-26720	181	3	]	]	X
fcis-26720	181	4	wang	wang	PROPN
fcis-26720	181	5	j	j	PROPN
fcis-26720	181	6	,	,	PUNCT
fcis-26720	181	7	xue	xue	PROPN
fcis-26720	181	8	m	m	PROPN
fcis-26720	181	9	,	,	PUNCT
fcis-26720	181	10	culhane	culhane	NOUN
fcis-26720	181	11	r	r	NOUN
fcis-26720	181	12	,	,	PUNCT
fcis-26720	181	13	et	et	PROPN
fcis-26720	182	1	al	al	PROPN
fcis-26720	182	2	.	.	PROPN
fcis-26720	182	3	speech	speech	NOUN
fcis-26720	182	4	emotion	emotion	NOUN
fcis-26720	182	5	recognition	recognition	NOUN
fcis-26720	182	6	with	with	ADP
fcis-26720	182	7	dual	dual	ADJ
fcis-26720	182	8	-	-	PUNCT
fcis-26720	182	9	sequence	sequence	NOUN
fcis-26720	182	10	lstm	lstm	ADJ
fcis-26720	182	11	architecture	architecture	NOUN
fcis-26720	182	12	[	[	X
fcis-26720	182	13	c]//icassp	c]//icassp	PROPN
fcis-26720	182	14	20202020	20202020	NUM
fcis-26720	182	15	ieee	ieee	NOUN
fcis-26720	182	16	international	international	ADJ
fcis-26720	182	17	conference	conference	NOUN
fcis-26720	182	18	on	on	ADP
fcis-26720	182	19	acoustics	acoustic	NOUN
fcis-26720	182	20	,	,	PUNCT
fcis-26720	182	21	speech	speech	NOUN
fcis-26720	182	22	and	and	CCONJ
fcis-26720	182	23	signal	signal	NOUN
fcis-26720	182	24	processing	processing	NOUN
fcis-26720	182	25	(	(	PUNCT
fcis-26720	182	26	icassp	icassp	PROPN
fcis-26720	182	27	)	)	PUNCT
fcis-26720	182	28	.	.	PUNCT
fcis-26720	183	1	ieee	ieee	PROPN
fcis-26720	183	2	,	,	PUNCT
fcis-26720	183	3	2020	2020	NUM
fcis-26720	183	4	:	:	PUNCT
fcis-26720	183	5	6474	6474	NUM
fcis-26720	183	6	-	-	SYM
fcis-26720	183	7	6478	6478	NUM
fcis-26720	183	8	.	.	PUNCT
fcis-26720	184	1	[	[	X
fcis-26720	184	2	10	10	NUM
fcis-26720	184	3	]	]	X
fcis-26720	184	4	liu	liu	PROPN
fcis-26720	184	5	g	g	PROPN
fcis-26720	184	6	,	,	PUNCT
fcis-26720	184	7	he	he	PRON
fcis-26720	184	8	w	w	PROPN
fcis-26720	184	9	,	,	PUNCT
fcis-26720	184	10	jin	jin	PROPN
fcis-26720	184	11	b.	b.	PROPN
fcis-26720	184	12	feature	feature	NOUN
fcis-26720	184	13	fusion	fusion	NOUN
fcis-26720	184	14	of	of	ADP
fcis-26720	184	15	speech	speech	NOUN
fcis-26720	184	16	emotion	emotion	NOUN
fcis-26720	184	17	recognition	recognition	NOUN
fcis-26720	184	18	based	base	VERB
fcis-26720	184	19	on	on	ADP
fcis-26720	184	20	deep	deep	ADJ
fcis-26720	184	21	learning	learning	NOUN
fcis-26720	184	22	[	[	PUNCT
fcis-26720	184	23	c]//2018	c]//2018	PRON
fcis-26720	184	24	international	international	ADJ
fcis-26720	184	25	conference	conference	NOUN
fcis-26720	184	26	on	on	ADP
fcis-26720	184	27	network	network	NOUN
fcis-26720	184	28	infrastructure	infrastructure	NOUN
fcis-26720	184	29	and	and	CCONJ
fcis-26720	184	30	digital	digital	ADJ
fcis-26720	184	31	content	content	NOUN
fcis-26720	184	32	(	(	PUNCT
fcis-26720	184	33	icnidc	icnidc	NOUN
fcis-26720	184	34	)	)	PUNCT
fcis-26720	184	35	.	.	PUNCT
fcis-26720	185	1	ieee	ieee	PROPN
fcis-26720	185	2	,	,	PUNCT
fcis-26720	185	3	2018	2018	NUM
fcis-26720	185	4	:	:	PUNCT
fcis-26720	185	5	193	193	NUM
fcis-26720	185	6	-	-	SYM
fcis-26720	185	7	197	197	NUM
fcis-26720	185	8	.	.	PUNCT
fcis-26720	186	1	[	[	X
fcis-26720	186	2	11	11	NUM
fcis-26720	186	3	]	]	X
fcis-26720	186	4	guo	guo	PROPN
fcis-26720	186	5	l	l	PROPN
fcis-26720	186	6	,	,	PUNCT
fcis-26720	186	7	wang	wang	PROPN
fcis-26720	186	8	l	l	PROPN
fcis-26720	186	9	,	,	PUNCT
fcis-26720	186	10	dang	dang	PROPN
fcis-26720	186	11	j	j	PROPN
fcis-26720	186	12	,	,	PUNCT
fcis-26720	186	13	et	et	PROPN
fcis-26720	186	14	al	al	PROPN
fcis-26720	186	15	.	.	PUNCT
fcis-26720	187	1	a	a	DET
fcis-26720	187	2	feature	feature	NOUN
fcis-26720	187	3	fusion	fusion	NOUN
fcis-26720	187	4	method	method	NOUN
fcis-26720	187	5	based	base	VERB
fcis-26720	187	6	on	on	ADP
fcis-26720	187	7	extreme	extreme	ADJ
fcis-26720	187	8	learning	learning	NOUN
fcis-26720	187	9	machine	machine	NOUN
fcis-26720	187	10	for	for	ADP
fcis-26720	187	11	speech	speech	NOUN
fcis-26720	187	12	emotion	emotion	NOUN
fcis-26720	187	13	recognition	recognition	NOUN
fcis-26720	187	14	[	[	X
fcis-26720	187	15	c]//2018	c]//2018	NOUN
fcis-26720	187	16	ieee	ieee	PROPN
fcis-26720	187	17	international	international	ADJ
fcis-26720	187	18	conference	conference	NOUN
fcis-26720	187	19	on	on	ADP
fcis-26720	187	20	acoustics	acoustic	NOUN
fcis-26720	187	21	,	,	PUNCT
fcis-26720	187	22	speech	speech	NOUN
fcis-26720	187	23	and	and	CCONJ
fcis-26720	187	24	signal	signal	NOUN
fcis-26720	187	25	processing	processing	NOUN
fcis-26720	187	26	(	(	PUNCT
fcis-26720	187	27	icassp	icassp	PROPN
fcis-26720	187	28	)	)	PUNCT
fcis-26720	187	29	.	.	PUNCT
fcis-26720	188	1	ieee	ieee	NOUN
fcis-26720	188	2	,	,	PUNCT
fcis-26720	188	3	2018	2018	NUM
fcis-26720	188	4	:	:	PUNCT
fcis-26720	188	5	2666	2666	NUM
fcis-26720	188	6	−	−	NOUN
fcis-26720	188	7	2670	2670	NUM
fcis-26720	188	8	.	.	PUNCT
fcis-26720	189	1	[	[	X
fcis-26720	189	2	12	12	NUM
fcis-26720	189	3	]	]	X
fcis-26720	189	4	bandela	bandela	NOUN
fcis-26720	189	5	s	s	PROPN
fcis-26720	189	6	r	r	PROPN
fcis-26720	189	7	,	,	PUNCT
fcis-26720	189	8	kumar	kumar	PROPN
fcis-26720	189	9	t	t	PROPN
fcis-26720	189	10	k.	k.	PROPN
fcis-26720	189	11	stressed	stress	VERB
fcis-26720	189	12	speech	speech	NOUN
fcis-26720	189	13	emotion	emotion	NOUN
fcis-26720	189	14	recognition	recognition	NOUN
fcis-26720	189	15	using	use	VERB
fcis-26720	189	16	feature	feature	NOUN
fcis-26720	189	17	fusion	fusion	NOUN
fcis-26720	189	18	of	of	ADP
fcis-26720	189	19	teager	teager	NOUN
fcis-26720	189	20	energy	energy	NOUN
fcis-26720	189	21	operator	operator	NOUN
fcis-26720	189	22	and	and	CCONJ
fcis-26720	189	23	mfcc	mfcc	NOUN
fcis-26720	190	1	[	[	X
fcis-26720	190	2	c]//2017	c]//2017	X
fcis-26720	190	3	8th	8th	ADJ
fcis-26720	190	4	international	international	ADJ
fcis-26720	190	5	conference	conference	NOUN
fcis-26720	190	6	on	on	ADP
fcis-26720	190	7	computing	computing	NOUN
fcis-26720	190	8	,	,	PUNCT
fcis-26720	190	9	communication	communication	NOUN
fcis-26720	190	10	and	and	CCONJ
fcis-26720	190	11	networking	networking	NOUN
fcis-26720	190	12	technologies	technology	NOUN
fcis-26720	190	13	(	(	PUNCT
fcis-26720	190	14	icccnt	icccnt	ADJ
fcis-26720	190	15	)	)	PUNCT
fcis-26720	190	16	.	.	PUNCT
fcis-26720	191	1	ieee	ieee	PROPN
fcis-26720	191	2	,	,	PUNCT
fcis-26720	191	3	2017	2017	NUM
fcis-26720	191	4	:	:	SYM
fcis-26720	191	5	1	1	NUM
fcis-26720	191	6	-	-	SYM
fcis-26720	191	7	5	5	NUM
fcis-26720	191	8	.	.	PUNCT
fcis-26720	192	1	[	[	X
fcis-26720	192	2	13	13	NUM
fcis-26720	192	3	]	]	PUNCT
fcis-26720	192	4	lieskovská	lieskovská	ADJ
fcis-26720	192	5	e	e	PROPN
fcis-26720	192	6	,	,	PUNCT
fcis-26720	192	7	jakubec	jakubec	PROPN
fcis-26720	192	8	m	m	PROPN
fcis-26720	192	9	,	,	PUNCT
fcis-26720	192	10	jarina	jarina	PROPN
fcis-26720	192	11	r	r	PROPN
fcis-26720	192	12	,	,	PUNCT
fcis-26720	192	13	et	et	PROPN
fcis-26720	192	14	al	al	PROPN
fcis-26720	192	15	.	.	PUNCT
fcis-26720	193	1	a	a	DET
fcis-26720	193	2	review	review	NOUN
fcis-26720	193	3	on	on	ADP
fcis-26720	193	4	speech	speech	NOUN
fcis-26720	193	5	emotion	emotion	NOUN
fcis-26720	193	6	recognition	recognition	NOUN
fcis-26720	193	7	using	use	VERB
fcis-26720	193	8	deep	deep	ADJ
fcis-26720	193	9	learning	learning	NOUN
fcis-26720	193	10	and	and	CCONJ
fcis-26720	193	11	attention	attention	NOUN
fcis-26720	193	12	mechanism	mechanism	NOUN
fcis-26720	194	1	[	[	X
fcis-26720	194	2	j	j	X
fcis-26720	194	3	]	]	X
fcis-26720	194	4	.	.	PUNCT
fcis-26720	195	1	electronics	electronic	NOUN
fcis-26720	195	2	,	,	PUNCT
fcis-26720	195	3	2021	2021	NUM
fcis-26720	195	4	,	,	PUNCT
fcis-26720	195	5	10	10	NUM
fcis-26720	195	6	(	(	PUNCT
fcis-26720	195	7	10	10	NUM
fcis-26720	195	8	):	):	PUNCT
fcis-26720	195	9	1163	1163	NUM
fcis-26720	195	10	.	.	PUNCT
fcis-26720	196	1	[	[	X
fcis-26720	196	2	14	14	NUM
fcis-26720	196	3	]	]	X
fcis-26720	196	4	niu	niu	PROPN
fcis-26720	196	5	z	z	PROPN
fcis-26720	196	6	,	,	PUNCT
fcis-26720	196	7	zhong	zhong	PROPN
fcis-26720	196	8	g	g	PROPN
fcis-26720	196	9	,	,	PUNCT
fcis-26720	196	10	yu	yu	PROPN
fcis-26720	196	11	h.	h.	PROPN
fcis-26720	196	12	a	a	DET
fcis-26720	196	13	review	review	NOUN
fcis-26720	196	14	on	on	ADP
fcis-26720	196	15	the	the	DET
fcis-26720	196	16	attention	attention	NOUN
fcis-26720	196	17	mechanism	mechanism	NOUN
fcis-26720	196	18	of	of	ADP
fcis-26720	196	19	deep	deep	ADJ
fcis-26720	196	20	learning	learning	NOUN
fcis-26720	196	21	[	[	X
fcis-26720	196	22	j	j	X
fcis-26720	196	23	]	]	X
fcis-26720	196	24	.	.	PUNCT
fcis-26720	197	1	neurocomputing	neurocomputing	NOUN
fcis-26720	197	2	,	,	PUNCT
fcis-26720	197	3	2021	2021	NUM
fcis-26720	197	4	,	,	PUNCT
fcis-26720	197	5	452	452	NUM
fcis-26720	197	6	:	:	SYM
fcis-26720	197	7	48	48	NUM
fcis-26720	197	8	-	-	SYM
fcis-26720	197	9	62	62	NUM
fcis-26720	197	10	.	.	PUNCT
fcis-26720	198	1	[	[	X
fcis-26720	198	2	15	15	NUM
fcis-26720	198	3	]	]	X
fcis-26720	198	4	vaswani	vaswani	ADJ
fcis-26720	198	5	a.	a.	NOUN
fcis-26720	198	6	attention	attention	NOUN
fcis-26720	198	7	is	be	AUX
fcis-26720	198	8	all	all	PRON
fcis-26720	198	9	you	you	PRON
fcis-26720	198	10	need	need	VERB
fcis-26720	198	11	[	[	X
fcis-26720	198	12	j	j	X
fcis-26720	198	13	]	]	X
fcis-26720	198	14	.	.	PUNCT
fcis-26720	199	1	advances	advance	NOUN
fcis-26720	199	2	in	in	ADP
fcis-26720	199	3	neural	neural	ADJ
fcis-26720	199	4	information	information	NOUN
fcis-26720	199	5	processing	processing	NOUN
fcis-26720	199	6	systems	system	NOUN
fcis-26720	199	7	,	,	PUNCT
fcis-26720	199	8	2017	2017	NUM
fcis-26720	199	9	.	.	PUNCT
fcis-26720	200	1	[	[	X
fcis-26720	200	2	16	16	NUM
fcis-26720	200	3	]	]	X
fcis-26720	200	4	ma	ma	PROPN
fcis-26720	200	5	j	j	PROPN
fcis-26720	200	6	,	,	PUNCT
fcis-26720	200	7	tang	tang	PROPN
fcis-26720	200	8	h	h	PROPN
fcis-26720	200	9	,	,	PUNCT
fcis-26720	200	10	zheng	zheng	PROPN
fcis-26720	200	11	w	w	PROPN
fcis-26720	200	12	l	l	PROPN
fcis-26720	200	13	,	,	PUNCT
fcis-26720	200	14	et	et	PROPN
fcis-26720	200	15	al	al	PROPN
fcis-26720	200	16	.	.	PUNCT
fcis-26720	201	1	emotion	emotion	NOUN
fcis-26720	201	2	recognition	recognition	NOUN
fcis-26720	201	3	using	use	VERB
fcis-26720	201	4	multimodal	multimodal	ADJ
fcis-26720	201	5	residual	residual	ADJ
fcis-26720	201	6	lstm	lstm	NOUN
fcis-26720	201	7	network	network	NOUN
fcis-26720	201	8	[	[	X
fcis-26720	201	9	c]//proceedings	c]//proceeding	NOUN
fcis-26720	201	10	of	of	ADP
fcis-26720	201	11	the	the	DET
fcis-26720	201	12	27th	27th	ADJ
fcis-26720	201	13	acm	acm	PROPN
fcis-26720	201	14	international	international	ADJ
fcis-26720	201	15	conference	conference	NOUN
fcis-26720	201	16	on	on	ADP
fcis-26720	201	17	multimedia	multimedia	NOUN
fcis-26720	201	18	.	.	PUNCT
fcis-26720	202	1	2019	2019	NUM
fcis-26720	202	2	:	:	PUNCT
fcis-26720	202	3	176183	176183	NUM
fcis-26720	202	4	.	.	PUNCT
fcis-26720	203	1	[	[	X
fcis-26720	203	2	17	17	NUM
fcis-26720	203	3	]	]	PUNCT
fcis-26720	203	4	xie	xie	PROPN
fcis-26720	203	5	y	y	PROPN
fcis-26720	203	6	,	,	PUNCT
fcis-26720	203	7	liang	liang	PROPN
fcis-26720	203	8	r	r	PROPN
fcis-26720	203	9	,	,	PUNCT
fcis-26720	203	10	liang	liang	PROPN
fcis-26720	203	11	z	z	PROPN
fcis-26720	203	12	,	,	PUNCT
fcis-26720	203	13	et	et	PROPN
fcis-26720	203	14	al	al	PROPN
fcis-26720	203	15	.	.	PROPN
fcis-26720	203	16	speech	speech	NOUN
fcis-26720	203	17	emotion	emotion	NOUN
fcis-26720	203	18	classification	classification	NOUN
fcis-26720	203	19	using	use	VERB
fcis-26720	203	20	attention	attention	NOUN
fcis-26720	203	21	-	-	PUNCT
fcis-26720	203	22	based	base	VERB
fcis-26720	203	23	lstm	lstm	NOUN
fcis-26720	204	1	[	[	X
fcis-26720	204	2	j	j	X
fcis-26720	204	3	]	]	X
fcis-26720	204	4	.	.	PUNCT
fcis-26720	205	1	ieee	ieee	PROPN
fcis-26720	205	2	/	/	SYM
fcis-26720	205	3	acm	acm	PROPN
fcis-26720	205	4	transactions	transaction	NOUN
fcis-26720	205	5	on	on	ADP
fcis-26720	205	6	audio	audio	NOUN
fcis-26720	205	7	,	,	PUNCT
fcis-26720	205	8	speech	speech	NOUN
fcis-26720	205	9	,	,	PUNCT
fcis-26720	205	10	and	and	CCONJ
fcis-26720	205	11	language	language	NOUN
fcis-26720	205	12	processing	processing	NOUN
fcis-26720	205	13	,	,	PUNCT
fcis-26720	205	14	2019	2019	NUM
fcis-26720	205	15	,	,	PUNCT
fcis-26720	205	16	27	27	NUM
fcis-26720	205	17	(	(	PUNCT
fcis-26720	205	18	11	11	NUM
fcis-26720	205	19	):	):	PUNCT
fcis-26720	205	20	16751685	16751685	NUM
fcis-26720	205	21	.	.	PUNCT
fcis-26720	206	1	[	[	X
fcis-26720	206	2	18	18	NUM
fcis-26720	206	3	]	]	X
fcis-26720	206	4	pengmao	pengmao	PROPN
fcis-26720	206	5	tashi	tashi	PROPN
fcis-26720	206	6	,	,	PUNCT
fcis-26720	206	7	cai	cai	PROPN
fcis-26720	206	8	zhijie	zhijie	PROPN
fcis-26720	206	9	,	,	PUNCT
fcis-26720	206	10	cai	cai	PROPN
fcis-26720	206	11	rang	ring	VERB
fcis-26720	206	12	zhuoma	zhuoma	PROPN
fcis-26720	206	13	.	.	PUNCT
fcis-26720	207	1	construction	construction	NOUN
fcis-26720	207	2	of	of	ADP
fcis-26720	207	3	tibetan	tibetan	ADJ
fcis-26720	207	4	emotional	emotional	ADJ
fcis-26720	207	5	speech	speech	NOUN
fcis-26720	207	6	database	database	NOUN
fcis-26720	208	1	[	[	X
fcis-26720	208	2	j	j	X
fcis-26720	208	3	]	]	X
fcis-26720	208	4	.	.	PUNCT
fcis-26720	209	1	journal	journal	PROPN
fcis-26720	209	2	of	of	ADP
fcis-26720	209	3	peking	peking	PROPN
fcis-26720	209	4	university	university	PROPN
fcis-26720	209	5	(	(	PUNCT
fcis-26720	209	6	natural	natural	ADJ
fcis-26720	209	7	science	science	NOUN
fcis-26720	209	8	edition	edition	NOUN
fcis-26720	209	9	)	)	PUNCT
fcis-26720	209	10	,	,	PUNCT
fcis-26720	209	11	2023	2023	NUM
fcis-26720	209	12	,	,	PUNCT
fcis-26720	209	13	59	59	NUM
fcis-26720	209	14	(	(	PUNCT
fcis-26720	209	15	05	05	NUM
fcis-26720	209	16	):	):	PUNCT
fcis-26720	209	17	773	773	NUM
fcis-26720	209	18	-	-	SYM
fcis-26720	209	19	781	781	NUM
fcis-26720	209	20	.	.	PUNCT
fcis-26720	210	1	doi	doi	NOUN
fcis-26720	210	2	:	:	PUNCT
fcis-26720	210	3	10.13209	10.13209	NUM
fcis-26720	210	4	/	/	SYM
fcis-26720	210	5	j.0479	j.0479	PROPN
fcis-26720	210	6	-	-	PUNCT
fcis-26720	210	7	8023.2022	8023.2022	NOUN
fcis-26720	210	8	.	.	PUNCT
fcis-26720	211	1	121	121	NUM
fcis-26720	211	2	.	.	PUNCT
fcis-26720	212	1	[	[	X
fcis-26720	212	2	19	19	NUM
fcis-26720	212	3	]	]	PUNCT
fcis-26720	212	4	gu	gu	PROPN
fcis-26720	212	5	zeyue	zeyue	PROPN
fcis-26720	212	6	,	,	PUNCT
fcis-26720	212	7	bianbawangdui	bianbawangdui	PROPN
fcis-26720	212	8	,	,	PUNCT
fcis-26720	212	9	qi	qi	PROPN
fcis-26720	212	10	jindong	jindong	PROPN
fcis-26720	212	11	.	.	PUNCT
fcis-26720	213	1	tibetan	tibetan	ADJ
fcis-26720	213	2	speech	speech	PROPN
fcis-26720	213	3	emotion	emotion	NOUN
fcis-26720	213	4	recognition	recognition	NOUN
fcis-26720	213	5	based	base	VERB
fcis-26720	213	6	on	on	ADP
fcis-26720	213	7	multi	multi	ADJ
fcis-26720	213	8	-	-	ADJ
fcis-26720	213	9	feature	feature	ADJ
fcis-26720	213	10	fusion	fusion	NOUN
fcis-26720	214	1	[	[	X
fcis-26720	214	2	j	j	X
fcis-26720	214	3	]	]	X
fcis-26720	214	4	.	.	PUNCT
fcis-26720	215	1	modern	modern	ADJ
fcis-26720	215	2	electronic	electronic	ADJ
fcis-26720	215	3	technology	technology	NOUN
fcis-26720	215	4	,	,	PUNCT
fcis-26720	215	5	2023	2023	NUM
fcis-26720	215	6	,	,	PUNCT
fcis-26720	215	7	46	46	NUM
fcis-26720	215	8	(	(	PUNCT
fcis-26720	215	9	21	21	NUM
fcis-26720	215	10	):	):	PUNCT
fcis-26720	215	11	129	129	NUM
fcis-26720	215	12	-	-	SYM
fcis-26720	215	13	133	133	NUM
fcis-26720	215	14	.	.	PUNCT
fcis-26720	216	1	doi	doi	NOUN
fcis-26720	216	2	:	:	PUNCT
fcis-26720	216	3	10	10	NUM
fcis-26720	216	4	.	.	PUNCT
fcis-26720	217	1	16652/	16652/	NUM
fcis-26720	217	2	j.issn.1004	j.issn.1004	NOUN
fcis-26720	217	3	-	-	PUNCT
fcis-26720	217	4	373x	373x	NUM
fcis-26720	217	5	.	.	PUNCT
fcis-26720	218	1	2023.21	2023.21	NUM
fcis-26720	218	2	.	.	NOUN
fcis-26720	218	3	024	024	NUM
fcis-26720	218	4	.	.	PUNCT
fcis-26720	219	1	[	[	X
fcis-26720	219	2	20	20	NUM
fcis-26720	219	3	]	]	X
fcis-26720	219	4	sabour	sabour	NOUN
fcis-26720	219	5	s	s	PROPN
fcis-26720	219	6	,	,	PUNCT
fcis-26720	219	7	frosst	frosst	ADJ
fcis-26720	219	8	n	n	CCONJ
fcis-26720	219	9	,	,	PUNCT
fcis-26720	219	10	hinton	hinton	PROPN
fcis-26720	219	11	g	g	PROPN
fcis-26720	219	12	e.	e.	PROPN
fcis-26720	219	13	dynamic	dynamic	ADJ
fcis-26720	219	14	routing	routing	NOUN
fcis-26720	219	15	between	between	ADP
fcis-26720	219	16	capsules	capsule	NOUN
fcis-26720	219	17	[	[	X
fcis-26720	219	18	j	j	X
fcis-26720	219	19	]	]	X
fcis-26720	219	20	.	.	PUNCT
fcis-26720	220	1	advances	advance	NOUN
fcis-26720	220	2	in	in	ADP
fcis-26720	220	3	neural	neural	ADJ
fcis-26720	220	4	information	information	NOUN
fcis-26720	220	5	processing	processing	NOUN
fcis-26720	220	6	systems	system	NOUN
fcis-26720	220	7	,	,	PUNCT
fcis-26720	220	8	2017	2017	NUM
fcis-26720	220	9	,	,	PUNCT
fcis-26720	220	10	30	30	NUM
fcis-26720	220	11	.	.	PUNCT
fcis-26720	221	1	[	[	X
fcis-26720	221	2	21	21	NUM
fcis-26720	221	3	]	]	X
fcis-26720	221	4	lian	lian	PROPN
fcis-26720	221	5	z	z	NOUN
fcis-26720	221	6	,	,	PUNCT
fcis-26720	221	7	liu	liu	PROPN
fcis-26720	221	8	b	b	PROPN
fcis-26720	221	9	,	,	PUNCT
fcis-26720	221	10	tao	tao	PROPN
fcis-26720	221	11	j.	j.	PROPN
fcis-26720	221	12	decn	decn	PROPN
fcis-26720	221	13	:	:	PUNCT
fcis-26720	221	14	dialogical	dialogical	ADJ
fcis-26720	221	15	emotion	emotion	NOUN
fcis-26720	221	16	correction	correction	NOUN
fcis-26720	221	17	network	network	NOUN
fcis-26720	221	18	for	for	ADP
fcis-26720	221	19	conversational	conversational	ADJ
fcis-26720	221	20	emotion	emotion	NOUN
fcis-26720	221	21	recognition	recognition	NOUN
fcis-26720	222	1	[	[	X
fcis-26720	222	2	j	j	X
fcis-26720	222	3	]	]	X
fcis-26720	222	4	.	.	PUNCT
fcis-26720	223	1	neurocomputing	neurocomputing	NOUN
fcis-26720	223	2	,	,	PUNCT
fcis-26720	223	3	2021	2021	NUM
fcis-26720	223	4	,	,	PUNCT
fcis-26720	223	5	454	454	NUM
fcis-26720	223	6	:	:	SYM
fcis-26720	223	7	483	483	NUM
fcis-26720	223	8	-	-	SYM
fcis-26720	223	9	495	495	NUM
fcis-26720	223	10	.	.	PUNCT
fcis-26720	224	1	[	[	X
fcis-26720	224	2	22	22	NUM
fcis-26720	224	3	]	]	X
fcis-26720	224	4	li	li	PROPN
fcis-26720	224	5	s	s	PROPN
fcis-26720	224	6	,	,	PUNCT
fcis-26720	224	7	xing	xing	PROPN
fcis-26720	224	8	x	x	PROPN
fcis-26720	224	9	,	,	PUNCT
fcis-26720	224	10	fan	fan	PROPN
fcis-26720	224	11	w	w	PROPN
fcis-26720	224	12	,	,	PUNCT
fcis-26720	224	13	et	et	PROPN
fcis-26720	224	14	al	al	PROPN
fcis-26720	224	15	.	.	PROPN
fcis-26720	225	1	spatiotemporal	spatiotemporal	ADJ
fcis-26720	225	2	and	and	CCONJ
fcis-26720	225	3	frequential	frequential	ADJ
fcis-26720	225	4	cascaded	cascade	VERB
fcis-26720	225	5	attention	attention	NOUN
fcis-26720	225	6	networks	network	NOUN
fcis-26720	225	7	for	for	ADP
fcis-26720	225	8	speech	speech	NOUN
fcis-26720	225	9	emotion	emotion	NOUN
fcis-26720	225	10	recognition	recognition	NOUN
fcis-26720	226	1	[	[	X
fcis-26720	226	2	j	j	X
fcis-26720	226	3	]	]	X
fcis-26720	226	4	.	.	PUNCT
fcis-26720	227	1	neurocomputing	neurocomputing	NOUN
fcis-26720	227	2	,	,	PUNCT
fcis-26720	227	3	2021	2021	NUM
fcis-26720	227	4	,	,	PUNCT
fcis-26720	227	5	448	448	NUM
fcis-26720	227	6	:	:	SYM
fcis-26720	227	7	238	238	NUM
fcis-26720	227	8	-	-	SYM
fcis-26720	227	9	248	248	NUM
fcis-26720	227	10	.	.	PUNCT
fcis-26720	228	1	[	[	X
fcis-26720	228	2	23	23	NUM
fcis-26720	228	3	]	]	X
fcis-26720	228	4	chen	chen	PROPN
fcis-26720	228	5	j	j	PROPN
fcis-26720	228	6	,	,	PUNCT
fcis-26720	228	7	liu	liu	PROPN
fcis-26720	228	8	z.	z.	PROPN
fcis-26720	228	9	mask	mask	VERB
fcis-26720	228	10	dynamic	dynamic	ADJ
fcis-26720	228	11	routing	routing	NOUN
fcis-26720	228	12	to	to	ADP
fcis-26720	228	13	combined	combined	ADJ
fcis-26720	228	14	model	model	NOUN
fcis-26720	228	15	of	of	ADP
fcis-26720	228	16	deep	deep	ADJ
fcis-26720	228	17	capsule	capsule	NOUN
fcis-26720	228	18	network	network	NOUN
fcis-26720	228	19	and	and	CCONJ
fcis-26720	228	20	u	u	NOUN
fcis-26720	228	21	-	-	NOUN
fcis-26720	228	22	net	net	ADJ
fcis-26720	229	1	[	[	X
fcis-26720	229	2	j	j	X
fcis-26720	229	3	]	]	X
fcis-26720	229	4	.	.	PUNCT
fcis-26720	230	1	ieee	ieee	NOUN
fcis-26720	230	2	transactions	transaction	NOUN
fcis-26720	230	3	on	on	ADP
fcis-26720	230	4	neural	neural	ADJ
fcis-26720	230	5	networks	network	NOUN
fcis-26720	230	6	and	and	CCONJ
fcis-26720	230	7	learning	learning	NOUN
fcis-26720	230	8	systems	system	NOUN
fcis-26720	230	9	,	,	PUNCT
fcis-26720	230	10	2020	2020	NUM
fcis-26720	230	11	,	,	PUNCT
fcis-26720	230	12	31	31	NUM
fcis-26720	230	13	(	(	PUNCT
fcis-26720	230	14	7	7	NUM
fcis-26720	230	15	):	):	PUNCT
fcis-26720	230	16	2653	2653	NUM
fcis-26720	230	17	-	-	SYM
fcis-26720	230	18	2664	2664	NUM
fcis-26720	230	19	.	.	PUNCT
fcis-26720	231	1	[	[	X
fcis-26720	231	2	24	24	NUM
fcis-26720	231	3	]	]	X
fcis-26720	231	4	wen	wen	NOUN
fcis-26720	231	5	x	x	PROPN
fcis-26720	231	6	c	c	X
fcis-26720	231	7	,	,	PUNCT
fcis-26720	231	8	ye	ye	PRON
fcis-26720	231	9	j	j	PROPN
fcis-26720	231	10	x	x	PROPN
fcis-26720	231	11	,	,	PUNCT
fcis-26720	231	12	luo	luo	PROPN
fcis-26720	231	13	y	y	PROPN
fcis-26720	231	14	,	,	PUNCT
fcis-26720	231	15	et	et	PROPN
fcis-26720	231	16	al	al	PROPN
fcis-26720	231	17	.	.	PUNCT
fcis-26720	231	18	ctl	ctl	PROPN
fcis-26720	231	19	-	-	PUNCT
fcis-26720	231	20	mtnet	mtnet	PROPN
fcis-26720	231	21	:	:	PUNCT
fcis-26720	231	22	a	a	DET
fcis-26720	231	23	novel	novel	ADJ
fcis-26720	231	24	capsnet	capsnet	NOUN
fcis-26720	231	25	and	and	CCONJ
fcis-26720	231	26	transfer	transfer	VERB
fcis-26720	231	27	learning	learning	NOUN
fcis-26720	231	28	-	-	PUNCT
fcis-26720	231	29	based	base	VERB
fcis-26720	231	30	mixed	mixed	ADJ
fcis-26720	231	31	task	task	NOUN
fcis-26720	231	32	net	net	NOUN
fcis-26720	231	33	for	for	ADP
fcis-26720	231	34	the	the	DET
fcis-26720	231	35	single	single	ADJ
fcis-26720	231	36	-	-	PUNCT
fcis-26720	231	37	corpus	corpus	NOUN
fcis-26720	231	38	and	and	CCONJ
fcis-26720	231	39	cross	cross	ADJ
fcis-26720	231	40	-	-	ADJ
fcis-26720	231	41	corpus	corpus	ADJ
fcis-26720	231	42	speech	speech	NOUN
fcis-26720	231	43	emotion	emotion	NOUN
fcis-26720	231	44	recognition	recognition	NOUN
fcis-26720	231	45	[	[	X
fcis-26720	231	46	j	j	X
fcis-26720	231	47	]	]	X
fcis-26720	231	48	.	.	PUNCT
fcis-26720	232	1	arxiv	arxiv	PROPN
fcis-26720	232	2	preprint	preprint	PROPN
fcis-26720	232	3	arxiv	arxiv	PROPN
fcis-26720	232	4	:	:	PUNCT
fcis-26720	232	5	2207.10644	2207.10644	NUM
fcis-26720	232	6	,	,	PUNCT
fcis-26720	232	7	2022	2022	NUM
fcis-26720	232	8	.	.	PUNCT
fcis-26720	233	1	[	[	X
fcis-26720	233	2	25	25	NUM
fcis-26720	233	3	]	]	X
fcis-26720	233	4	livingstone	livingstone	PROPN
fcis-26720	233	5	s	s	PART
fcis-26720	233	6	r	r	NOUN
fcis-26720	233	7	,	,	PUNCT
fcis-26720	233	8	russo	russo	PROPN
fcis-26720	233	9	f	f	PROPN
fcis-26720	233	10	a.	a.	NOUN
fcis-26720	233	11	the	the	DET
fcis-26720	233	12	ryerson	ryerson	NOUN
fcis-26720	233	13	audio	audio	ADJ
fcis-26720	233	14	-	-	ADJ
fcis-26720	233	15	visual	visual	ADJ
fcis-26720	233	16	database	database	NOUN
fcis-26720	233	17	of	of	ADP
fcis-26720	233	18	emotional	emotional	ADJ
fcis-26720	233	19	speech	speech	NOUN
fcis-26720	233	20	and	and	CCONJ
fcis-26720	233	21	song	song	NOUN
fcis-26720	233	22	(	(	PUNCT
fcis-26720	233	23	ravdess	ravdess	NOUN
fcis-26720	233	24	):	):	PUNCT
fcis-26720	233	25	a	a	DET
fcis-26720	233	26	dynamic	dynamic	ADJ
fcis-26720	233	27	,	,	PUNCT
fcis-26720	233	28	multimodal	multimodal	ADJ
fcis-26720	233	29	set	set	NOUN
fcis-26720	233	30	of	of	ADP
fcis-26720	233	31	facial	facial	ADJ
fcis-26720	233	32	and	and	CCONJ
fcis-26720	233	33	vocal	vocal	ADJ
fcis-26720	233	34	expressions	expression	NOUN
fcis-26720	233	35	in	in	ADP
fcis-26720	233	36	north	north	ADJ
fcis-26720	233	37	american	american	ADJ
fcis-26720	233	38	english	english	PROPN
fcis-26720	234	1	[	[	X
fcis-26720	234	2	j	j	X
fcis-26720	234	3	]	]	X
fcis-26720	234	4	.	.	PUNCT
fcis-26720	235	1	plos	plos	PROPN
fcis-26720	235	2	one	one	NUM
fcis-26720	235	3	,	,	PUNCT
fcis-26720	235	4	2018	2018	NUM
fcis-26720	235	5	,	,	PUNCT
fcis-26720	235	6	13	13	NUM
fcis-26720	235	7	(	(	PUNCT
fcis-26720	235	8	5	5	NUM
fcis-26720	235	9	):	):	PUNCT
fcis-26720	235	10	e0196391	e0196391	PROPN
fcis-26720	235	11	.	.	PUNCT
fcis-26720	236	1	[	[	X
fcis-26720	236	2	26	26	NUM
fcis-26720	236	3	]	]	X
fcis-26720	236	4	qu	qu	X
fcis-26720	236	5	aitangs	aitangs	NOUN
fcis-26720	236	6	"	"	PUNCT
fcis-26720	236	7	research	research	NOUN
fcis-26720	236	8	on	on	ADP
fcis-26720	236	9	tibetan	tibetan	ADJ
fcis-26720	236	10	finals	final	NOUN
fcis-26720	236	11	"	"	PUNCT
fcis-26720	236	12	[	[	X
fcis-26720	236	13	m	m	X
fcis-26720	236	14	]	]	X
fcis-26720	236	15	qinghai	qinghai	PROPN
fcis-26720	236	16	ethnic	ethnic	ADJ
fcis-26720	236	17	publishing	publishing	PROPN
fcis-26720	236	18	house	house	PROPN
fcis-26720	236	19	.	.	PUNCT
fcis-26720	237	1	july	july	PROPN
fcis-26720	237	2	1991	1991	NUM
fcis-26720	237	3	[	[	X
fcis-26720	237	4	27	27	NUM
fcis-26720	237	5	]	]	X
fcis-26720	237	6	yang	yang	PROPN
fcis-26720	237	7	jie	jie	PROPN
fcis-26720	237	8	,	,	PUNCT
fcis-26720	237	9	li	li	PROPN
fcis-26720	237	10	yonghong	yonghong	PROPN
fcis-26720	237	11	,	,	PUNCT
fcis-26720	237	12	hu	hu	PROPN
fcis-26720	237	13	axu	axu	PROPN
fcis-26720	237	14	,	,	PUNCT
fcis-26720	237	15	et	et	PROPN
fcis-26720	237	16	al	al	PROPN
fcis-26720	237	17	.	.	PROPN
fcis-26720	237	18	study	study	PROPN
fcis-26720	237	19	on	on	ADP
fcis-26720	237	20	tone	tone	NOUN
fcis-26720	237	21	and	and	CCONJ
fcis-26720	237	22	laryngeal	laryngeal	ADJ
fcis-26720	237	23	plug	plug	NOUN
fcis-26720	237	24	rhyme	rhyme	ADJ
fcis-26720	237	25	perception	perception	NOUN
fcis-26720	237	26	in	in	ADP
fcis-26720	237	27	tibetan	tibetan	PROPN
fcis-26720	237	28	lhasa	lhasa	PROPN
fcis-26720	237	29	dialect	dialect	NOUN
fcis-26720	238	1	[	[	X
fcis-26720	238	2	j	j	X
fcis-26720	238	3	]	]	X
fcis-26720	238	4	.	.	PUNCT
fcis-26720	239	1	national	national	ADJ
fcis-26720	239	2	languages	language	NOUN
fcis-26720	239	3	,	,	PUNCT
fcis-26720	239	4	2023	2023	NUM
fcis-26720	239	5	,	,	PUNCT
fcis-26720	239	6	(	(	PUNCT
fcis-26720	239	7	04	04	NUM
fcis-26720	239	8	):	):	PUNCT
fcis-26720	239	9	101	101	NUM
fcis-26720	239	10	-	-	SYM
fcis-26720	239	11	110	110	NUM
fcis-26720	239	12	.	.	PUNCT
fcis-26720	240	1	[	[	X
fcis-26720	240	2	28	28	NUM
fcis-26720	240	3	]	]	X
fcis-26720	240	4	ye	ye	PROPN
fcis-26720	240	5	j	j	PROPN
fcis-26720	240	6	,	,	PUNCT
fcis-26720	240	7	wen	wen	PROPN
fcis-26720	240	8	x	x	PROPN
fcis-26720	240	9	c	c	PROPN
fcis-26720	240	10	,	,	PUNCT
fcis-26720	240	11	wei	wei	PROPN
fcis-26720	240	12	y	y	PROPN
fcis-26720	240	13	,	,	PUNCT
fcis-26720	240	14	et	et	PROPN
fcis-26720	240	15	al	al	PROPN
fcis-26720	240	16	.	.	PUNCT
fcis-26720	241	1	temporal	temporal	ADJ
fcis-26720	241	2	modeling	modeling	NOUN
fcis-26720	241	3	matters	matter	NOUN
fcis-26720	241	4	:	:	PUNCT
fcis-26720	241	5	a	a	DET
fcis-26720	241	6	novel	novel	ADJ
fcis-26720	241	7	temporal	temporal	ADJ
fcis-26720	241	8	emotional	emotional	ADJ
fcis-26720	241	9	modeling	modeling	NOUN
fcis-26720	241	10	approach	approach	NOUN
fcis-26720	241	11	for	for	ADP
fcis-26720	241	12	speech	speech	NOUN
fcis-26720	241	13	emotion	emotion	NOUN
fcis-26720	241	14	recognition	recognition	NOUN
fcis-26720	242	1	[	[	X
fcis-26720	242	2	c]//icassp	c]//icassp	PROPN
fcis-26720	242	3	2023	2023	NUM
fcis-26720	242	4	-	-	SYM
fcis-26720	242	5	2023	2023	NUM
fcis-26720	242	6	ieee	ieee	NOUN
fcis-26720	242	7	international	international	ADJ
fcis-26720	242	8	conference	conference	NOUN
fcis-26720	242	9	on	on	ADP
fcis-26720	242	10	acoustics	acoustic	NOUN
fcis-26720	242	11	,	,	PUNCT
fcis-26720	242	12	speech	speech	NOUN
fcis-26720	242	13	and	and	CCONJ
fcis-26720	242	14	signal	signal	NOUN
fcis-26720	242	15	processing	processing	NOUN
fcis-26720	242	16	(	(	PUNCT
fcis-26720	242	17	icassp	icassp	PROPN
fcis-26720	242	18	)	)	PUNCT
fcis-26720	242	19	.	.	PUNCT
fcis-26720	243	1	ieee	ieee	NOUN
fcis-26720	243	2	,	,	PUNCT
fcis-26720	243	3	2023	2023	NUM
fcis-26720	243	4	:	:	PUNCT
fcis-26720	243	5	1	1	NUM
fcis-26720	243	6	-	-	SYM
fcis-26720	243	7	5	5	NUM
fcis-26720	243	8	.	.	PUNCT
fcis-26720	244	1	[	[	X
fcis-26720	244	2	29	29	NUM
fcis-26720	244	3	]	]	PUNCT
fcis-26720	244	4	sadok	sadok	PROPN
fcis-26720	244	5	s	s	PART
fcis-26720	244	6	,	,	PUNCT
fcis-26720	244	7	leglaive	leglaive	PROPN
fcis-26720	244	8	s	s	PROPN
fcis-26720	244	9	,	,	PUNCT
fcis-26720	244	10	séguier	séguier	PROPN
fcis-26720	244	11	r.	r.	PROPN
fcis-26720	244	12	a	a	DET
fcis-26720	244	13	vector	vector	NOUN
fcis-26720	244	14	quantized	quantize	VERB
fcis-26720	244	15	masked	mask	VERB
fcis-26720	244	16	autoencoder	autoencoder	NOUN
fcis-26720	244	17	for	for	ADP
fcis-26720	244	18	speech	speech	NOUN
fcis-26720	244	19	emotion	emotion	NOUN
fcis-26720	244	20	recognition	recognition	NOUN
fcis-26720	244	21	[	[	X
fcis-26720	244	22	c]//2023	c]//2023	PROPN
fcis-26720	244	23	ieee	ieee	NOUN
fcis-26720	244	24	international	international	ADJ
fcis-26720	244	25	conference	conference	NOUN
fcis-26720	244	26	on	on	ADP
fcis-26720	244	27	acoustics	acoustic	NOUN
fcis-26720	244	28	,	,	PUNCT
fcis-26720	244	29	speech	speech	NOUN
fcis-26720	244	30	,	,	PUNCT
fcis-26720	244	31	and	and	CCONJ
fcis-26720	244	32	signal	signal	NOUN
fcis-26720	244	33	processing	processing	NOUN
fcis-26720	244	34	workshops	workshop	NOUN
fcis-26720	244	35	(	(	PUNCT
fcis-26720	244	36	icasspw	icasspw	NOUN
fcis-26720	244	37	)	)	PUNCT
fcis-26720	244	38	.	.	PUNCT
fcis-26720	245	1	ieee	ieee	NOUN
fcis-26720	245	2	,	,	PUNCT
fcis-26720	245	3	2023	2023	NUM
fcis-26720	245	4	:	:	PUNCT
fcis-26720	245	5	1	1	NUM
fcis-26720	245	6	-	-	SYM
fcis-26720	245	7	5	5	NUM
fcis-26720	245	8	.	.	PUNCT