id	sid	tid	token	lemma	pos
fcis-5510	1	1	frontiers	frontier	NOUN
fcis-5510	1	2	in	in	ADP
fcis-5510	1	3	computing	computing	NOUN
fcis-5510	1	4	and	and	CCONJ
fcis-5510	1	5	intelligent	intelligent	ADJ
fcis-5510	1	6	systems	system	NOUN
fcis-5510	1	7	issn	issn	VERB
fcis-5510	1	8	:	:	PUNCT
fcis-5510	1	9	2832	2832	NUM
fcis-5510	1	10	-	-	SYM
fcis-5510	1	11	6024	6024	NUM
fcis-5510	1	12	|	|	NOUN
fcis-5510	1	13	vol	vol	NOUN
fcis-5510	1	14	.	.	PROPN
fcis-5510	2	1	2	2	NUM
fcis-5510	2	2	,	,	PUNCT
fcis-5510	2	3	no	no	INTJ
fcis-5510	2	4	.	.	NOUN
fcis-5510	2	5	3	3	NUM
fcis-5510	2	6	,	,	PUNCT
fcis-5510	2	7	2022	2022	NUM
fcis-5510	2	8	123	123	NUM
fcis-5510	2	9	research	research	NOUN
fcis-5510	2	10	on	on	ADP
fcis-5510	2	11	algorithm	algorithm	NOUN
fcis-5510	2	12	of	of	ADP
fcis-5510	2	13	video	video	NOUN
fcis-5510	2	14	analysis	analysis	NOUN
fcis-5510	2	15	system	system	NOUN
fcis-5510	2	16	based	base	VERB
fcis-5510	2	17	on	on	ADP
fcis-5510	2	18	text	text	NOUN
fcis-5510	2	19	error	error	NOUN
fcis-5510	2	20	correction	correction	NOUN
fcis-5510	2	21	jinjin	jinjin	NOUN
fcis-5510	2	22	wang	wang	PROPN
fcis-5510	2	23	*	*	PROPN
fcis-5510	2	24	,	,	PUNCT
fcis-5510	2	25	yang	yang	PROPN
fcis-5510	2	26	qin	qin	PROPN
fcis-5510	2	27	,	,	PUNCT
fcis-5510	2	28	jiahao	jiahao	PROPN
fcis-5510	2	29	shi	shi	PROPN
fcis-5510	2	30	,	,	PUNCT
fcis-5510	2	31	jiachen	jiachen	PROPN
fcis-5510	2	32	luo	luo	PROPN
fcis-5510	2	33	,	,	PUNCT
fcis-5510	2	34	guo	guo	PROPN
fcis-5510	2	35	huang	huang	PROPN
fcis-5510	2	36	,	,	PUNCT
fcis-5510	2	37	jiaqi	jiaqi	PROPN
fcis-5510	2	38	lu	lu	PROPN
fcis-5510	2	39	school	school	PROPN
fcis-5510	2	40	of	of	ADP
fcis-5510	2	41	information	information	NOUN
fcis-5510	2	42	engineering	engineering	NOUN
fcis-5510	2	43	,	,	PUNCT
fcis-5510	2	44	yancheng	yancheng	PROPN
fcis-5510	2	45	teachers	teachers	PROPN
fcis-5510	2	46	university	university	PROPN
fcis-5510	2	47	,	,	PUNCT
fcis-5510	2	48	yancheng	yancheng	PROPN
fcis-5510	2	49	,	,	PUNCT
fcis-5510	2	50	224000	224000	NUM
fcis-5510	2	51	,	,	PUNCT
fcis-5510	2	52	china	china	PROPN
fcis-5510	2	53	*	*	PUNCT
fcis-5510	2	54	corresponding	correspond	VERB
fcis-5510	2	55	author	author	NOUN
fcis-5510	2	56	:	:	PUNCT
fcis-5510	2	57	jinjin	jinjin	PROPN
fcis-5510	2	58	wang	wang	PROPN
fcis-5510	2	59	(	(	PUNCT
fcis-5510	2	60	email	email	NOUN
fcis-5510	2	61	:	:	PUNCT
fcis-5510	2	62	wangjj@yctu.edu.cn	wangjj@yctu.edu.cn	NOUN
fcis-5510	2	63	)	)	PUNCT
fcis-5510	2	64	abstract	abstract	NOUN
fcis-5510	2	65	:	:	PUNCT
fcis-5510	2	66	when	when	SCONJ
fcis-5510	2	67	making	make	VERB
fcis-5510	2	68	a	a	DET
fcis-5510	2	69	video	video	NOUN
fcis-5510	2	70	,	,	PUNCT
fcis-5510	2	71	if	if	SCONJ
fcis-5510	2	72	the	the	DET
fcis-5510	2	73	video	video	NOUN
fcis-5510	2	74	has	have	VERB
fcis-5510	2	75	a	a	DET
fcis-5510	2	76	language	language	NOUN
fcis-5510	2	77	organization	organization	NOUN
fcis-5510	2	78	error	error	NOUN
fcis-5510	2	79	,	,	PUNCT
fcis-5510	2	80	it	it	PRON
fcis-5510	2	81	needs	need	VERB
fcis-5510	2	82	to	to	PART
fcis-5510	2	83	be	be	AUX
fcis-5510	2	84	re	re	VERB
fcis-5510	2	85	-	-	VERB
fcis-5510	2	86	recorded	recorded	ADJ
fcis-5510	2	87	.	.	PUNCT
fcis-5510	3	1	it	it	PRON
fcis-5510	3	2	is	be	AUX
fcis-5510	3	3	not	not	PART
fcis-5510	3	4	possible	possible	ADJ
fcis-5510	3	5	to	to	PART
fcis-5510	3	6	remove	remove	VERB
fcis-5510	3	7	inappropriate	inappropriate	ADJ
fcis-5510	3	8	or	or	CCONJ
fcis-5510	3	9	unnatural	unnatural	ADJ
fcis-5510	3	10	pronunciation	pronunciation	NOUN
fcis-5510	3	11	parts	part	NOUN
fcis-5510	3	12	of	of	ADP
fcis-5510	3	13	the	the	DET
fcis-5510	3	14	recording	recording	NOUN
fcis-5510	3	15	more	more	ADV
fcis-5510	3	16	effectively	effectively	ADV
fcis-5510	3	17	.	.	PUNCT
fcis-5510	4	1	in	in	ADP
fcis-5510	4	2	response	response	NOUN
fcis-5510	4	3	to	to	ADP
fcis-5510	4	4	this	this	DET
fcis-5510	4	5	problem	problem	NOUN
fcis-5510	4	6	,	,	PUNCT
fcis-5510	4	7	this	this	DET
fcis-5510	4	8	paper	paper	NOUN
fcis-5510	4	9	studies	study	VERB
fcis-5510	4	10	the	the	DET
fcis-5510	4	11	speech	speech	NOUN
fcis-5510	4	12	extraction	extraction	NOUN
fcis-5510	4	13	,	,	PUNCT
fcis-5510	4	14	error	error	NOUN
fcis-5510	4	15	correction	correction	NOUN
fcis-5510	4	16	and	and	CCONJ
fcis-5510	4	17	synthesis	synthesis	NOUN
fcis-5510	4	18	of	of	ADP
fcis-5510	4	19	video	video	NOUN
fcis-5510	4	20	,	,	PUNCT
fcis-5510	4	21	which	which	PRON
fcis-5510	4	22	is	be	AUX
fcis-5510	4	23	divided	divide	VERB
fcis-5510	4	24	into	into	ADP
fcis-5510	4	25	three	three	NUM
fcis-5510	4	26	parts	part	NOUN
fcis-5510	4	27	:	:	PUNCT
fcis-5510	4	28	(	(	PUNCT
fcis-5510	4	29	1	1	X
fcis-5510	4	30	)	)	PUNCT
fcis-5510	4	31	speech	speech	NOUN
fcis-5510	4	32	segmentation	segmentation	NOUN
fcis-5510	4	33	and	and	CCONJ
fcis-5510	4	34	speech	speech	NOUN
fcis-5510	4	35	-	-	PUNCT
fcis-5510	4	36	to	to	ADP
fcis-5510	4	37	-	-	PUNCT
fcis-5510	4	38	text	text	NOUN
fcis-5510	4	39	of	of	ADP
fcis-5510	4	40	video	video	NOUN
fcis-5510	4	41	;	;	PUNCT
fcis-5510	4	42	(	(	PUNCT
fcis-5510	4	43	2	2	X
fcis-5510	4	44	)	)	PUNCT
fcis-5510	4	45	text	text	NOUN
fcis-5510	4	46	recognition	recognition	NOUN
fcis-5510	4	47	error	error	NOUN
fcis-5510	4	48	correction	correction	NOUN
fcis-5510	4	49	;	;	PUNCT
fcis-5510	4	50	(	(	PUNCT
fcis-5510	4	51	3	3	X
fcis-5510	4	52	)	)	PUNCT
fcis-5510	4	53	text	text	NOUN
fcis-5510	4	54	-	-	PUNCT
fcis-5510	4	55	to	to	ADP
fcis-5510	4	56	-	-	PUNCT
fcis-5510	4	57	speech	speech	NOUN
fcis-5510	4	58	and	and	CCONJ
fcis-5510	4	59	video	video	NOUN
fcis-5510	4	60	speech	speech	NOUN
fcis-5510	4	61	synthesis	synthesis	NOUN
fcis-5510	4	62	.	.	PUNCT
fcis-5510	5	1	for	for	ADP
fcis-5510	5	2	the	the	DET
fcis-5510	5	3	first	first	ADJ
fcis-5510	5	4	part	part	NOUN
fcis-5510	5	5	,	,	PUNCT
fcis-5510	5	6	we	we	PRON
fcis-5510	5	7	applied	apply	VERB
fcis-5510	5	8	the	the	DET
fcis-5510	5	9	staged	stage	VERB
fcis-5510	5	10	and	and	CCONJ
fcis-5510	5	11	efficient	efficient	ADJ
fcis-5510	5	12	algorithm	algorithm	NOUN
fcis-5510	5	13	based	base	VERB
fcis-5510	5	14	on	on	ADP
fcis-5510	5	15	(	(	PUNCT
fcis-5510	5	16	bayesian	bayesian	NOUN
fcis-5510	5	17	information	information	NOUN
fcis-5510	5	18	criterion	criterion	NOUN
fcis-5510	5	19	)	)	PUNCT
fcis-5510	5	20	bic	bic	PROPN
fcis-5510	5	21	&	&	CCONJ
fcis-5510	5	22	(	(	PUNCT
fcis-5510	5	23	statistical	statistical	ADJ
fcis-5510	5	24	mean	mean	NOUN
fcis-5510	5	25	euclidean	euclidean	ADJ
fcis-5510	5	26	distance	distance	NOUN
fcis-5510	5	27	)	)	PUNCT
fcis-5510	5	28	medist	medist	NOUN
fcis-5510	5	29	to	to	PART
fcis-5510	5	30	segment	segment	VERB
fcis-5510	5	31	the	the	DET
fcis-5510	5	32	video	video	NOUN
fcis-5510	5	33	voice	voice	NOUN
fcis-5510	5	34	,	,	PUNCT
fcis-5510	5	35	and	and	CCONJ
fcis-5510	5	36	then	then	ADV
fcis-5510	5	37	,	,	PUNCT
fcis-5510	5	38	the	the	DET
fcis-5510	5	39	segmented	segment	VERB
fcis-5510	5	40	audio	audio	NOUN
fcis-5510	5	41	is	be	AUX
fcis-5510	5	42	subtracted	subtract	VERB
fcis-5510	5	43	to	to	PART
fcis-5510	5	44	reduce	reduce	VERB
fcis-5510	5	45	noise	noise	NOUN
fcis-5510	5	46	,	,	PUNCT
fcis-5510	5	47	and	and	CCONJ
fcis-5510	5	48	finally	finally	ADV
fcis-5510	5	49	converted	convert	VERB
fcis-5510	5	50	to	to	ADP
fcis-5510	5	51	text	text	NOUN
fcis-5510	5	52	using	use	VERB
fcis-5510	5	53	the	the	DET
fcis-5510	5	54	iflytek	iflytek	NOUN
fcis-5510	5	55	interface	interface	NOUN
fcis-5510	5	56	.	.	PUNCT
fcis-5510	6	1	for	for	ADP
fcis-5510	6	2	the	the	DET
fcis-5510	6	3	second	second	ADJ
fcis-5510	6	4	part	part	NOUN
fcis-5510	6	5	,	,	PUNCT
fcis-5510	6	6	we	we	PRON
fcis-5510	6	7	apply	apply	VERB
fcis-5510	6	8	the	the	DET
fcis-5510	6	9	(	(	PUNCT
fcis-5510	6	10	double	double	ADJ
fcis-5510	6	11	automatic	automatic	ADJ
fcis-5510	6	12	error	error	NOUN
fcis-5510	6	13	correction	correction	NOUN
fcis-5510	6	14	)	)	PUNCT
fcis-5510	6	15	daec	daec	ADJ
fcis-5510	6	16	algorithm	algorithm	NOUN
fcis-5510	6	17	to	to	ADP
fcis-5510	6	18	text	text	NOUN
fcis-5510	6	19	error	error	NOUN
fcis-5510	6	20	correction	correction	NOUN
fcis-5510	6	21	.	.	PUNCT
fcis-5510	7	1	for	for	ADP
fcis-5510	7	2	the	the	DET
fcis-5510	7	3	third	third	ADJ
fcis-5510	7	4	part	part	NOUN
fcis-5510	7	5	,	,	PUNCT
fcis-5510	7	6	we	we	PRON
fcis-5510	7	7	use	use	VERB
fcis-5510	7	8	the	the	DET
fcis-5510	7	9	(	(	PUNCT
fcis-5510	7	10	improved	improve	VERB
fcis-5510	7	11	chinese	chinese	ADJ
fcis-5510	7	12	realtime	realtime	ADJ
fcis-5510	7	13	voice	voice	NOUN
fcis-5510	7	14	cloning	cloning	NOUN
fcis-5510	7	15	)	)	PUNCT
fcis-5510	8	1	i	i	PROPN
fcis-5510	8	2	-	-	PUNCT
fcis-5510	8	3	zhrtvc	zhrtvc	PROPN
fcis-5510	8	4	for	for	ADP
fcis-5510	8	5	text	text	NOUN
fcis-5510	8	6	-	-	PUNCT
fcis-5510	8	7	to	to	ADP
fcis-5510	8	8	-	-	PUNCT
fcis-5510	8	9	speech	speech	NOUN
fcis-5510	8	10	.	.	PUNCT
fcis-5510	9	1	then	then	ADV
fcis-5510	9	2	merge	merge	VERB
fcis-5510	9	3	the	the	DET
fcis-5510	9	4	voice	voice	NOUN
fcis-5510	9	5	into	into	ADP
fcis-5510	9	6	the	the	DET
fcis-5510	9	7	video	video	NOUN
fcis-5510	9	8	.	.	PUNCT
fcis-5510	10	1	the	the	DET
fcis-5510	10	2	simulation	simulation	NOUN
fcis-5510	10	3	result	result	NOUN
fcis-5510	10	4	shows	show	VERB
fcis-5510	10	5	that	that	SCONJ
fcis-5510	10	6	the	the	DET
fcis-5510	10	7	staged	stage	VERB
fcis-5510	10	8	and	and	CCONJ
fcis-5510	10	9	efficient	efficient	ADJ
fcis-5510	10	10	algorithm	algorithm	NOUN
fcis-5510	10	11	based	base	VERB
fcis-5510	10	12	on	on	ADP
fcis-5510	10	13	bic	bic	PROPN
fcis-5510	10	14	&	&	CCONJ
fcis-5510	10	15	medist	medist	PROPN
fcis-5510	10	16	,	,	PUNCT
fcis-5510	10	17	which	which	PRON
fcis-5510	10	18	accurately	accurately	ADV
fcis-5510	10	19	segmented	segment	VERB
fcis-5510	10	20	by	by	ADP
fcis-5510	10	21	sentences	sentence	NOUN
fcis-5510	10	22	,	,	PUNCT
fcis-5510	10	23	can	can	AUX
fcis-5510	10	24	identify	identify	VERB
fcis-5510	10	25	audio	audio	NOUN
fcis-5510	10	26	with	with	ADP
fcis-5510	10	27	dialect	dialect	NOUN
fcis-5510	10	28	accents	accent	NOUN
fcis-5510	10	29	,	,	PUNCT
fcis-5510	10	30	and	and	CCONJ
fcis-5510	10	31	has	have	VERB
fcis-5510	10	32	high	high	ADJ
fcis-5510	10	33	accuracy	accuracy	NOUN
fcis-5510	10	34	in	in	ADP
fcis-5510	10	35	translating	translate	VERB
fcis-5510	10	36	to	to	ADP
fcis-5510	10	37	text	text	NOUN
fcis-5510	10	38	,	,	PUNCT
fcis-5510	10	39	up	up	ADP
fcis-5510	10	40	to	to	ADP
fcis-5510	10	41	an	an	DET
fcis-5510	10	42	average	average	NOUN
fcis-5510	10	43	of	of	ADP
fcis-5510	10	44	95.8	95.8	NUM
fcis-5510	10	45	%	%	NOUN
fcis-5510	10	46	.	.	PUNCT
fcis-5510	11	1	daec	daec	ADJ
fcis-5510	11	2	algorithm	algorithm	NOUN
fcis-5510	11	3	has	have	VERB
fcis-5510	11	4	a	a	DET
fcis-5510	11	5	high	high	ADJ
fcis-5510	11	6	error	error	NOUN
fcis-5510	11	7	correction	correction	NOUN
fcis-5510	11	8	rate	rate	NOUN
fcis-5510	11	9	.	.	PUNCT
fcis-5510	12	1	the	the	DET
fcis-5510	12	2	audio	audio	ADJ
fcis-5510	12	3	prosody	prosody	NOUN
fcis-5510	12	4	accuracy	accuracy	NOUN
fcis-5510	12	5	after	after	ADP
fcis-5510	12	6	synthesis	synthesis	NOUN
fcis-5510	12	7	is	be	AUX
fcis-5510	12	8	high	high	ADJ
fcis-5510	12	9	.	.	PUNCT
fcis-5510	13	1	zvtow	zvtow	ADJ
fcis-5510	13	2	text	text	NOUN
fcis-5510	13	3	-	-	PUNCT
fcis-5510	13	4	to	to	ADP
fcis-5510	13	5	-	-	PUNCT
fcis-5510	13	6	speech	speech	NOUN
fcis-5510	13	7	(	(	PUNCT
fcis-5510	13	8	mean	mean	VERB
fcis-5510	13	9	opinion	opinion	NOUN
fcis-5510	13	10	score	score	NOUN
fcis-5510	13	11	)	)	PUNCT
fcis-5510	13	12	mos	mos	NOUN
fcis-5510	13	13	up	up	ADP
fcis-5510	13	14	to	to	ADP
fcis-5510	13	15	4.5	4.5	NUM
fcis-5510	13	16	.	.	PUNCT
fcis-5510	14	1	keywords	keyword	NOUN
fcis-5510	14	2	:	:	PUNCT
fcis-5510	14	3	speech	speech	NOUN
fcis-5510	14	4	segmentation	segmentation	NOUN
fcis-5510	14	5	;	;	PUNCT
fcis-5510	14	6	text	text	NOUN
fcis-5510	14	7	recognition	recognition	NOUN
fcis-5510	14	8	error	error	NOUN
fcis-5510	14	9	correction	correction	NOUN
fcis-5510	14	10	;	;	PUNCT
fcis-5510	14	11	text	text	NOUN
fcis-5510	14	12	-	-	PUNCT
fcis-5510	14	13	to	to	ADP
fcis-5510	14	14	-	-	PUNCT
fcis-5510	14	15	speech	speech	NOUN
fcis-5510	14	16	;	;	PUNCT
fcis-5510	14	17	video	video	NOUN
fcis-5510	14	18	speech	speech	NOUN
fcis-5510	14	19	synthesis	synthesis	NOUN
fcis-5510	14	20	.	.	PUNCT
fcis-5510	15	1	1	1	X
fcis-5510	15	2	.	.	X
fcis-5510	15	3	introduction	introduction	NOUN
fcis-5510	15	4	and	and	CCONJ
fcis-5510	15	5	related	related	ADJ
fcis-5510	15	6	work	work	NOUN
fcis-5510	15	7	at	at	ADP
fcis-5510	15	8	present	present	ADJ
fcis-5510	15	9	,	,	PUNCT
fcis-5510	15	10	the	the	DET
fcis-5510	15	11	courseware	courseware	NOUN
fcis-5510	15	12	content	content	NOUN
fcis-5510	15	13	and	and	CCONJ
fcis-5510	15	14	artificial	artificial	ADJ
fcis-5510	15	15	voice	voice	NOUN
fcis-5510	15	16	in	in	ADP
fcis-5510	15	17	the	the	DET
fcis-5510	15	18	online	online	ADJ
fcis-5510	15	19	recording	recording	NOUN
fcis-5510	15	20	of	of	ADP
fcis-5510	15	21	educational	educational	ADJ
fcis-5510	15	22	videos	video	NOUN
fcis-5510	15	23	often	often	ADV
fcis-5510	15	24	appear	appear	VERB
fcis-5510	15	25	as	as	ADP
fcis-5510	15	26	a	a	DET
fcis-5510	15	27	whole	whole	NOUN
fcis-5510	15	28	,	,	PUNCT
fcis-5510	15	29	and	and	CCONJ
fcis-5510	15	30	when	when	SCONJ
fcis-5510	15	31	the	the	DET
fcis-5510	15	32	video	video	NOUN
fcis-5510	15	33	needs	need	VERB
fcis-5510	15	34	to	to	PART
fcis-5510	15	35	be	be	AUX
fcis-5510	15	36	processed	process	VERB
fcis-5510	15	37	,	,	PUNCT
fcis-5510	15	38	more	more	ADV
fcis-5510	15	39	professional	professional	ADJ
fcis-5510	15	40	multimedia	multimedia	NOUN
fcis-5510	15	41	tools	tool	NOUN
fcis-5510	15	42	and	and	CCONJ
fcis-5510	15	43	teams	team	NOUN
fcis-5510	15	44	are	be	AUX
fcis-5510	15	45	often	often	ADV
fcis-5510	15	46	required	require	VERB
fcis-5510	15	47	to	to	PART
fcis-5510	15	48	complete	complete	VERB
fcis-5510	15	49	the	the	DET
fcis-5510	15	50	processing	processing	NOUN
fcis-5510	15	51	.	.	PUNCT
fcis-5510	16	1	what	what	PRON
fcis-5510	16	2	knowledge	knowledge	NOUN
fcis-5510	16	3	points	point	VERB
fcis-5510	16	4	need	need	VERB
fcis-5510	16	5	to	to	PART
fcis-5510	16	6	be	be	AUX
fcis-5510	16	7	explained	explain	VERB
fcis-5510	16	8	for	for	ADP
fcis-5510	16	9	how	how	SCONJ
fcis-5510	16	10	long	long	ADJ
fcis-5510	16	11	,	,	PUNCT
fcis-5510	16	12	how	how	SCONJ
fcis-5510	16	13	long	long	ADV
fcis-5510	16	14	do	do	AUX
fcis-5510	16	15	you	you	PRON
fcis-5510	16	16	need	need	VERB
fcis-5510	16	17	to	to	PART
fcis-5510	16	18	code	code	VERB
fcis-5510	16	19	demonstration	demonstration	NOUN
fcis-5510	16	20	cases	case	NOUN
fcis-5510	16	21	,	,	PUNCT
fcis-5510	16	22	whether	whether	SCONJ
fcis-5510	16	23	the	the	DET
fcis-5510	16	24	language	language	NOUN
fcis-5510	16	25	organization	organization	NOUN
fcis-5510	16	26	is	be	AUX
fcis-5510	16	27	reasonable	reasonable	ADJ
fcis-5510	16	28	,	,	PUNCT
fcis-5510	16	29	there	there	PRON
fcis-5510	16	30	is	be	VERB
fcis-5510	16	31	no	no	DET
fcis-5510	16	32	more	more	ADV
fcis-5510	16	33	effective	effective	ADJ
fcis-5510	16	34	tool	tool	NOUN
fcis-5510	16	35	to	to	PART
fcis-5510	16	36	assist	assist	VERB
fcis-5510	16	37	the	the	DET
fcis-5510	16	38	analysis	analysis	NOUN
fcis-5510	16	39	,	,	PUNCT
fcis-5510	16	40	and	and	CCONJ
fcis-5510	16	41	often	often	ADV
fcis-5510	16	42	the	the	DET
fcis-5510	16	43	recorder	recorder	NOUN
fcis-5510	16	44	often	often	ADV
fcis-5510	16	45	re	re	VERB
fcis-5510	16	46	-	-	NOUN
fcis-5510	16	47	records	record	NOUN
fcis-5510	16	48	after	after	ADP
fcis-5510	16	49	the	the	DET
fcis-5510	16	50	language	language	NOUN
fcis-5510	16	51	organization	organization	NOUN
fcis-5510	16	52	error	error	NOUN
fcis-5510	16	53	,	,	PUNCT
fcis-5510	16	54	and	and	CCONJ
fcis-5510	16	55	can	can	AUX
fcis-5510	16	56	not	not	PART
fcis-5510	16	57	more	more	ADV
fcis-5510	16	58	effectively	effectively	ADV
fcis-5510	16	59	remove	remove	VERB
fcis-5510	16	60	the	the	DET
fcis-5510	16	61	inappropriate	inappropriate	ADJ
fcis-5510	16	62	or	or	CCONJ
fcis-5510	16	63	unnatural	unnatural	ADJ
fcis-5510	16	64	pronunciation	pronunciation	NOUN
fcis-5510	16	65	part	part	NOUN
fcis-5510	16	66	of	of	ADP
fcis-5510	16	67	the	the	DET
fcis-5510	16	68	recording	recording	NOUN
fcis-5510	16	69	.	.	PUNCT
fcis-5510	17	1	therefore	therefore	ADV
fcis-5510	17	2	,	,	PUNCT
fcis-5510	17	3	video	video	NOUN
fcis-5510	17	4	analysis	analysis	NOUN
fcis-5510	17	5	schemes	scheme	NOUN
fcis-5510	17	6	based	base	VERB
fcis-5510	17	7	on	on	ADP
fcis-5510	17	8	text	text	NOUN
fcis-5510	17	9	error	error	NOUN
fcis-5510	17	10	correction	correction	NOUN
fcis-5510	17	11	have	have	AUX
fcis-5510	17	12	become	become	VERB
fcis-5510	17	13	a	a	DET
fcis-5510	17	14	research	research	NOUN
fcis-5510	17	15	hotspot.chen	hotspot.chen	PROPN
fcis-5510	17	16	lijiang	lijiang	PROPN
fcis-5510	17	17	et	et	PROPN
fcis-5510	17	18	al	al	PROPN
fcis-5510	17	19	.	.	PUNCT
fcis-5510	18	1	[	[	X
fcis-5510	18	2	1	1	X
fcis-5510	18	3	]	]	PUNCT
fcis-5510	18	4	studied	study	VERB
fcis-5510	18	5	the	the	DET
fcis-5510	18	6	synthesis	synthesis	NOUN
fcis-5510	18	7	of	of	ADP
fcis-5510	18	8	speech	speech	NOUN
fcis-5510	18	9	with	with	ADP
fcis-5510	18	10	corresponding	correspond	VERB
fcis-5510	18	11	text	text	NOUN
fcis-5510	18	12	and	and	CCONJ
fcis-5510	18	13	egg	egg	NOUN
fcis-5510	18	14	signals	signal	NOUN
fcis-5510	18	15	,	,	PUNCT
fcis-5510	18	16	and	and	CCONJ
fcis-5510	18	17	obtained	obtain	VERB
fcis-5510	18	18	a	a	DET
fcis-5510	18	19	mel	mel	PROPN
fcis-5510	18	20	cepstral	cepstral	ADJ
fcis-5510	18	21	distortion	distortion	NOUN
fcis-5510	18	22	(	(	PUNCT
fcis-5510	18	23	mcd	mcd	NOUN
fcis-5510	18	24	)	)	PUNCT
fcis-5510	18	25	of	of	ADP
fcis-5510	18	26	5.877	5.877	NUM
fcis-5510	18	27	and	and	CCONJ
fcis-5510	18	28	an	an	DET
fcis-5510	18	29	average	average	ADJ
fcis-5510	18	30	opinion	opinion	NOUN
fcis-5510	18	31	score	score	NOUN
fcis-5510	18	32	(	(	PUNCT
fcis-5510	18	33	mos	mos	NOUN
fcis-5510	18	34	)	)	PUNCT
fcis-5510	18	35	of	of	ADP
fcis-5510	18	36	3.87	3.87	NUM
fcis-5510	18	37	based	base	VERB
fcis-5510	18	38	on	on	ADP
fcis-5510	18	39	the	the	DET
fcis-5510	18	40	improved	improved	ADJ
fcis-5510	18	41	tacotron-2	tacotron-2	NUM
fcis-5510	18	42	model	model	NOUN
fcis-5510	18	43	,	,	PUNCT
fcis-5510	18	44	and	and	CCONJ
fcis-5510	18	45	achieved	achieve	VERB
fcis-5510	18	46	an	an	DET
fcis-5510	18	47	improvement	improvement	NOUN
fcis-5510	18	48	of	of	ADP
fcis-5510	18	49	0.42	0.42	NUM
fcis-5510	18	50	with	with	ADP
fcis-5510	18	51	a	a	DET
fcis-5510	18	52	relatively	relatively	ADV
fcis-5510	18	53	small	small	ADJ
fcis-5510	18	54	model	model	NOUN
fcis-5510	18	55	size	size	NOUN
fcis-5510	18	56	.	.	PUNCT
fcis-5510	19	1	a	a	DET
fcis-5510	19	2	fine	fine	ADV
fcis-5510	19	3	-	-	PUNCT
fcis-5510	19	4	grained	grain	VERB
fcis-5510	19	5	fundamental	fundamental	ADJ
fcis-5510	19	6	frequency	frequency	NOUN
fcis-5510	19	7	modification	modification	NOUN
fcis-5510	19	8	method	method	NOUN
fcis-5510	19	9	is	be	AUX
fcis-5510	19	10	proposed	propose	VERB
fcis-5510	19	11	,	,	PUNCT
fcis-5510	19	12	which	which	PRON
fcis-5510	19	13	adjusts	adjust	VERB
fcis-5510	19	14	the	the	DET
fcis-5510	19	15	fundamental	fundamental	ADJ
fcis-5510	19	16	frequency	frequency	NOUN
fcis-5510	19	17	according	accord	VERB
fcis-5510	19	18	to	to	ADP
fcis-5510	19	19	the	the	DET
fcis-5510	19	20	egg	egg	NOUN
fcis-5510	19	21	signal	signal	NOUN
fcis-5510	19	22	.	.	PUNCT
fcis-5510	20	1	compared	compare	VERB
fcis-5510	20	2	with	with	ADP
fcis-5510	20	3	the	the	DET
fcis-5510	20	4	unmodified	unmodified	ADJ
fcis-5510	20	5	method	method	NOUN
fcis-5510	20	6	,	,	PUNCT
fcis-5510	20	7	the	the	DET
fcis-5510	20	8	mcd	mcd	NOUN
fcis-5510	20	9	is	be	AUX
fcis-5510	20	10	lower	low	ADJ
fcis-5510	20	11	at	at	ADP
fcis-5510	20	12	5.781	5.781	NUM
fcis-5510	20	13	and	and	CCONJ
fcis-5510	20	14	the	the	DET
fcis-5510	20	15	mos	mos	NOUN
fcis-5510	20	16	is	be	AUX
fcis-5510	20	17	higher	high	ADJ
fcis-5510	20	18	at	at	ADP
fcis-5510	20	19	3.94.jinsong	3.94.jinsong	NUM
fcis-5510	20	20	zhang	zhang	PROPN
fcis-5510	20	21	et	et	PROPN
fcis-5510	20	22	al	al	PROPN
fcis-5510	20	23	.	.	PUNCT
fcis-5510	21	1	[	[	X
fcis-5510	21	2	2	2	NUM
fcis-5510	21	3	]	]	PUNCT
fcis-5510	21	4	studied	study	VERB
fcis-5510	21	5	robust	robust	ADJ
fcis-5510	21	6	tone	tone	NOUN
fcis-5510	21	7	recognition	recognition	NOUN
fcis-5510	21	8	based	base	VERB
fcis-5510	21	9	on	on	ADP
fcis-5510	21	10	multi	multi	ADJ
fcis-5510	21	11	-	-	ADJ
fcis-5510	21	12	level	level	ADJ
fcis-5510	21	13	core	core	NOUN
fcis-5510	21	14	framework	framework	NOUN
fcis-5510	21	15	;	;	PUNCT
fcis-5510	21	16	the	the	DET
fcis-5510	21	17	contribution	contribution	NOUN
fcis-5510	21	18	of	of	ADP
fcis-5510	21	19	tone	tone	NOUN
fcis-5510	21	20	information	information	NOUN
fcis-5510	21	21	to	to	ADP
fcis-5510	21	22	pinyin	pinyin	NOUN
fcis-5510	21	23	-	-	PUNCT
fcis-5510	21	24	to	to	ADP
fcis-5510	21	25	-	-	PUNCT
fcis-5510	21	26	text	text	NOUN
fcis-5510	21	27	conversion	conversion	NOUN
fcis-5510	21	28	,	,	PUNCT
fcis-5510	21	29	based	base	VERB
fcis-5510	21	30	on	on	ADP
fcis-5510	21	31	mutual	mutual	ADJ
fcis-5510	21	32	information	information	NOUN
fcis-5510	21	33	to	to	PART
fcis-5510	21	34	find	find	VERB
fcis-5510	21	35	tones	tone	NOUN
fcis-5510	21	36	depends	depend	VERB
fcis-5510	21	37	on	on	ADP
fcis-5510	21	38	information	information	NOUN
fcis-5510	21	39	phonemes	phoneme	NOUN
fcis-5510	21	40	.	.	PUNCT
fcis-5510	22	1	these	these	DET
fcis-5510	22	2	studies	study	NOUN
fcis-5510	22	3	and	and	CCONJ
fcis-5510	22	4	their	their	PRON
fcis-5510	22	5	results	result	NOUN
fcis-5510	22	6	are	be	AUX
fcis-5510	22	7	applicable	applicable	ADJ
fcis-5510	22	8	and	and	CCONJ
fcis-5510	22	9	instructive	instructive	ADJ
fcis-5510	22	10	for	for	ADP
fcis-5510	22	11	the	the	DET
fcis-5510	22	12	development	development	NOUN
fcis-5510	22	13	of	of	ADP
fcis-5510	22	14	tone	tone	NOUN
fcis-5510	22	15	processing	processing	NOUN
fcis-5510	22	16	techniques	technique	NOUN
fcis-5510	22	17	in	in	ADP
fcis-5510	22	18	chinese	chinese	ADJ
fcis-5510	22	19	capt	capt	PROPN
fcis-5510	22	20	techniques	technique	NOUN
fcis-5510	22	21	.	.	PUNCT
fcis-5510	23	1	a.	a.	PROPN
fcis-5510	23	2	mouchtaris	mouchtaris	PROPN
fcis-5510	23	3	et	et	PROPN
fcis-5510	23	4	al	al	PROPN
fcis-5510	23	5	.	.	PUNCT
fcis-5510	24	1	[	[	X
fcis-5510	24	2	3	3	X
fcis-5510	24	3	]	]	PUNCT
fcis-5510	24	4	address	address	NOUN
fcis-5510	24	5	the	the	DET
fcis-5510	24	6	more	more	ADV
fcis-5510	24	7	general	general	ADJ
fcis-5510	24	8	problem	problem	NOUN
fcis-5510	24	9	of	of	ADP
fcis-5510	24	10	multichannel	multichannel	ADJ
fcis-5510	24	11	audio	audio	ADJ
fcis-5510	24	12	synthesis	synthesis	NOUN
fcis-5510	24	13	by	by	ADP
fcis-5510	24	14	extending	extend	VERB
fcis-5510	24	15	the	the	DET
fcis-5510	24	16	model	model	NOUN
fcis-5510	24	17	used	use	VERB
fcis-5510	24	18	for	for	ADP
fcis-5510	24	19	the	the	DET
fcis-5510	24	20	recomposition	recomposition	NOUN
fcis-5510	24	21	problem	problem	NOUN
fcis-5510	24	22	,	,	PUNCT
fcis-5510	24	23	namely	namely	ADV
fcis-5510	24	24	how	how	SCONJ
fcis-5510	24	25	to	to	PART
fcis-5510	24	26	fully	fully	ADV
fcis-5510	24	27	synthesize	synthesize	VERB
fcis-5510	24	28	a	a	DET
fcis-5510	24	29	multichannel	multichannel	NOUN
fcis-5510	24	30	recording	recording	NOUN
fcis-5510	24	31	from	from	ADP
fcis-5510	24	32	a	a	DET
fcis-5510	24	33	particular	particular	ADJ
fcis-5510	24	34	stereo	stereo	NOUN
fcis-5510	24	35	or	or	CCONJ
fcis-5510	24	36	mono	mono	NOUN
fcis-5510	24	37	recording	recording	NOUN
fcis-5510	24	38	,	,	PUNCT
fcis-5510	24	39	by	by	ADP
fcis-5510	24	40	adapting	adapt	VERB
fcis-5510	24	41	the	the	DET
fcis-5510	24	42	recomposition	recomposition	NOUN
fcis-5510	24	43	transformation	transformation	NOUN
fcis-5510	24	44	parameters	parameter	NOUN
fcis-5510	24	45	to	to	ADP
fcis-5510	24	46	the	the	DET
fcis-5510	24	47	statistical	statistical	ADJ
fcis-5510	24	48	properties	property	NOUN
fcis-5510	24	49	of	of	ADP
fcis-5510	24	50	the	the	DET
fcis-5510	24	51	record	record	NOUN
fcis-5510	24	52	that	that	PRON
fcis-5510	24	53	we	we	PRON
fcis-5510	24	54	wish	wish	VERB
fcis-5510	24	55	to	to	PART
fcis-5510	24	56	enhance	enhance	VERB
fcis-5510	24	57	.	.	PUNCT
fcis-5510	25	1	when	when	SCONJ
fcis-5510	25	2	a	a	DET
fcis-5510	25	3	particular	particular	ADJ
fcis-5510	25	4	model	model	NOUN
fcis-5510	25	5	is	be	AUX
fcis-5510	25	6	applied	apply	VERB
fcis-5510	25	7	to	to	ADP
fcis-5510	25	8	a	a	DET
fcis-5510	25	9	different	different	ADJ
fcis-5510	25	10	context	context	NOUN
fcis-5510	25	11	(	(	PUNCT
fcis-5510	25	12	speaker	speaker	NOUN
fcis-5510	25	13	,	,	PUNCT
fcis-5510	25	14	language	language	NOUN
fcis-5510	25	15	,	,	PUNCT
fcis-5510	25	16	or	or	CCONJ
fcis-5510	25	17	channel	channel	NOUN
fcis-5510	25	18	)	)	PUNCT
fcis-5510	25	19	,	,	PUNCT
fcis-5510	25	20	this	this	DET
fcis-5510	25	21	parameter	parameter	NOUN
fcis-5510	25	22	adaptation	adaptation	NOUN
fcis-5510	25	23	is	be	AUX
fcis-5510	25	24	similar	similar	ADJ
fcis-5510	25	25	to	to	ADP
fcis-5510	25	26	task	task	NOUN
fcis-5510	25	27	adaptation	adaptation	NOUN
fcis-5510	25	28	used	use	VERB
fcis-5510	25	29	in	in	ADP
fcis-5510	25	30	speech	speech	NOUN
fcis-5510	25	31	recognition	recognition	NOUN
fcis-5510	25	32	.	.	PUNCT
fcis-5510	26	1	mahmut	mahmut	NOUN
fcis-5510	26	2	emilianerman	emilianerman	NOUN
fcis-5510	26	3	et	et	PROPN
fcis-5510	26	4	al	al	PROPN
fcis-5510	26	5	.	.	PUNCT
fcis-5510	27	1	[	[	X
fcis-5510	27	2	4	4	X
fcis-5510	27	3	]	]	PUNCT
fcis-5510	27	4	proposes	propose	VERB
fcis-5510	27	5	a	a	DET
fcis-5510	27	6	method	method	NOUN
fcis-5510	27	7	of	of	ADP
fcis-5510	27	8	audio	audio	ADJ
fcis-5510	27	9	file	file	NOUN
fcis-5510	27	10	segmentation	segmentation	NOUN
fcis-5510	27	11	that	that	PRON
fcis-5510	27	12	attempts	attempt	VERB
fcis-5510	27	13	to	to	PART
fcis-5510	27	14	alleviate	alleviate	VERB
fcis-5510	27	15	the	the	DET
fcis-5510	27	16	problem	problem	NOUN
fcis-5510	27	17	of	of	ADP
fcis-5510	27	18	different	different	ADJ
fcis-5510	27	19	individuals	individual	NOUN
fcis-5510	27	20	having	have	VERB
fcis-5510	27	21	different	different	ADJ
fcis-5510	27	22	durations	duration	NOUN
fcis-5510	27	23	for	for	ADP
fcis-5510	27	24	the	the	DET
fcis-5510	27	25	same	same	ADJ
fcis-5510	27	26	utterance	utterance	NOUN
fcis-5510	27	27	.	.	PUNCT
fcis-5510	28	1	a	a	DET
fcis-5510	28	2	method	method	NOUN
fcis-5510	28	3	for	for	ADP
fcis-5510	28	4	determining	determine	VERB
fcis-5510	28	5	the	the	DET
fcis-5510	28	6	maximum	maximum	ADJ
fcis-5510	28	7	cross	cross	ADJ
fcis-5510	28	8	-	-	ADJ
fcis-5510	28	9	correlation	correlation	ADJ
fcis-5510	28	10	value	value	NOUN
fcis-5510	28	11	between	between	ADP
fcis-5510	28	12	two	two	NUM
fcis-5510	28	13	audio	audio	ADJ
fcis-5510	28	14	files	file	NOUN
fcis-5510	28	15	and	and	CCONJ
fcis-5510	28	16	a	a	DET
fcis-5510	28	17	subsequent	subsequent	ADJ
fcis-5510	28	18	automatic	automatic	ADJ
fcis-5510	28	19	segmentation	segmentation	NOUN
fcis-5510	28	20	method	method	NOUN
fcis-5510	28	21	are	be	AUX
fcis-5510	28	22	described	describe	VERB
fcis-5510	28	23	in	in	ADP
fcis-5510	28	24	order	order	NOUN
fcis-5510	28	25	to	to	PART
fcis-5510	28	26	extract	extract	VERB
fcis-5510	28	27	two	two	NUM
fcis-5510	28	28	valid	valid	ADJ
fcis-5510	28	29	sound	sound	NOUN
fcis-5510	28	30	samples	sample	NOUN
fcis-5510	28	31	of	of	ADP
fcis-5510	28	32	target	target	NOUN
fcis-5510	28	33	consonants	consonant	NOUN
fcis-5510	28	34	,	,	PUNCT
fcis-5510	28	35	with	with	ADP
fcis-5510	28	36	the	the	DET
fcis-5510	28	37	aim	aim	NOUN
fcis-5510	28	38	of	of	ADP
fcis-5510	28	39	preprocessing	preprocesse	VERB
fcis-5510	28	40	the	the	DET
fcis-5510	28	41	audio	audio	ADJ
fcis-5510	28	42	file	file	NOUN
fcis-5510	28	43	and	and	CCONJ
fcis-5510	28	44	feeding	feed	VERB
fcis-5510	28	45	the	the	DET
fcis-5510	28	46	evenly	evenly	ADV
fcis-5510	28	47	trimmed	trim	VERB
fcis-5510	28	48	audio	audio	ADJ
fcis-5510	28	49	samples	sample	NOUN
fcis-5510	28	50	to	to	ADP
fcis-5510	28	51	a	a	DET
fcis-5510	28	52	computerized	computerized	ADJ
fcis-5510	28	53	ssd	ssd	NOUN
fcis-5510	28	54	screening	screening	NOUN
fcis-5510	28	55	system	system	NOUN
fcis-5510	28	56	.	.	PUNCT
fcis-5510	29	1	dabbabi	dabbabi	PROPN
fcis-5510	29	2	karim	karim	PROPN
fcis-5510	29	3	et	et	PROPN
fcis-5510	29	4	al	al	PROPN
fcis-5510	29	5	.	.	PUNCT
fcis-5510	30	1	[	[	X
fcis-5510	30	2	5	5	X
fcis-5510	30	3	]	]	X
fcis-5510	30	4	propose	propose	VERB
fcis-5510	30	5	an	an	DET
fcis-5510	30	6	optimized	optimize	VERB
fcis-5510	30	7	audio	audio	NOUN
fcis-5510	30	8	classification	classification	NOUN
fcis-5510	30	9	and	and	CCONJ
fcis-5510	30	10	segmentation	segmentation	NOUN
fcis-5510	30	11	algorithm	algorithm	NOUN
fcis-5510	30	12	for	for	ADP
fcis-5510	30	13	dividing	divide	VERB
fcis-5510	30	14	stacked	stack	VERB
fcis-5510	30	15	audio	audio	ADJ
fcis-5510	30	16	streams	stream	NOUN
fcis-5510	30	17	into	into	ADP
fcis-5510	30	18	10	10	NUM
fcis-5510	30	19	main	main	ADJ
fcis-5510	30	20	audio	audio	NOUN
fcis-5510	30	21	types	type	NOUN
fcis-5510	30	22	based	base	VERB
fcis-5510	30	23	on	on	ADP
fcis-5510	30	24	their	their	PRON
fcis-5510	30	25	content	content	NOUN
fcis-5510	30	26	.	.	PUNCT
fcis-5510	31	1	they	they	PRON
fcis-5510	31	2	tested	test	VERB
fcis-5510	31	3	knn	knn	PROPN
fcis-5510	31	4	,	,	PUNCT
fcis-5510	31	5	svm	svm	NOUN
fcis-5510	31	6	and	and	CCONJ
fcis-5510	31	7	gasom	gasom	NOUN
fcis-5510	31	8	algorithms	algorithm	NOUN
fcis-5510	31	9	on	on	ADP
fcis-5510	31	10	two	two	NUM
fcis-5510	31	11	audio	audio	ADJ
fcis-5510	31	12	classification	classification	NOUN
fcis-5510	31	13	systems	system	NOUN
fcis-5510	31	14	.	.	PUNCT
fcis-5510	32	1	in	in	ADP
fcis-5510	32	2	the	the	DET
fcis-5510	32	3	first	first	ADJ
fcis-5510	32	4	system	system	NOUN
fcis-5510	32	5	using	use	VERB
fcis-5510	32	6	gasom	gasom	NOUN
fcis-5510	32	7	algorithm	algorithm	NOUN
fcis-5510	32	8	and	and	CCONJ
fcis-5510	32	9	leave	leave	VERB
fcis-5510	32	10	-	-	PUNCT
fcis-5510	32	11	one	one	NUM
fcis-5510	32	12	verification	verification	NOUN
fcis-5510	32	13	technique	technique	NOUN
fcis-5510	32	14	,	,	PUNCT
fcis-5510	32	15	the	the	DET
fcis-5510	32	16	average	average	ADJ
fcis-5510	32	17	accuracy	accuracy	NOUN
fcis-5510	32	18	of	of	ADP
fcis-5510	32	19	music	music	NOUN
fcis-5510	32	20	/	/	SYM
fcis-5510	32	21	ambient	ambient	NOUN
fcis-5510	32	22	sound	sound	NOUN
fcis-5510	32	23	discrimination	discrimination	NOUN
fcis-5510	32	24	reached	reach	VERB
fcis-5510	32	25	99.17	99.17	NUM
fcis-5510	32	26	%	%	NOUN
fcis-5510	32	27	.	.	PUNCT
fcis-5510	33	1	compared	compare	VERB
fcis-5510	33	2	with	with	ADP
fcis-5510	33	3	knn	knn	PROPN
fcis-5510	33	4	and	and	CCONJ
fcis-5510	33	5	svm	svm	ADJ
fcis-5510	33	6	algorithms	algorithm	NOUN
fcis-5510	33	7	,	,	PUNCT
fcis-5510	33	8	gasom	gasom	NOUN
fcis-5510	33	9	algorithm	algorithm	NOUN
fcis-5510	33	10	always	always	ADV
fcis-5510	33	11	achieves	achieve	VERB
fcis-5510	33	12	the	the	DET
fcis-5510	33	13	best	good	ADJ
fcis-5510	33	14	performance	performance	NOUN
fcis-5510	33	15	results	result	NOUN
fcis-5510	33	16	.	.	PUNCT
fcis-5510	34	1	2	2	X
fcis-5510	34	2	.	.	X
fcis-5510	34	3	video	video	NOUN
fcis-5510	34	4	analysis	analysis	NOUN
fcis-5510	34	5	2.1	2.1	NUM
fcis-5510	34	6	.	.	PUNCT
fcis-5510	34	7	speech	speech	NOUN
fcis-5510	34	8	segmentation	segmentation	NOUN
fcis-5510	34	9	and	and	CCONJ
fcis-5510	34	10	speech	speech	NOUN
fcis-5510	34	11	-	-	PUNCT
fcis-5510	34	12	to	to	ADP
fcis-5510	34	13	-	-	PUNCT
fcis-5510	34	14	text	text	NOUN
fcis-5510	34	15	of	of	ADP
fcis-5510	34	16	video	video	NOUN
fcis-5510	34	17	in	in	ADP
fcis-5510	34	18	terms	term	NOUN
fcis-5510	34	19	of	of	ADP
fcis-5510	34	20	audio	audio	ADJ
fcis-5510	34	21	segmentation	segmentation	NOUN
fcis-5510	34	22	,	,	PUNCT
fcis-5510	34	23	we	we	PRON
fcis-5510	34	24	applied	apply	VERB
fcis-5510	34	25	the	the	DET
fcis-5510	34	26	staged	stage	VERB
fcis-5510	34	27	and	and	CCONJ
fcis-5510	34	28	efficient	efficient	ADJ
fcis-5510	34	29	algorithm	algorithm	NOUN
fcis-5510	34	30	based	base	VERB
fcis-5510	34	31	on	on	ADP
fcis-5510	34	32	bic	bic	PROPN
fcis-5510	34	33	&	&	CCONJ
fcis-5510	34	34	medist	medist	PROPN
fcis-5510	34	35	to	to	PART
fcis-5510	34	36	segment	segment	VERB
fcis-5510	34	37	the	the	DET
fcis-5510	34	38	video	video	NOUN
fcis-5510	34	39	voice	voice	NOUN
fcis-5510	34	40	.	.	PUNCT
fcis-5510	35	1	firstly	firstly	ADV
fcis-5510	35	2	,	,	PUNCT
fcis-5510	35	3	the	the	DET
fcis-5510	35	4	audio	audio	NOUN
fcis-5510	35	5	separated	separate	VERB
fcis-5510	35	6	from	from	ADP
fcis-5510	35	7	the	the	DET
fcis-5510	35	8	video	video	NOUN
fcis-5510	35	9	is	be	AUX
fcis-5510	35	10	subjected	subject	VERB
fcis-5510	35	11	to	to	ADP
fcis-5510	35	12	spectral	spectral	ADJ
fcis-5510	35	13	reduction	reduction	NOUN
fcis-5510	35	14	and	and	CCONJ
fcis-5510	35	15	noise	noise	NOUN
fcis-5510	35	16	reduction	reduction	NOUN
fcis-5510	35	17	processing	processing	NOUN
fcis-5510	35	18	;	;	PUNCT
fcis-5510	35	19	then	then	ADV
fcis-5510	35	20	,	,	PUNCT
fcis-5510	35	21	the	the	DET
fcis-5510	35	22	noise	noise	NOUN
fcis-5510	35	23	-	-	PUNCT
fcis-5510	35	24	reduced	reduce	VERB
fcis-5510	35	25	speech	speech	NOUN
fcis-5510	35	26	is	be	AUX
fcis-5510	35	27	segmented	segment	VERB
fcis-5510	35	28	in	in	ADP
fcis-5510	35	29	stages	stage	NOUN
fcis-5510	35	30	;	;	PUNCT
fcis-5510	35	31	finally	finally	ADV
fcis-5510	35	32	,	,	PUNCT
fcis-5510	35	33	call	call	VERB
fcis-5510	35	34	iflytek	iflytek	PROPN
fcis-5510	35	35	interface	interface	NOUN
fcis-5510	35	36	,	,	PUNCT
fcis-5510	35	37	complete	complete	ADJ
fcis-5510	35	38	voice	voice	NOUN
fcis-5510	35	39	-	-	PUNCT
fcis-5510	35	40	to	to	ADP
fcis-5510	35	41	-	-	PUNCT
fcis-5510	35	42	text	text	NOUN
fcis-5510	35	43	.	.	PUNCT
fcis-5510	36	1	124	124	NUM
fcis-5510	36	2	figure	figure	NOUN
fcis-5510	36	3	1	1	NUM
fcis-5510	36	4	.	.	PUNCT
fcis-5510	37	1	the	the	DET
fcis-5510	37	2	process	process	NOUN
fcis-5510	37	3	of	of	ADP
fcis-5510	37	4	staged	stage	VERB
fcis-5510	37	5	efficient	efficient	ADJ
fcis-5510	37	6	algorithms	algorithm	NOUN
fcis-5510	37	7	based	base	VERB
fcis-5510	37	8	on	on	ADP
fcis-5510	37	9	bic	bic	PROPN
fcis-5510	37	10	and	and	CCONJ
fcis-5510	37	11	medist	medist	NOUN
fcis-5510	37	12	2.2	2.2	NUM
fcis-5510	37	13	.	.	PUNCT
fcis-5510	38	1	text	text	NOUN
fcis-5510	38	2	recognition	recognition	NOUN
fcis-5510	38	3	error	error	NOUN
fcis-5510	38	4	correction	correction	NOUN
fcis-5510	38	5	in	in	ADP
fcis-5510	38	6	this	this	DET
fcis-5510	38	7	part	part	NOUN
fcis-5510	38	8	,	,	PUNCT
fcis-5510	38	9	we	we	PRON
fcis-5510	38	10	use	use	VERB
fcis-5510	38	11	the	the	DET
fcis-5510	38	12	daec	daec	ADJ
fcis-5510	38	13	algorithm	algorithm	NOUN
fcis-5510	38	14	to	to	PART
fcis-5510	38	15	correct	correct	VERB
fcis-5510	38	16	the	the	DET
fcis-5510	38	17	text	text	NOUN
fcis-5510	38	18	.	.	PUNCT
fcis-5510	39	1	first	first	ADV
fcis-5510	39	2	,	,	PUNCT
fcis-5510	39	3	automatic	automatic	ADJ
fcis-5510	39	4	error	error	NOUN
fcis-5510	39	5	correction	correction	NOUN
fcis-5510	39	6	is	be	AUX
fcis-5510	39	7	performed	perform	VERB
fcis-5510	39	8	on	on	ADP
fcis-5510	39	9	the	the	DET
fcis-5510	39	10	converted	convert	VERB
fcis-5510	39	11	text	text	NOUN
fcis-5510	39	12	(	(	PUNCT
fcis-5510	39	13	manual	manual	ADJ
fcis-5510	39	14	error	error	NOUN
fcis-5510	39	15	correction	correction	NOUN
fcis-5510	39	16	is	be	AUX
fcis-5510	39	17	also	also	ADV
fcis-5510	39	18	supported	support	VERB
fcis-5510	39	19	)	)	PUNCT
fcis-5510	39	20	;	;	PUNCT
fcis-5510	39	21	then	then	ADV
fcis-5510	39	22	,	,	PUNCT
fcis-5510	39	23	the	the	DET
fcis-5510	39	24	corrected	correct	VERB
fcis-5510	39	25	text	text	NOUN
fcis-5510	39	26	is	be	AUX
fcis-5510	39	27	audited	audit	VERB
fcis-5510	39	28	,	,	PUNCT
fcis-5510	39	29	including	include	VERB
fcis-5510	39	30	hot	hot	ADJ
fcis-5510	39	31	word	word	NOUN
fcis-5510	39	32	frequency	frequency	NOUN
fcis-5510	39	33	audit	audit	NOUN
fcis-5510	39	34	,	,	PUNCT
fcis-5510	39	35	video	video	NOUN
fcis-5510	39	36	(	(	PUNCT
fcis-5510	39	37	and	and	CCONJ
fcis-5510	39	38	corresponding	correspond	VERB
fcis-5510	39	39	voice	voice	NOUN
fcis-5510	39	40	and	and	CCONJ
fcis-5510	39	41	text	text	NOUN
fcis-5510	39	42	)	)	PUNCT
fcis-5510	39	43	modification	modification	NOUN
fcis-5510	39	44	(	(	PUNCT
fcis-5510	39	45	editing	editing	NOUN
fcis-5510	39	46	)	)	PUNCT
fcis-5510	39	47	times	time	NOUN
fcis-5510	39	48	and	and	CCONJ
fcis-5510	39	49	location	location	NOUN
fcis-5510	39	50	audit	audit	NOUN
fcis-5510	39	51	,	,	PUNCT
fcis-5510	39	52	video	video	NOUN
fcis-5510	39	53	upload	upload	NOUN
fcis-5510	39	54	and	and	CCONJ
fcis-5510	39	55	video	video	NOUN
fcis-5510	39	56	editing	editing	NOUN
fcis-5510	39	57	number	number	NOUN
fcis-5510	39	58	proportion	proportion	NOUN
fcis-5510	39	59	audit	audit	NOUN
fcis-5510	39	60	,	,	PUNCT
fcis-5510	39	61	etc	etc	X
fcis-5510	39	62	.	.	X
fcis-5510	39	63	finally	finally	ADV
fcis-5510	39	64	,	,	PUNCT
fcis-5510	39	65	the	the	DET
fcis-5510	39	66	audit	audit	NOUN
fcis-5510	39	67	results	result	NOUN
fcis-5510	39	68	are	be	AUX
fcis-5510	39	69	visualized	visualize	VERB
fcis-5510	39	70	in	in	ADP
fcis-5510	39	71	the	the	DET
fcis-5510	39	72	developed	develop	VERB
fcis-5510	39	73	teaching	teaching	NOUN
fcis-5510	39	74	video	video	NOUN
fcis-5510	39	75	voice	voice	NOUN
fcis-5510	39	76	extraction	extraction	NOUN
fcis-5510	39	77	and	and	CCONJ
fcis-5510	39	78	audit	audit	NOUN
fcis-5510	39	79	system	system	NOUN
fcis-5510	39	80	.	.	PUNCT
fcis-5510	40	1	as	as	SCONJ
fcis-5510	40	2	shown	show	VERB
fcis-5510	40	3	in	in	ADP
fcis-5510	40	4	figure	figure	NOUN
fcis-5510	40	5	2	2	NUM
fcis-5510	40	6	.	.	PUNCT
fcis-5510	40	7	figure	figure	NOUN
fcis-5510	40	8	2	2	NUM
fcis-5510	40	9	.	.	PUNCT
fcis-5510	41	1	the	the	DET
fcis-5510	41	2	daec	daec	ADJ
fcis-5510	41	3	algorithm	algorithm	NOUN
fcis-5510	41	4	video	video	NOUN
fcis-5510	41	5	voice	voice	NOUN
fcis-5510	41	6	text	text	NOUN
fcis-5510	41	7	audit	audit	NOUN
fcis-5510	41	8	solution	solution	NOUN
fcis-5510	41	9	model	model	NOUN
fcis-5510	41	10	input	input	NOUN
fcis-5510	41	11	:	:	PUNCT
fcis-5510	41	12	the	the	DET
fcis-5510	41	13	input	input	NOUN
fcis-5510	41	14	embeddings	embedding	NOUN
fcis-5510	41	15	of	of	ADP
fcis-5510	41	16	the	the	DET
fcis-5510	41	17	entire	entire	ADJ
fcis-5510	41	18	model	model	NOUN
fcis-5510	41	19	framework	framework	NOUN
fcis-5510	41	20	are	be	AUX
fcis-5510	41	21	composed	compose	VERB
fcis-5510	41	22	of	of	ADP
fcis-5510	41	23	the	the	DET
fcis-5510	41	24	addition	addition	NOUN
fcis-5510	41	25	and	and	CCONJ
fcis-5510	41	26	embedding	embed	VERB
fcis-5510	41	27	of	of	ADP
fcis-5510	41	28	word	word	NOUN
fcis-5510	41	29	embedding	embed	VERB
fcis-5510	41	30	,	,	PUNCT
fcis-5510	41	31	position	position	NOUN
fcis-5510	41	32	embedding	embed	VERB
fcis-5510	41	33	,	,	PUNCT
fcis-5510	41	34	and	and	CCONJ
fcis-5510	41	35	segment	segment	NOUN
fcis-5510	41	36	embedding	embed	VERB
fcis-5510	41	37	of	of	ADP
fcis-5510	41	38	each	each	DET
fcis-5510	41	39	character	character	NOUN
fcis-5510	41	40	in	in	ADP
fcis-5510	41	41	the	the	DET
fcis-5510	41	42	text	text	NOUN
fcis-5510	41	43	sentence	sentence	NOUN
fcis-5510	41	44	.	.	PUNCT
fcis-5510	42	1	therefore	therefore	ADV
fcis-5510	42	2	,	,	PUNCT
fcis-5510	42	3	it	it	PRON
fcis-5510	42	4	can	can	AUX
fcis-5510	42	5	be	be	AUX
fcis-5510	42	6	seen	see	VERB
fcis-5510	42	7	that	that	SCONJ
fcis-5510	42	8	the	the	DET
fcis-5510	42	9	input	input	NOUN
fcis-5510	42	10	of	of	ADP
fcis-5510	42	11	the	the	DET
fcis-5510	42	12	model	model	NOUN
fcis-5510	42	13	framework	framework	NOUN
fcis-5510	42	14	of	of	ADP
fcis-5510	42	15	the	the	DET
fcis-5510	42	16	whole	whole	ADJ
fcis-5510	42	17	method	method	NOUN
fcis-5510	42	18	is	be	AUX
fcis-5510	42	19	actually	actually	ADV
fcis-5510	42	20	the	the	DET
fcis-5510	42	21	same	same	ADJ
fcis-5510	42	22	as	as	ADP
fcis-5510	42	23	the	the	DET
fcis-5510	42	24	general	general	ADJ
fcis-5510	42	25	input	input	NOUN
fcis-5510	42	26	form	form	NOUN
fcis-5510	42	27	of	of	ADP
fcis-5510	42	28	the	the	DET
fcis-5510	42	29	daec	daec	ADJ
fcis-5510	42	30	algorithm	algorithm	NOUN
fcis-5510	42	31	.	.	PUNCT
fcis-5510	43	1	𝑒𝑖	𝑒𝑖	X
fcis-5510	43	2	=	=	PROPN
fcis-5510	43	3	𝑤𝑜𝑟𝑑	𝑤𝑜𝑟𝑑	PROPN
fcis-5510	43	4	𝑒𝑚𝑏𝑒𝑑𝑑𝑖𝑛𝑔(𝑒𝑖	𝑒𝑚𝑏𝑒𝑑𝑑𝑖𝑛𝑔(𝑒𝑖	PROPN
fcis-5510	43	5	)	)	PUNCT
fcis-5510	43	6	+	+	CCONJ
fcis-5510	43	7	𝑝𝑜𝑠𝑖𝑡𝑖𝑜𝑛	𝑝𝑜𝑠𝑖𝑡𝑖𝑜𝑛	PROPN
fcis-5510	43	8	𝑒𝑚𝑏𝑒𝑑𝑑𝑖𝑛𝑔(𝑥𝑖	𝑒𝑚𝑏𝑒𝑑𝑑𝑖𝑛𝑔(𝑥𝑖	PROPN
fcis-5510	43	9	)	)	PUNCT
fcis-5510	44	1	+	+	CCONJ
fcis-5510	44	2	𝑠𝑒𝑔𝑚𝑒𝑛𝑡	𝑠𝑒𝑔𝑚𝑒𝑛𝑡	PROPN
fcis-5510	44	3	𝑒𝑚𝑏𝑒𝑑𝑑𝑖𝑛𝑔(𝑥𝑖	𝑒𝑚𝑏𝑒𝑑𝑑𝑖𝑛𝑔(𝑥𝑖	PROPN
fcis-5510	44	4	)	)	PUNCT
fcis-5510	44	5	(	(	PUNCT
fcis-5510	44	6	1	1	X
fcis-5510	44	7	)	)	PUNCT
fcis-5510	44	8	detection	detection	NOUN
fcis-5510	44	9	network	network	NOUN
fcis-5510	44	10	:	:	PUNCT
fcis-5510	44	11	the	the	DET
fcis-5510	44	12	detection	detection	NOUN
fcis-5510	44	13	network	network	NOUN
fcis-5510	44	14	in	in	ADP
fcis-5510	44	15	the	the	DET
fcis-5510	44	16	daec	daec	ADJ
fcis-5510	44	17	algorithm	algorithm	NOUN
fcis-5510	44	18	framework	framework	NOUN
fcis-5510	44	19	is	be	AUX
fcis-5510	44	20	essentially	essentially	ADV
fcis-5510	44	21	a	a	DET
fcis-5510	44	22	bidirectional	bidirectional	ADJ
fcis-5510	44	23	gru	gru	NOUN
fcis-5510	44	24	model	model	NOUN
fcis-5510	44	25	(	(	PUNCT
fcis-5510	44	26	bi	bi	NOUN
fcis-5510	44	27	-	-	NOUN
fcis-5510	44	28	gru	gru	NOUN
fcis-5510	44	29	)	)	PUNCT
fcis-5510	44	30	.	.	PUNCT
fcis-5510	45	1	the	the	DET
fcis-5510	45	2	bi	bi	PROPN
fcis-5510	45	3	-	-	PROPN
fcis-5510	45	4	gru	gru	NOUN
fcis-5510	45	5	model	model	NOUN
fcis-5510	45	6	encodes	encode	NOUN
fcis-5510	45	7	each	each	DET
fcis-5510	45	8	text	text	NOUN
fcis-5510	45	9	sequence	sequence	NOUN
fcis-5510	45	10	forward	forward	ADV
fcis-5510	45	11	and	and	CCONJ
fcis-5510	45	12	backward	backward	ADV
fcis-5510	45	13	,	,	PUNCT
fcis-5510	45	14	and	and	CCONJ
fcis-5510	45	15	then	then	ADV
fcis-5510	45	16	merges	merge	VERB
fcis-5510	45	17	the	the	DET
fcis-5510	45	18	hidden	hidden	ADJ
fcis-5510	45	19	state	state	NOUN
fcis-5510	45	20	of	of	ADP
fcis-5510	45	21	the	the	DET
fcis-5510	45	22	forward	forward	ADJ
fcis-5510	45	23	coding	coding	NOUN
fcis-5510	45	24	of	of	ADP
fcis-5510	45	25	the	the	DET
fcis-5510	45	26	last	last	ADJ
fcis-5510	45	27	hidden	hide	VERB
fcis-5510	45	28	layer	layer	NOUN
fcis-5510	45	29	chinese	chinese	PROPN
fcis-5510	45	30	the	the	DET
fcis-5510	45	31	sequence	sequence	NOUN
fcis-5510	45	32	horizontally	horizontally	ADV
fcis-5510	45	33	with	with	ADP
fcis-5510	45	34	the	the	DET
fcis-5510	45	35	hidden	hidden	ADJ
fcis-5510	45	36	state	state	NOUN
fcis-5510	45	37	of	of	ADP
fcis-5510	45	38	reverse	reverse	NOUN
fcis-5510	45	39	encoding	encoding	NOUN
fcis-5510	45	40	,	,	PUNCT
fcis-5510	45	41	and	and	CCONJ
fcis-5510	45	42	the	the	DET
fcis-5510	45	43	calculation	calculation	NOUN
fcis-5510	45	44	process	process	NOUN
fcis-5510	45	45	of	of	ADP
fcis-5510	45	46	the	the	DET
fcis-5510	45	47	bi	bi	PROPN
fcis-5510	45	48	-	-	PROPN
fcis-5510	45	49	gru	gru	NOUN
fcis-5510	45	50	model	model	NOUN
fcis-5510	45	51	is	be	AUX
fcis-5510	45	52	as	as	SCONJ
fcis-5510	45	53	follows	follow	VERB
fcis-5510	45	54	:	:	PUNCT
fcis-5510	45	55	ℎ𝑖	ℎ𝑖	PROPN
fcis-5510	45	56	𝑑⃗⃗	𝑑⃗⃗	PROPN
fcis-5510	45	57	⃗⃗	⃗⃗	PROPN
fcis-5510	45	58	=	=	SYM
fcis-5510	45	59	𝐺𝑅𝑈(ℎ𝑖−1	𝐺𝑅𝑈(ℎ𝑖−1	PROPN
fcis-5510	45	60	𝑑⃗⃗	𝑑⃗⃗	PROPN
fcis-5510	45	61	⃗⃗	⃗⃗	PROPN
fcis-5510	45	62	⃗⃗	⃗⃗	PROPN
fcis-5510	45	63	⃗⃗	⃗⃗	PROPN
fcis-5510	45	64	,	,	PUNCT
fcis-5510	45	65	𝑒𝑖	𝑒𝑖	PROPN
fcis-5510	45	66	)	)	PUNCT
fcis-5510	45	67	(	(	PUNCT
fcis-5510	45	68	2	2	X
fcis-5510	45	69	)	)	PUNCT
fcis-5510	45	70	ℎ𝑖	ℎ𝑖	NOUN
fcis-5510	45	71	𝑑⃖⃗	𝑑⃖⃗	PRON
fcis-5510	45	72	⃗⃗⃗	⃗⃗⃗	PROPN
fcis-5510	45	73	=	=	SYM
fcis-5510	45	74	gru(ℎ𝑖+1	gru(ℎ𝑖+1	PROPN
fcis-5510	45	75	𝑑⃖⃗	𝑑⃖⃗	PROPN
fcis-5510	45	76	⃗⃗⃗⃗	⃗⃗⃗⃗	PUNCT
fcis-5510	46	1	⃗⃗⃗	⃗⃗⃗	PROPN
fcis-5510	46	2	,	,	PUNCT
fcis-5510	46	3	𝑒𝑖	𝑒𝑖	PROPN
fcis-5510	46	4	)	)	PUNCT
fcis-5510	46	5	(	(	PUNCT
fcis-5510	46	6	3	3	X
fcis-5510	46	7	)	)	PUNCT
fcis-5510	46	8	ℎ𝑖	ℎ𝑖	NOUN
fcis-5510	46	9	𝑑	𝑑	NOUN
fcis-5510	46	10	=	=	PUNCT
fcis-5510	47	1	[	[	X
fcis-5510	47	2	ℎ𝑖	ℎ𝑖	ADP
fcis-5510	47	3	𝑑⃗⃗	𝑑⃗⃗	PROPN
fcis-5510	47	4	⃗⃗	⃗⃗	PROPN
fcis-5510	47	5	;	;	PUNCT
fcis-5510	47	6	ℎ𝑖	ℎ𝑖	NOUN
fcis-5510	47	7	𝑑⃖⃗	𝑑⃖⃗	PROPN
fcis-5510	47	8	⃗⃗⃗	⃗⃗⃗	NOUN
fcis-5510	47	9	]	]	PUNCT
fcis-5510	47	10	(	(	PUNCT
fcis-5510	47	11	4	4	X
fcis-5510	47	12	)	)	PUNCT
fcis-5510	47	13	ℎ𝑖	ℎ𝑖	NOUN
fcis-5510	47	14	𝑑is	𝑑is	VERB
fcis-5510	47	15	the	the	DET
fcis-5510	47	16	hidden	hidden	ADJ
fcis-5510	47	17	state	state	NOUN
fcis-5510	47	18	of	of	ADP
fcis-5510	47	19	the	the	DET
fcis-5510	47	20	embedded	embed	VERB
fcis-5510	47	21	ei	ei	NOUN
fcis-5510	47	22	of	of	ADP
fcis-5510	47	23	the	the	DET
fcis-5510	47	24	character	character	NOUN
fcis-5510	47	25	i	i	PRON
fcis-5510	47	26	in	in	ADP
fcis-5510	47	27	the	the	DET
fcis-5510	47	28	text	text	NOUN
fcis-5510	47	29	sequence	sequence	NOUN
fcis-5510	47	30	in	in	ADP
fcis-5510	47	31	the	the	DET
fcis-5510	47	32	last	last	ADJ
fcis-5510	47	33	hidden	hide	VERB
fcis-5510	47	34	layer	layer	NOUN
fcis-5510	47	35	after	after	ADP
fcis-5510	47	36	being	be	AUX
fcis-5510	47	37	calculated	calculate	VERB
fcis-5510	47	38	by	by	ADP
fcis-5510	47	39	the	the	DET
fcis-5510	47	40	bi	bi	PROPN
fcis-5510	47	41	-	-	PROPN
fcis-5510	47	42	gru	gru	NOUN
fcis-5510	47	43	model	model	NOUN
fcis-5510	47	44	?	?	PUNCT
fcis-5510	48	1	the	the	DET
fcis-5510	48	2	number	number	NOUN
fcis-5510	48	3	of	of	ADP
fcis-5510	48	4	hidden	hide	VERB
fcis-5510	48	5	layer	layer	NOUN
fcis-5510	48	6	dimensions	dimension	NOUN
fcis-5510	48	7	of	of	ADP
fcis-5510	48	8	the	the	DET
fcis-5510	48	9	reference	reference	NOUN
fcis-5510	48	10	bi	bi	PROPN
fcis-5510	48	11	-	-	ADJ
fcis-5510	48	12	gru	gru	NOUN
fcis-5510	48	13	model	model	NOUN
fcis-5510	48	14	is	be	AUX
fcis-5510	48	15	set	set	VERB
fcis-5510	48	16	to	to	ADP
fcis-5510	48	17	256	256	NUM
fcis-5510	48	18	,	,	PUNCT
fcis-5510	48	19	and	and	CCONJ
fcis-5510	48	20	the	the	DET
fcis-5510	48	21	number	number	NOUN
fcis-5510	48	22	of	of	ADP
fcis-5510	48	23	hidden	hide	VERB
fcis-5510	48	24	layer	layer	NOUN
fcis-5510	48	25	output	output	NOUN
fcis-5510	48	26	dimensions	dimension	NOUN
fcis-5510	48	27	after	after	ADP
fcis-5510	48	28	bidirectional	bidirectional	ADJ
fcis-5510	48	29	encoding	encoding	NOUN
fcis-5510	48	30	is	be	AUX
fcis-5510	48	31	512	512	NUM
fcis-5510	48	32	.	.	PUNCT
fcis-5510	49	1	the	the	DET
fcis-5510	49	2	hid	hid	NOUN
fcis-5510	49	3	calculated	calculate	VERB
fcis-5510	49	4	by	by	ADP
fcis-5510	49	5	the	the	DET
fcis-5510	49	6	bigru	bigru	NOUN
fcis-5510	49	7	model	model	NOUN
fcis-5510	49	8	is	be	AUX
fcis-5510	49	9	then	then	ADV
fcis-5510	49	10	fed	feed	VERB
fcis-5510	49	11	into	into	ADP
fcis-5510	49	12	two	two	NUM
fcis-5510	49	13	fully	fully	ADV
fcis-5510	49	14	connected	connected	ADJ
fcis-5510	49	15	layers	layer	NOUN
fcis-5510	49	16	.	.	PUNCT
fcis-5510	50	1	loss	loss	NOUN
fcis-5510	50	2	function	function	NOUN
fcis-5510	50	3	:	:	PUNCT
fcis-5510	50	4	as	as	SCONJ
fcis-5510	50	5	shown	show	VERB
fcis-5510	50	6	in	in	ADP
fcis-5510	50	7	the	the	DET
fcis-5510	50	8	following	following	ADJ
fcis-5510	50	9	schema	schema	NOUN
fcis-5510	50	10	,	,	PUNCT
fcis-5510	50	11	ℒ𝑑	ℒ𝑑	PROPN
fcis-5510	50	12	represents	represent	VERB
fcis-5510	50	13	the	the	DET
fcis-5510	50	14	cross	cross	ADJ
fcis-5510	50	15	-	-	ADJ
fcis-5510	50	16	entropy	entropy	ADJ
fcis-5510	50	17	loss	loss	NOUN
fcis-5510	50	18	calculated	calculate	VERB
fcis-5510	50	19	by	by	ADP
fcis-5510	50	20	the	the	DET
fcis-5510	50	21	final	final	ADJ
fcis-5510	50	22	output	output	NOUN
fcis-5510	50	23	value	value	NOUN
fcis-5510	50	24	of	of	ADP
fcis-5510	50	25	the	the	DET
fcis-5510	50	26	detection	detection	NOUN
fcis-5510	50	27	network	network	NOUN
fcis-5510	50	28	of	of	ADP
fcis-5510	50	29	the	the	DET
fcis-5510	50	30	error	error	NOUN
fcis-5510	50	31	detection	detection	NOUN
fcis-5510	50	32	network	network	NOUN
fcis-5510	50	33	;	;	PUNCT
fcis-5510	50	34	ℒ𝑐	ℒ𝑐	PROPN
fcis-5510	50	35	represents	represent	VERB
fcis-5510	50	36	the	the	DET
fcis-5510	50	37	cross	cross	ADJ
fcis-5510	50	38	-	-	ADJ
fcis-5510	50	39	entropy	entropy	ADJ
fcis-5510	50	40	loss	loss	NOUN
fcis-5510	50	41	calculated	calculate	VERB
fcis-5510	50	42	for	for	ADP
fcis-5510	50	43	the	the	DET
fcis-5510	50	44	output	output	NOUN
fcis-5510	50	45	value	value	NOUN
fcis-5510	50	46	after	after	ADP
fcis-5510	50	47	the	the	DET
fcis-5510	50	48	error	error	NOUN
fcis-5510	50	49	correction	correction	NOUN
fcis-5510	50	50	network	network	NOUN
fcis-5510	50	51	.	.	PUNCT
fcis-5510	51	1	ℒ𝑑	ℒ𝑑	PROPN
fcis-5510	51	2	=	=	SYM
fcis-5510	51	3	−∑log𝑃𝑑(𝑔𝑖|𝑋	−∑log𝑃𝑑(𝑔𝑖|𝑋	X
fcis-5510	51	4	)	)	PUNCT
fcis-5510	51	5	𝑛	𝑛	PRON
fcis-5510	51	6	𝑖=1	𝑖=1	PROPN
fcis-5510	51	7	(	(	PUNCT
fcis-5510	51	8	5	5	X
fcis-5510	51	9	)	)	PUNCT
fcis-5510	51	10	ℒ𝑐	ℒ𝑐	NOUN
fcis-5510	51	11	=	=	SYM
fcis-5510	51	12	−∑log𝑃𝑐(𝑦𝑖|𝑋	−∑log𝑃𝑐(𝑦𝑖|𝑋	NOUN
fcis-5510	51	13	)	)	PUNCT
fcis-5510	52	1	𝑛	𝑛	PRON
fcis-5510	52	2	𝑖=1	𝑖=1	PROPN
fcis-5510	52	3	(	(	PUNCT
fcis-5510	52	4	6	6	NUM
fcis-5510	52	5	)	)	PUNCT
fcis-5510	52	6	the	the	DET
fcis-5510	52	7	loss	loss	NOUN
fcis-5510	52	8	function	function	NOUN
fcis-5510	52	9	of	of	ADP
fcis-5510	52	10	the	the	DET
fcis-5510	52	11	whole	whole	ADJ
fcis-5510	52	12	daec	daec	ADJ
fcis-5510	52	13	algorithm	algorithm	NOUN
fcis-5510	52	14	is	be	AUX
fcis-5510	52	15	composed	compose	VERB
fcis-5510	52	16	of	of	ADP
fcis-5510	52	17	the	the	DET
fcis-5510	52	18	loss	loss	NOUN
fcis-5510	52	19	function	function	NOUN
fcis-5510	52	20	of	of	ADP
fcis-5510	52	21	the	the	DET
fcis-5510	52	22	detection	detection	NOUN
fcis-5510	52	23	network	network	NOUN
fcis-5510	52	24	of	of	ADP
fcis-5510	52	25	the	the	DET
fcis-5510	52	26	error	error	NOUN
fcis-5510	52	27	detection	detection	NOUN
fcis-5510	52	28	network	network	NOUN
fcis-5510	52	29	and	and	CCONJ
fcis-5510	52	30	the	the	DET
fcis-5510	52	31	loss	loss	NOUN
fcis-5510	52	32	function	function	NOUN
fcis-5510	52	33	of	of	ADP
fcis-5510	52	34	the	the	DET
fcis-5510	52	35	error	error	NOUN
fcis-5510	52	36	correction	correction	NOUN
fcis-5510	52	37	network	network	NOUN
fcis-5510	52	38	correction	correction	NOUN
fcis-5510	52	39	network	network	NOUN
fcis-5510	52	40	:	:	PUNCT
fcis-5510	52	41	ℒ	ℒ	NOUN
fcis-5510	52	42	=	=	SYM
fcis-5510	52	43	λ	λ	NOUN
fcis-5510	52	44	∙	∙	PROPN
fcis-5510	52	45	ℒ𝑐	ℒ𝑐	PROPN
fcis-5510	52	46	+	+	CCONJ
fcis-5510	52	47	(	(	PUNCT
fcis-5510	52	48	1	1	NUM
fcis-5510	52	49	−	−	PROPN
fcis-5510	52	50	λ	λ	NOUN
fcis-5510	52	51	)	)	PUNCT
fcis-5510	52	52	∙	∙	PROPN
fcis-5510	52	53	ℒ𝑑	ℒ𝑑	PROPN
fcis-5510	52	54	(	(	PUNCT
fcis-5510	52	55	7	7	NUM
fcis-5510	52	56	)	)	PUNCT
fcis-5510	52	57	the	the	DET
fcis-5510	52	58	above	above	ADJ
fcis-5510	52	59	equation	equation	NOUN
fcis-5510	52	60	is	be	AUX
fcis-5510	52	61	the	the	DET
fcis-5510	52	62	loss	loss	NOUN
fcis-5510	52	63	function	function	NOUN
fcis-5510	52	64	representation	representation	NOUN
fcis-5510	52	65	of	of	ADP
fcis-5510	52	66	daec	daec	ADJ
fcis-5510	52	67	algorithm	algorithm	NOUN
fcis-5510	52	68	.	.	PUNCT
fcis-5510	53	1	in	in	ADP
fcis-5510	53	2	the	the	DET
fcis-5510	53	3	formula	formula	NOUN
fcis-5510	53	4	,	,	PUNCT
fcis-5510	53	5	1−λ	1−λ	NUM
fcis-5510	53	6	and	and	CCONJ
fcis-5510	53	7	λ	λ	PROPN
fcis-5510	53	8	are	be	AUX
fcis-5510	53	9	the	the	DET
fcis-5510	53	10	linear	linear	ADJ
fcis-5510	53	11	combination	combination	NOUN
fcis-5510	53	12	coefficients	coefficient	NOUN
fcis-5510	53	13	of	of	ADP
fcis-5510	53	14	the	the	DET
fcis-5510	53	15	loss	loss	NOUN
fcis-5510	53	16	function	function	NOUN
fcis-5510	53	17	of	of	ADP
fcis-5510	53	18	the	the	DET
fcis-5510	53	19	detection	detection	NOUN
fcis-5510	53	20	network	network	NOUN
fcis-5510	53	21	and	and	CCONJ
fcis-5510	53	22	the	the	DET
fcis-5510	53	23	loss	loss	NOUN
fcis-5510	53	24	function	function	NOUN
fcis-5510	53	25	of	of	ADP
fcis-5510	53	26	the	the	DET
fcis-5510	53	27	correction	correction	NOUN
fcis-5510	53	28	network	network	NOUN
fcis-5510	53	29	,	,	PUNCT
fcis-5510	53	30	that	that	ADV
fcis-5510	53	31	is	is	ADV
fcis-5510	53	32	,	,	PUNCT
fcis-5510	53	33	the	the	DET
fcis-5510	53	34	linear	linear	ADJ
fcis-5510	53	35	combination	combination	NOUN
fcis-5510	53	36	of	of	ADP
fcis-5510	53	37	the	the	DET
fcis-5510	53	38	respective	respective	ADJ
fcis-5510	53	39	loss	loss	NOUN
fcis-5510	53	40	functions	function	NOUN
fcis-5510	53	41	of	of	ADP
fcis-5510	53	42	the	the	DET
fcis-5510	53	43	two	two	NUM
fcis-5510	53	44	networks	network	NOUN
fcis-5510	53	45	is	be	AUX
fcis-5510	53	46	the	the	DET
fcis-5510	53	47	total	total	ADJ
fcis-5510	53	48	loss	loss	NOUN
fcis-5510	53	49	function	function	NOUN
fcis-5510	53	50	of	of	ADP
fcis-5510	53	51	the	the	DET
fcis-5510	53	52	final	final	ADJ
fcis-5510	53	53	softmasked	softmasked	ADJ
fcis-5510	53	54	bert	bert	PROPN
fcis-5510	53	55	model	model	NOUN
fcis-5510	53	56	.	.	PUNCT
fcis-5510	54	1	2.3	2.3	NUM
fcis-5510	54	2	.	.	PUNCT
fcis-5510	55	1	text	text	NOUN
fcis-5510	55	2	-	-	PUNCT
fcis-5510	55	3	to	to	ADP
fcis-5510	55	4	-	-	PUNCT
fcis-5510	55	5	speech	speech	NOUN
fcis-5510	55	6	and	and	CCONJ
fcis-5510	55	7	video	video	NOUN
fcis-5510	55	8	speech	speech	NOUN
fcis-5510	55	9	synthesis	synthesis	NOUN
fcis-5510	55	10	in	in	ADP
fcis-5510	55	11	this	this	DET
fcis-5510	55	12	part	part	NOUN
fcis-5510	55	13	,	,	PUNCT
fcis-5510	55	14	we	we	PRON
fcis-5510	55	15	use	use	VERB
fcis-5510	55	16	the	the	DET
fcis-5510	55	17	i	i	PROPN
fcis-5510	55	18	-	-	NOUN
fcis-5510	55	19	zhrtvc	zhrtvc	PROPN
fcis-5510	55	20	for	for	ADP
fcis-5510	55	21	text	text	NOUN
fcis-5510	55	22	-	-	PUNCT
fcis-5510	55	23	to	to	ADP
fcis-5510	55	24	-	-	PUNCT
fcis-5510	55	25	speech	speech	NOUN
fcis-5510	55	26	.	.	PUNCT
fcis-5510	56	1	then	then	ADV
fcis-5510	56	2	merge	merge	VERB
fcis-5510	56	3	the	the	DET
fcis-5510	56	4	voice	voice	NOUN
fcis-5510	56	5	into	into	ADP
fcis-5510	56	6	the	the	DET
fcis-5510	56	7	video	video	NOUN
fcis-5510	56	8	.	.	PUNCT
fcis-5510	57	1	the	the	DET
fcis-5510	57	2	prototype	prototype	NOUN
fcis-5510	57	3	is	be	AUX
fcis-5510	57	4	zhrtvc	zhrtvc	PROPN
fcis-5510	57	5	,	,	PUNCT
fcis-5510	57	6	using	use	VERB
fcis-5510	57	7	machine	machine	NOUN
fcis-5510	57	8	learning	learning	NOUN
fcis-5510	57	9	algorithms	algorithm	NOUN
fcis-5510	57	10	,	,	PUNCT
fcis-5510	57	11	zhvoice	zhvoice	PROPN
fcis-5510	57	12	is	be	AUX
fcis-5510	57	13	the	the	DET
fcis-5510	57	14	chinese	chinese	ADJ
fcis-5510	57	15	speech	speech	NOUN
fcis-5510	57	16	corpus	corpus	NOUN
fcis-5510	57	17	,	,	PUNCT
fcis-5510	57	18	the	the	DET
fcis-5510	57	19	voice	voice	NOUN
fcis-5510	57	20	is	be	AUX
fcis-5510	57	21	clearer	clear	ADJ
fcis-5510	57	22	and	and	CCONJ
fcis-5510	57	23	more	more	ADV
fcis-5510	57	24	natural	natural	ADJ
fcis-5510	57	25	,	,	PUNCT
fcis-5510	57	26	including	include	VERB
fcis-5510	57	27	8	8	NUM
fcis-5510	57	28	open	open	ADJ
fcis-5510	57	29	source	source	NOUN
fcis-5510	57	30	datasets	dataset	NOUN
fcis-5510	57	31	,	,	PUNCT
fcis-5510	57	32	3200	3200	NUM
fcis-5510	57	33	speakers	speaker	NOUN
fcis-5510	57	34	,	,	PUNCT
fcis-5510	57	35	900	900	NUM
fcis-5510	57	36	hours	hour	NOUN
fcis-5510	57	37	of	of	ADP
fcis-5510	57	38	speech	speech	NOUN
fcis-5510	57	39	,	,	PUNCT
fcis-5510	57	40	13	13	NUM
fcis-5510	57	41	million	million	NUM
fcis-5510	57	42	words	word	NOUN
fcis-5510	57	43	.	.	PUNCT
fcis-5510	58	1	large	large	ADJ
fcis-5510	58	2	corpus	corpus	NOUN
fcis-5510	58	3	splicing	splicing	NOUN
fcis-5510	58	4	speech	speech	NOUN
fcis-5510	58	5	synthesis	synthesis	NOUN
fcis-5510	58	6	system	system	NOUN
fcis-5510	58	7	based	base	VERB
fcis-5510	58	8	on	on	ADP
fcis-5510	58	9	statistical	statistical	ADJ
fcis-5510	58	10	rules	rule	NOUN
fcis-5510	58	11	ultra	ultra	ADJ
fcis-5510	58	12	-	-	ADJ
fcis-5510	58	13	largescale	largescale	ADJ
fcis-5510	58	14	sound	sound	NOUN
fcis-5510	58	15	library	library	NOUN
fcis-5510	58	16	production	production	NOUN
fcis-5510	58	17	:	:	PUNCT
fcis-5510	58	18	corpus	corpus	NOUN
fcis-5510	58	19	design	design	NOUN
fcis-5510	58	20	;	;	PUNCT
fcis-5510	58	21	soundbank	soundbank	PROPN
fcis-5510	58	22	recording	recording	NOUN
fcis-5510	58	23	;	;	PUNCT
fcis-5510	58	24	fine	fine	ADJ
fcis-5510	58	25	segmentation	segmentation	NOUN
fcis-5510	58	26	;	;	PUNCT
fcis-5510	58	27	prosodic	prosodic	ADJ
fcis-5510	58	28	callouts	callout	NOUN
fcis-5510	58	29	;	;	PUNCT
fcis-5510	58	30	the	the	DET
fcis-5510	58	31	sound	sound	ADJ
fcis-5510	58	32	quality	quality	NOUN
fcis-5510	58	33	is	be	AUX
fcis-5510	58	34	good	good	ADJ
fcis-5510	58	35	,	,	PUNCT
fcis-5510	58	36	the	the	DET
fcis-5510	58	37	difference	difference	NOUN
fcis-5510	58	38	in	in	ADP
fcis-5510	58	39	sound	sound	ADJ
fcis-5510	58	40	quality	quality	NOUN
fcis-5510	58	41	of	of	ADP
fcis-5510	58	42	recording	record	VERB
fcis-5510	58	43	synthesis	synthesis	NOUN
fcis-5510	58	44	is	be	AUX
fcis-5510	58	45	small	small	ADJ
fcis-5510	58	46	,	,	PUNCT
fcis-5510	58	47	and	and	CCONJ
fcis-5510	58	48	the	the	DET
fcis-5510	58	49	naturalness	naturalness	NOUN
fcis-5510	58	50	of	of	ADP
fcis-5510	58	51	normal	normal	ADJ
fcis-5510	58	52	sentences	sentence	NOUN
fcis-5510	58	53	is	be	AUX
fcis-5510	58	54	good	good	ADJ
fcis-5510	58	55	.	.	PUNCT
fcis-5510	59	1	the	the	DET
fcis-5510	59	2	zhvoice	zhvoice	NOUN
fcis-5510	59	3	corpus	corpus	NOUN
fcis-5510	59	4	can	can	AUX
fcis-5510	59	5	be	be	AUX
fcis-5510	59	6	used	use	VERB
fcis-5510	59	7	to	to	PART
fcis-5510	59	8	train	train	VERB
fcis-5510	59	9	the	the	DET
fcis-5510	59	10	base	base	NOUN
fcis-5510	59	11	model	model	NOUN
fcis-5510	59	12	of	of	ADP
fcis-5510	59	13	speech	speech	NOUN
fcis-5510	59	14	cloning	cloning	NOUN
fcis-5510	59	15	.	.	PUNCT
fcis-5510	60	1	ge2e_pretrained_iwater.pt	ge2e_pretrained_iwater.pt	NOUN
fcis-5510	60	2	:	:	PUNCT
fcis-5510	60	3	a	a	DET
fcis-5510	60	4	speech	speech	NOUN
fcis-5510	60	5	encoder	encoder	NOUN
fcis-5510	60	6	model	model	NOUN
fcis-5510	60	7	trained	train	VERB
fcis-5510	60	8	with	with	ADP
fcis-5510	60	9	chinese	chinese	ADJ
fcis-5510	60	10	open	open	ADJ
fcis-5510	60	11	-	-	PUNCT
fcis-5510	60	12	source	source	NOUN
fcis-5510	60	13	speech	speech	NOUN
fcis-5510	60	14	corpus	corpus	NOUN
fcis-5510	60	15	.	.	PUNCT
fcis-5510	61	1	mel	mel	PROPN
fcis-5510	61	2	-	-	PUNCT
fcis-5510	61	3	spectrogram	spectrogram	NOUN
fcis-5510	61	4	:	:	PUNCT
fcis-5510	61	5	vocoder	vocoder	VERB
fcis-5510	61	6	the	the	DET
fcis-5510	61	7	process	process	NOUN
fcis-5510	61	8	of	of	ADP
fcis-5510	61	9	i	i	PROPN
fcis-5510	61	10	-	-	PUNCT
fcis-5510	61	11	zhrtvc	zhrtvc	PROPN
fcis-5510	61	12	for	for	ADP
fcis-5510	61	13	text	text	NOUN
fcis-5510	61	14	-	-	PUNCT
fcis-5510	61	15	to	to	ADP
fcis-5510	61	16	-	-	PUNCT
fcis-5510	61	17	speech	speech	NOUN
fcis-5510	61	18	and	and	CCONJ
fcis-5510	61	19	merge	merge	VERB
fcis-5510	61	20	the	the	DET
fcis-5510	61	21	voice	voice	NOUN
fcis-5510	61	22	into	into	ADP
fcis-5510	61	23	the	the	DET
fcis-5510	61	24	video	video	NOUN
fcis-5510	61	25	as	as	SCONJ
fcis-5510	61	26	shown	show	VERB
fcis-5510	61	27	in	in	ADP
fcis-5510	61	28	figure	figure	NOUN
fcis-5510	61	29	3	3	NUM
fcis-5510	61	30	.	.	PUNCT
fcis-5510	61	31	first	first	ADV
fcis-5510	61	32	obtain	obtain	VERB
fcis-5510	61	33	the	the	DET
fcis-5510	61	34	text	text	NOUN
fcis-5510	61	35	that	that	PRON
fcis-5510	61	36	needs	need	VERB
fcis-5510	61	37	to	to	PART
fcis-5510	61	38	be	be	AUX
fcis-5510	61	39	replaced	replace	VERB
fcis-5510	61	40	;	;	PUNCT
fcis-5510	61	41	then	then	ADV
fcis-5510	61	42	,	,	PUNCT
fcis-5510	61	43	the	the	DET
fcis-5510	61	44	text	text	NOUN
fcis-5510	61	45	-	-	PUNCT
fcis-5510	61	46	to	to	ADP
fcis-5510	61	47	-	-	PUNCT
fcis-5510	61	48	speech	speech	NOUN
fcis-5510	61	49	encoder	encoder	NOUN
fcis-5510	61	50	retrained	retrain	VERB
fcis-5510	61	51	based	base	VERB
fcis-5510	61	52	on	on	ADP
fcis-5510	61	53	the	the	DET
fcis-5510	61	54	fusion	fusion	NOUN
fcis-5510	61	55	data_aishell	data_aishell	NOUN
fcis-5510	61	56	dataset	dataset	VERB
fcis-5510	61	57	automatically	automatically	ADV
fcis-5510	61	58	converts	convert	VERB
fcis-5510	61	59	text	text	NOUN
fcis-5510	61	60	into	into	ADP
fcis-5510	61	61	speech	speech	NOUN
fcis-5510	61	62	(	(	PUNCT
fcis-5510	61	63	and	and	CCONJ
fcis-5510	61	64	also	also	ADV
fcis-5510	61	65	supports	support	VERB
fcis-5510	61	66	manual	manual	ADJ
fcis-5510	61	67	recording	recording	NOUN
fcis-5510	61	68	of	of	ADP
fcis-5510	61	69	speech	speech	NOUN
fcis-5510	61	70	)	)	PUNCT
fcis-5510	61	71	.	.	PUNCT
fcis-5510	62	1	then	then	ADV
fcis-5510	62	2	,	,	PUNCT
fcis-5510	62	3	the	the	DET
fcis-5510	62	4	converted	converted	ADJ
fcis-5510	62	5	voice	voice	NOUN
fcis-5510	62	6	is	be	AUX
fcis-5510	62	7	normalized	normalize	VERB
fcis-5510	62	8	and	and	CCONJ
fcis-5510	62	9	integrated	integrate	VERB
fcis-5510	62	10	with	with	ADP
fcis-5510	62	11	the	the	DET
fcis-5510	62	12	original	original	ADJ
fcis-5510	62	13	voice	voice	NOUN
fcis-5510	62	14	intensity	intensity	NOUN
fcis-5510	62	15	,	,	PUNCT
fcis-5510	62	16	prosody	prosody	NOUN
fcis-5510	62	17	and	and	CCONJ
fcis-5510	62	18	duration	duration	NOUN
fcis-5510	62	19	.	.	PUNCT
fcis-5510	63	1	finally	finally	ADV
fcis-5510	63	2	,	,	PUNCT
fcis-5510	63	3	the	the	DET
fcis-5510	63	4	fused	fuse	VERB
fcis-5510	63	5	audio	audio	NOUN
fcis-5510	63	6	is	be	AUX
fcis-5510	63	7	merged	merge	VERB
fcis-5510	63	8	with	with	ADP
fcis-5510	63	9	the	the	DET
fcis-5510	63	10	silent	silent	ADJ
fcis-5510	63	11	video	video	NOUN
fcis-5510	63	12	to	to	PART
fcis-5510	63	13	get	get	VERB
fcis-5510	63	14	the	the	DET
fcis-5510	63	15	corrected	correct	VERB
fcis-5510	63	16	video	video	NOUN
fcis-5510	63	17	.	.	PUNCT
fcis-5510	64	1	figure	figure	NOUN
fcis-5510	64	2	3	3	NUM
fcis-5510	64	3	.	.	PUNCT
fcis-5510	65	1	the	the	DET
fcis-5510	65	2	process	process	NOUN
fcis-5510	65	3	of	of	ADP
fcis-5510	65	4	i	i	PROPN
fcis-5510	65	5	-	-	PUNCT
fcis-5510	65	6	zhrtvc	zhrtvc	PROPN
fcis-5510	65	7	for	for	ADP
fcis-5510	65	8	text	text	NOUN
fcis-5510	65	9	-	-	PUNCT
fcis-5510	65	10	to	to	ADP
fcis-5510	65	11	-	-	PUNCT
fcis-5510	65	12	speech	speech	NOUN
fcis-5510	65	13	and	and	CCONJ
fcis-5510	65	14	merge	merge	VERB
fcis-5510	65	15	the	the	DET
fcis-5510	65	16	voice	voice	NOUN
fcis-5510	65	17	into	into	ADP
fcis-5510	65	18	the	the	DET
fcis-5510	65	19	video	video	NOUN
fcis-5510	65	20	3	3	NUM
fcis-5510	65	21	.	.	PUNCT
fcis-5510	65	22	experiment	experiment	NOUN
fcis-5510	65	23	and	and	CCONJ
fcis-5510	65	24	result	result	VERB
fcis-5510	65	25	analysis	analysis	NOUN
fcis-5510	65	26	3.1	3.1	NUM
fcis-5510	65	27	.	.	PUNCT
fcis-5510	66	1	experimental	experimental	ADJ
fcis-5510	66	2	results	result	NOUN
fcis-5510	66	3	of	of	ADP
fcis-5510	66	4	the	the	DET
fcis-5510	66	5	staged	stage	VERB
fcis-5510	66	6	and	and	CCONJ
fcis-5510	66	7	efficient	efficient	ADJ
fcis-5510	66	8	algorithm	algorithm	NOUN
fcis-5510	66	9	based	base	VERB
fcis-5510	66	10	on	on	ADP
fcis-5510	66	11	bic	bic	PROPN
fcis-5510	66	12	&	&	CCONJ
fcis-5510	66	13	medist	medist	PROPN
fcis-5510	66	14	compared	compare	VERB
fcis-5510	66	15	with	with	ADP
fcis-5510	66	16	traditional	traditional	ADJ
fcis-5510	66	17	bic	bic	PROPN
fcis-5510	66	18	audio	audio	NOUN
fcis-5510	66	19	segmentation	segmentation	NOUN
fcis-5510	66	20	125	125	NUM
fcis-5510	66	21	algorithm	algorithm	NOUN
fcis-5510	66	22	,	,	PUNCT
fcis-5510	66	23	the	the	DET
fcis-5510	66	24	staged	stage	VERB
fcis-5510	66	25	and	and	CCONJ
fcis-5510	66	26	efficient	efficient	ADJ
fcis-5510	66	27	algorithm	algorithm	NOUN
fcis-5510	66	28	based	base	VERB
fcis-5510	66	29	on	on	ADP
fcis-5510	66	30	bic	bic	PROPN
fcis-5510	66	31	&	&	CCONJ
fcis-5510	66	32	medist	medist	PROPN
fcis-5510	66	33	is	be	AUX
fcis-5510	66	34	proposed	propose	VERB
fcis-5510	66	35	in	in	ADP
fcis-5510	66	36	this	this	DET
fcis-5510	66	37	paper	paper	NOUN
fcis-5510	66	38	.	.	PUNCT
fcis-5510	67	1	which	which	PRON
fcis-5510	67	2	discards	discard	VERB
fcis-5510	67	3	the	the	DET
fcis-5510	67	4	glr	glr	PROPN
fcis-5510	67	5	(	(	PUNCT
fcis-5510	67	6	general	general	ADJ
fcis-5510	67	7	likelihood	likelihood	NOUN
fcis-5510	67	8	)	)	PUNCT
fcis-5510	67	9	comparable	comparable	ADJ
fcis-5510	67	10	distance	distance	NOUN
fcis-5510	67	11	calculation	calculation	NOUN
fcis-5510	67	12	formulas	formula	NOUN
fcis-5510	67	13	such	such	ADJ
fcis-5510	67	14	as	as	ADP
fcis-5510	67	15	(	(	PUNCT
fcis-5510	67	16	kullback	kullback	NOUN
fcis-5510	67	17	-	-	PUNCT
fcis-5510	67	18	leibler)kl	leibler)kl	PROPN
fcis-5510	67	19	,	,	PUNCT
fcis-5510	67	20	etc	etc	X
fcis-5510	67	21	.	.	X
fcis-5510	67	22	,	,	PUNCT
fcis-5510	67	23	use	use	VERB
fcis-5510	67	24	the	the	DET
fcis-5510	67	25	statistical	statistical	ADJ
fcis-5510	67	26	mean	mean	NOUN
fcis-5510	67	27	euclidean	euclidean	ADJ
fcis-5510	67	28	distance	distance	NOUN
fcis-5510	67	29	medist	medist	NOUN
fcis-5510	67	30	,	,	PUNCT
fcis-5510	67	31	combined	combine	VERB
fcis-5510	67	32	with	with	ADP
fcis-5510	67	33	the	the	DET
fcis-5510	67	34	local	local	ADJ
fcis-5510	67	35	maximum	maximum	ADJ
fcis-5510	67	36	selection	selection	NOUN
fcis-5510	67	37	and	and	CCONJ
fcis-5510	67	38	significance	significance	NOUN
fcis-5510	67	39	detection	detection	NOUN
fcis-5510	67	40	.	.	PUNCT
fcis-5510	68	1	then	then	ADV
fcis-5510	68	2	use	use	VERB
fcis-5510	68	3	the	the	DET
fcis-5510	68	4	bic	bic	PROPN
fcis-5510	68	5	method	method	NOUN
fcis-5510	68	6	to	to	PART
fcis-5510	68	7	confirm	confirm	VERB
fcis-5510	68	8	the	the	DET
fcis-5510	68	9	candidate	candidate	NOUN
fcis-5510	68	10	segmentation	segmentation	NOUN
fcis-5510	68	11	points	point	NOUN
fcis-5510	68	12	.	.	PUNCT
fcis-5510	69	1	the	the	DET
fcis-5510	69	2	test	test	NOUN
fcis-5510	69	3	results	result	NOUN
fcis-5510	69	4	show	show	VERB
fcis-5510	69	5	that	that	SCONJ
fcis-5510	69	6	this	this	DET
fcis-5510	69	7	method	method	NOUN
fcis-5510	69	8	not	not	PART
fcis-5510	69	9	only	only	ADV
fcis-5510	69	10	improves	improve	VERB
fcis-5510	69	11	the	the	DET
fcis-5510	69	12	overall	overall	ADJ
fcis-5510	69	13	audio	audio	ADJ
fcis-5510	69	14	segmentation	segmentation	NOUN
fcis-5510	69	15	speed	speed	NOUN
fcis-5510	69	16	to	to	ADP
fcis-5510	69	17	a	a	DET
fcis-5510	69	18	large	large	ADJ
fcis-5510	69	19	extent	extent	NOUN
fcis-5510	69	20	,	,	PUNCT
fcis-5510	69	21	it	it	PRON
fcis-5510	69	22	is	be	AUX
fcis-5510	69	23	400	400	NUM
fcis-5510	69	24	times	time	NOUN
fcis-5510	69	25	higher	high	ADJ
fcis-5510	69	26	than	than	ADP
fcis-5510	69	27	the	the	DET
fcis-5510	69	28	traditional	traditional	ADJ
fcis-5510	69	29	bic	bic	NOUN
fcis-5510	69	30	method	method	NOUN
fcis-5510	69	31	,	,	PUNCT
fcis-5510	69	32	and	and	CCONJ
fcis-5510	69	33	the	the	DET
fcis-5510	69	34	deletion	deletion	NOUN
fcis-5510	69	35	error	error	NOUN
fcis-5510	69	36	rate	rate	NOUN
fcis-5510	69	37	mdr	mdr	PROPN
fcis-5510	69	38	is	be	AUX
fcis-5510	69	39	reduced	reduce	VERB
fcis-5510	69	40	by	by	ADP
fcis-5510	69	41	15.1	15.1	NUM
fcis-5510	69	42	%	%	NOUN
fcis-5510	69	43	.	.	PUNCT
fcis-5510	70	1	the	the	DET
fcis-5510	70	2	insertion	insertion	NOUN
fcis-5510	70	3	error	error	NOUN
fcis-5510	70	4	rate	rate	NOUN
fcis-5510	70	5	far	far	ADV
fcis-5510	70	6	is	be	AUX
fcis-5510	70	7	slightly	slightly	ADV
fcis-5510	70	8	reduced	reduce	VERB
fcis-5510	70	9	,	,	PUNCT
fcis-5510	70	10	which	which	PRON
fcis-5510	70	11	greatly	greatly	ADV
fcis-5510	70	12	improves	improve	VERB
fcis-5510	70	13	the	the	DET
fcis-5510	70	14	accuracy	accuracy	NOUN
fcis-5510	70	15	of	of	ADP
fcis-5510	70	16	text	text	NOUN
fcis-5510	70	17	-	-	PUNCT
fcis-5510	70	18	to	to	ADP
fcis-5510	70	19	-	-	PUNCT
fcis-5510	70	20	text	text	NOUN
fcis-5510	70	21	conversion	conversion	NOUN
fcis-5510	70	22	after	after	ADP
fcis-5510	70	23	speech	speech	NOUN
fcis-5510	70	24	extraction	extraction	NOUN
fcis-5510	70	25	.	.	PUNCT
fcis-5510	71	1	audio	audio	ADJ
fcis-5510	71	2	split	split	NOUN
fcis-5510	71	3	point	point	NOUN
fcis-5510	71	4	insertion	insertion	NOUN
fcis-5510	71	5	error	error	NOUN
fcis-5510	71	6	(	(	PUNCT
fcis-5510	71	7	fa	fa	NOUN
fcis-5510	71	8	)	)	PUNCT
fcis-5510	71	9	means	mean	VERB
fcis-5510	71	10	that	that	SCONJ
fcis-5510	71	11	a	a	DET
fcis-5510	71	12	split	split	NOUN
fcis-5510	71	13	point	point	NOUN
fcis-5510	71	14	that	that	PRON
fcis-5510	71	15	does	do	AUX
fcis-5510	71	16	not	not	PART
fcis-5510	71	17	actually	actually	ADV
fcis-5510	71	18	exist	exist	VERB
fcis-5510	71	19	is	be	AUX
fcis-5510	71	20	detected	detect	VERB
fcis-5510	71	21	;	;	PUNCT
fcis-5510	71	22	audio	audio	NOUN
fcis-5510	71	23	split	split	NOUN
fcis-5510	71	24	point	point	NOUN
fcis-5510	71	25	deletion	deletion	NOUN
fcis-5510	71	26	error	error	NOUN
fcis-5510	71	27	(	(	PUNCT
fcis-5510	71	28	md	md	PROPN
fcis-5510	71	29	)	)	PUNCT
fcis-5510	71	30	means	mean	VERB
fcis-5510	71	31	that	that	SCONJ
fcis-5510	71	32	a	a	DET
fcis-5510	71	33	split	split	NOUN
fcis-5510	71	34	point	point	NOUN
fcis-5510	71	35	that	that	PRON
fcis-5510	71	36	actually	actually	ADV
fcis-5510	71	37	exists	exist	VERB
fcis-5510	71	38	is	be	AUX
fcis-5510	71	39	not	not	PART
fcis-5510	71	40	detected	detect	VERB
fcis-5510	71	41	.	.	PUNCT
fcis-5510	72	1	fa	fa	INTJ
fcis-5510	72	2	(	(	PUNCT
fcis-5510	72	3	false	false	ADJ
fcis-5510	72	4	alarms	alarm	NOUN
fcis-5510	72	5	rate	rate	NOUN
fcis-5510	72	6	)	)	PUNCT
fcis-5510	72	7	and	and	CCONJ
fcis-5510	72	8	mdr	mdr	PROPN
fcis-5510	72	9	(	(	PUNCT
fcis-5510	72	10	missed	miss	VERB
fcis-5510	72	11	detection	detection	NOUN
fcis-5510	72	12	rate	rate	NOUN
fcis-5510	72	13	)	)	PUNCT
fcis-5510	72	14	are	be	AUX
fcis-5510	72	15	defined	define	VERB
fcis-5510	72	16	as	as	SCONJ
fcis-5510	72	17	follows	follow	VERB
fcis-5510	72	18	:	:	PUNCT
fcis-5510	73	1	𝐹𝐴𝑅	𝐹𝐴𝑅	PROPN
fcis-5510	73	2	=	=	SYM
fcis-5510	73	3	𝐹𝐴	𝐹𝐴	PROPN
fcis-5510	73	4	𝐴𝐶𝑃+𝐹𝐴	𝐴𝐶𝑃+𝐹𝐴	NOUN
fcis-5510	73	5	×	×	NOUN
fcis-5510	73	6	100	100	NUM
fcis-5510	73	7	%	%	NOUN
fcis-5510	73	8	(	(	PUNCT
fcis-5510	73	9	8)	8)	NUM
fcis-5510	73	10	𝑀𝐷𝑅	𝑀𝐷𝑅	NOUN
fcis-5510	73	11	=	=	SYM
fcis-5510	73	12	𝑀𝐷	𝑀𝐷	PROPN
fcis-5510	73	13	𝐴𝐶𝑃	𝐴𝐶𝑃	PROPN
fcis-5510	73	14	×	×	NOUN
fcis-5510	73	15	100	100	NUM
fcis-5510	73	16	%	%	NOUN
fcis-5510	73	17	(	(	PUNCT
fcis-5510	73	18	9	9	NUM
fcis-5510	73	19	)	)	PUNCT
fcis-5510	73	20	among	among	ADP
fcis-5510	73	21	them	they	PRON
fcis-5510	73	22	,	,	PUNCT
fcis-5510	73	23	acp	acp	PROPN
fcis-5510	73	24	(	(	PUNCT
fcis-5510	73	25	actual	actual	ADJ
fcis-5510	73	26	changing	change	VERB
fcis-5510	73	27	points	point	NOUN
fcis-5510	73	28	)	)	PUNCT
fcis-5510	73	29	represents	represent	VERB
fcis-5510	73	30	the	the	DET
fcis-5510	73	31	real	real	ADJ
fcis-5510	73	32	audio	audio	ADJ
fcis-5510	73	33	split	split	NOUN
fcis-5510	73	34	point	point	NOUN
fcis-5510	73	35	.	.	PUNCT
fcis-5510	74	1	similarly	similarly	ADV
fcis-5510	74	2	,	,	PUNCT
fcis-5510	74	3	we	we	PRON
fcis-5510	74	4	can	can	AUX
fcis-5510	74	5	define	define	VERB
fcis-5510	74	6	the	the	DET
fcis-5510	74	7	total	total	ADJ
fcis-5510	74	8	error	error	NOUN
fcis-5510	74	9	score	score	NOUN
fcis-5510	74	10	es	es	X
fcis-5510	74	11	(	(	PUNCT
fcis-5510	74	12	error	error	NOUN
fcis-5510	74	13	score	score	NOUN
fcis-5510	74	14	)	)	PUNCT
fcis-5510	74	15	es	es	NOUN
fcis-5510	74	16	=	=	PUNCT
fcis-5510	74	17	far	far	ADV
fcis-5510	74	18	+	+	CCONJ
fcis-5510	74	19	mdr	mdr	X
fcis-5510	74	20	(	(	PUNCT
fcis-5510	74	21	10	10	NUM
fcis-5510	74	22	)	)	PUNCT
fcis-5510	74	23	the	the	DET
fcis-5510	74	24	specific	specific	ADJ
fcis-5510	74	25	algorithm	algorithm	NOUN
fcis-5510	74	26	comparison	comparison	NOUN
fcis-5510	74	27	chart	chart	NOUN
fcis-5510	74	28	is	be	AUX
fcis-5510	74	29	shown	show	VERB
fcis-5510	74	30	in	in	ADP
fcis-5510	74	31	the	the	DET
fcis-5510	74	32	figure	figure	NOUN
fcis-5510	74	33	4	4	NUM
fcis-5510	74	34	.	.	PUNCT
fcis-5510	74	35	figure	figure	VERB
fcis-5510	74	36	4	4	NUM
fcis-5510	74	37	.	.	PUNCT
fcis-5510	75	1	the	the	DET
fcis-5510	75	2	specific	specific	ADJ
fcis-5510	75	3	algorithm	algorithm	NOUN
fcis-5510	75	4	comparison	comparison	NOUN
fcis-5510	75	5	table	table	NOUN
fcis-5510	75	6	1	1	NUM
fcis-5510	75	7	.	.	PUNCT
fcis-5510	76	1	accuracy	accuracy	NOUN
fcis-5510	76	2	comparison	comparison	NOUN
fcis-5510	76	3	of	of	ADP
fcis-5510	76	4	speech	speech	NOUN
fcis-5510	76	5	extraction	extraction	NOUN
fcis-5510	76	6	algorithms	algorithm	NOUN
fcis-5510	76	7	total	total	ADJ
fcis-5510	76	8	number	number	NOUN
fcis-5510	76	9	of	of	ADP
fcis-5510	76	10	test	test	NOUN
fcis-5510	76	11	cases	case	NOUN
fcis-5510	76	12	speechtospeech	speechtospeech	NOUN
fcis-5510	76	13	error	error	NOUN
fcis-5510	76	14	rate	rate	NOUN
fcis-5510	76	15	<	<	NOUN
fcis-5510	76	16	10	10	NUM
fcis-5510	76	17	%	%	NOUN
fcis-5510	76	18	speechto	speechto	NOUN
fcis-5510	76	19	-	-	NOUN
fcis-5510	76	20	speech	speech	NOUN
fcis-5510	76	21	error	error	NOUN
fcis-5510	76	22	rate	rate	NOUN
fcis-5510	76	23	<	<	NOUN
fcis-5510	76	24	5	5	NUM
fcis-5510	76	25	%	%	NOUN
fcis-5510	76	26	speechto	speechto	NOUN
fcis-5510	76	27	-	-	NOUN
fcis-5510	76	28	speech	speech	NOUN
fcis-5510	76	29	error	error	NOUN
fcis-5510	76	30	rate	rate	NOUN
fcis-5510	76	31	<3	<3	X
fcis-5510	76	32	%	%	NOUN
fcis-5510	76	33	new	new	ADJ
fcis-5510	76	34	design	design	NOUN
fcis-5510	76	35	speech	speech	NOUN
fcis-5510	76	36	extraction	extraction	NOUN
fcis-5510	76	37	algorithm	algorithm	NOUN
fcis-5510	76	38	28	28	NUM
fcis-5510	76	39	100	100	NUM
fcis-5510	76	40	%	%	NOUN
fcis-5510	76	41	92.80	92.80	NUM
fcis-5510	76	42	%	%	NOUN
fcis-5510	76	43	89.20	89.20	NUM
fcis-5510	76	44	%	%	NOUN
fcis-5510	76	45	traditional	traditional	ADJ
fcis-5510	76	46	speech	speech	NOUN
fcis-5510	76	47	extraction	extraction	NOUN
fcis-5510	76	48	algorithms	algorithm	NOUN
fcis-5510	76	49	29	29	NUM
fcis-5510	76	50	36	36	NUM
fcis-5510	76	51	%	%	NOUN
fcis-5510	76	52	10.70	10.70	NUM
fcis-5510	76	53	%	%	NOUN
fcis-5510	76	54	0.00	0.00	NUM
fcis-5510	76	55	%	%	NOUN
fcis-5510	76	56	in	in	ADP
fcis-5510	76	57	terms	term	NOUN
fcis-5510	76	58	of	of	ADP
fcis-5510	76	59	speech	speech	NOUN
fcis-5510	76	60	-	-	PUNCT
fcis-5510	76	61	to	to	ADP
fcis-5510	76	62	-	-	PUNCT
fcis-5510	76	63	text	text	NOUN
fcis-5510	76	64	noise	noise	NOUN
fcis-5510	76	65	reduction	reduction	NOUN
fcis-5510	76	66	in	in	ADP
fcis-5510	76	67	speech	speech	NOUN
fcis-5510	76	68	extraction	extraction	NOUN
fcis-5510	76	69	,	,	PUNCT
fcis-5510	76	70	by	by	ADP
fcis-5510	76	71	analyzing	analyze	VERB
fcis-5510	76	72	the	the	DET
fcis-5510	76	73	audio	audio	NOUN
fcis-5510	76	74	and	and	CCONJ
fcis-5510	76	75	processing	process	VERB
fcis-5510	76	76	the	the	DET
fcis-5510	76	77	audio	audio	NOUN
fcis-5510	76	78	through	through	ADP
fcis-5510	76	79	fft	fft	PROPN
fcis-5510	76	80	(	(	PUNCT
fcis-5510	76	81	fast	fast	ADJ
fcis-5510	76	82	fourier	fourier	NOUN
fcis-5510	76	83	transform	transform	NOUN
fcis-5510	76	84	)	)	PUNCT
fcis-5510	76	85	,	,	PUNCT
fcis-5510	76	86	noise	noise	NOUN
fcis-5510	76	87	estimation	estimation	NOUN
fcis-5510	76	88	and	and	CCONJ
fcis-5510	76	89	amplitude	amplitude	NOUN
fcis-5510	76	90	quadratic	quadratic	ADJ
fcis-5510	76	91	calculation	calculation	NOUN
fcis-5510	76	92	are	be	AUX
fcis-5510	76	93	performed	perform	VERB
fcis-5510	76	94	to	to	PART
fcis-5510	76	95	achieve	achieve	VERB
fcis-5510	76	96	spectral	spectral	ADJ
fcis-5510	76	97	reduction	reduction	NOUN
fcis-5510	76	98	,	,	PUNCT
fcis-5510	76	99	which	which	PRON
fcis-5510	76	100	can	can	AUX
fcis-5510	76	101	not	not	PART
fcis-5510	76	102	only	only	ADV
fcis-5510	76	103	noise	noise	NOUN
fcis-5510	76	104	reduction	reduction	NOUN
fcis-5510	76	105	,	,	PUNCT
fcis-5510	76	106	it	it	PRON
fcis-5510	76	107	also	also	ADV
fcis-5510	76	108	has	have	VERB
fcis-5510	76	109	a	a	DET
fcis-5510	76	110	certain	certain	ADJ
fcis-5510	76	111	ability	ability	NOUN
fcis-5510	76	112	to	to	PART
fcis-5510	76	113	recognize	recognize	VERB
fcis-5510	76	114	some	some	DET
fcis-5510	76	115	audio	audio	NOUN
fcis-5510	76	116	with	with	ADP
fcis-5510	76	117	dialect	dialect	ADJ
fcis-5510	76	118	accents	accent	NOUN
fcis-5510	76	119	except	except	SCONJ
fcis-5510	76	120	mandarin	mandarin	PROPN
fcis-5510	76	121	chinese	chinese	PROPN
fcis-5510	76	122	.	.	PUNCT
fcis-5510	77	1	after	after	ADP
fcis-5510	77	2	20,000	20,000	NUM
fcis-5510	77	3	audio	audio	ADJ
fcis-5510	77	4	tests	test	NOUN
fcis-5510	77	5	,	,	PUNCT
fcis-5510	77	6	we	we	PRON
fcis-5510	77	7	concluded	conclude	VERB
fcis-5510	77	8	that	that	SCONJ
fcis-5510	77	9	the	the	DET
fcis-5510	77	10	audio	audio	NOUN
fcis-5510	77	11	after	after	ADP
fcis-5510	77	12	noise	noise	NOUN
fcis-5510	77	13	reduction	reduction	NOUN
fcis-5510	77	14	can	can	AUX
fcis-5510	77	15	have	have	VERB
fcis-5510	77	16	an	an	DET
fcis-5510	77	17	average	average	ADJ
fcis-5510	77	18	recognition	recognition	NOUN
fcis-5510	77	19	accuracy	accuracy	NOUN
fcis-5510	77	20	of	of	ADP
fcis-5510	77	21	95.8	95.8	NUM
fcis-5510	77	22	%	%	NOUN
fcis-5510	77	23	,	,	PUNCT
fcis-5510	77	24	which	which	PRON
fcis-5510	77	25	can	can	AUX
fcis-5510	77	26	improve	improve	VERB
fcis-5510	77	27	the	the	DET
fcis-5510	77	28	accuracy	accuracy	NOUN
fcis-5510	77	29	of	of	ADP
fcis-5510	77	30	text	text	NOUN
fcis-5510	77	31	conversion	conversion	NOUN
fcis-5510	77	32	.	.	PUNCT
fcis-5510	78	1	combining	combine	VERB
fcis-5510	78	2	the	the	DET
fcis-5510	78	3	above	above	ADJ
fcis-5510	78	4	two	two	NUM
fcis-5510	78	5	points	point	NOUN
fcis-5510	78	6	and	and	CCONJ
fcis-5510	78	7	testing	test	VERB
fcis-5510	78	8	a	a	DET
fcis-5510	78	9	set	set	NOUN
fcis-5510	78	10	of	of	ADP
fcis-5510	78	11	java	java	PROPN
fcis-5510	78	12	videos	video	NOUN
fcis-5510	78	13	(	(	PUNCT
fcis-5510	78	14	28	28	NUM
fcis-5510	78	15	)	)	PUNCT
fcis-5510	78	16	,	,	PUNCT
fcis-5510	78	17	compared	compare	VERB
fcis-5510	78	18	with	with	ADP
fcis-5510	78	19	the	the	DET
fcis-5510	78	20	traditional	traditional	ADJ
fcis-5510	78	21	algorithm	algorithm	NOUN
fcis-5510	78	22	,	,	PUNCT
fcis-5510	78	23	the	the	DET
fcis-5510	78	24	new	new	ADJ
fcis-5510	78	25	algorithm	algorithm	NOUN
fcis-5510	78	26	can	can	AUX
fcis-5510	78	27	extract	extract	VERB
fcis-5510	78	28	language	language	NOUN
fcis-5510	78	29	with	with	ADP
fcis-5510	78	30	an	an	DET
fcis-5510	78	31	accuracy	accuracy	NOUN
fcis-5510	78	32	of	of	ADP
fcis-5510	78	33	95.8	95.8	NUM
fcis-5510	78	34	%	%	NOUN
fcis-5510	78	35	.	.	PUNCT
fcis-5510	79	1	there	there	PRON
fcis-5510	79	2	is	be	VERB
fcis-5510	79	3	no	no	DET
fcis-5510	79	4	error	error	NOUN
fcis-5510	79	5	in	in	ADP
fcis-5510	79	6	converting	convert	VERB
fcis-5510	79	7	most	most	ADJ
fcis-5510	79	8	audio	audio	ADJ
fcis-5510	79	9	segments	segment	NOUN
fcis-5510	79	10	to	to	ADP
fcis-5510	79	11	speech	speech	NOUN
fcis-5510	79	12	,	,	PUNCT
fcis-5510	79	13	and	and	CCONJ
fcis-5510	79	14	the	the	DET
fcis-5510	79	15	error	error	NOUN
fcis-5510	79	16	rate	rate	NOUN
fcis-5510	79	17	in	in	ADP
fcis-5510	79	18	a	a	DET
fcis-5510	79	19	small	small	ADJ
fcis-5510	79	20	number	number	NOUN
fcis-5510	79	21	of	of	ADP
fcis-5510	79	22	use	use	NOUN
fcis-5510	79	23	cases	case	NOUN
fcis-5510	79	24	can	can	AUX
fcis-5510	79	25	also	also	ADV
fcis-5510	79	26	be	be	AUX
fcis-5510	79	27	controlled	control	VERB
fcis-5510	79	28	below	below	ADP
fcis-5510	79	29	20	20	NUM
fcis-5510	79	30	%	%	NOUN
fcis-5510	79	31	.	.	PUNCT
fcis-5510	80	1	however	however	ADV
fcis-5510	80	2	,	,	PUNCT
fcis-5510	80	3	based	base	VERB
fcis-5510	80	4	on	on	ADP
fcis-5510	80	5	the	the	DET
fcis-5510	80	6	traditional	traditional	ADJ
fcis-5510	80	7	speech	speech	NOUN
fcis-5510	80	8	extraction	extraction	NOUN
fcis-5510	80	9	algorithm	algorithm	NOUN
fcis-5510	80	10	,	,	PUNCT
fcis-5510	80	11	there	there	PRON
fcis-5510	80	12	are	be	VERB
fcis-5510	80	13	large	large	ADJ
fcis-5510	80	14	errors	error	NOUN
fcis-5510	80	15	in	in	ADP
fcis-5510	80	16	the	the	DET
fcis-5510	80	17	speech	speech	NOUN
fcis-5510	80	18	extraction	extraction	NOUN
fcis-5510	80	19	of	of	ADP
fcis-5510	80	20	audio	audio	ADJ
fcis-5510	80	21	segments	segment	NOUN
fcis-5510	80	22	with	with	ADP
fcis-5510	80	23	high	high	ADJ
fcis-5510	80	24	noise	noise	NOUN
fcis-5510	80	25	or	or	CCONJ
fcis-5510	80	26	audio	audio	NOUN
fcis-5510	80	27	segments	segment	NOUN
fcis-5510	80	28	with	with	ADP
fcis-5510	80	29	low	low	ADJ
fcis-5510	80	30	segmentation	segmentation	NOUN
fcis-5510	80	31	accuracy	accuracy	NOUN
fcis-5510	80	32	,	,	PUNCT
fcis-5510	80	33	and	and	CCONJ
fcis-5510	80	34	the	the	DET
fcis-5510	80	35	correct	correct	ADJ
fcis-5510	80	36	rate	rate	NOUN
fcis-5510	80	37	is	be	AUX
fcis-5510	80	38	only	only	ADV
fcis-5510	80	39	35.3	35.3	NUM
fcis-5510	80	40	%	%	NOUN
fcis-5510	80	41	.	.	PUNCT
fcis-5510	81	1	the	the	DET
fcis-5510	81	2	test	test	NOUN
fcis-5510	81	3	results	result	NOUN
fcis-5510	81	4	are	be	AUX
fcis-5510	81	5	shown	show	VERB
fcis-5510	81	6	in	in	ADP
fcis-5510	81	7	table	table	NOUN
fcis-5510	81	8	1	1	NUM
fcis-5510	81	9	.	.	NOUN
fcis-5510	81	10	3.2	3.2	NUM
fcis-5510	81	11	.	.	PUNCT
fcis-5510	82	1	experimental	experimental	ADJ
fcis-5510	82	2	results	result	NOUN
fcis-5510	82	3	of	of	ADP
fcis-5510	82	4	the	the	DET
fcis-5510	82	5	daec	daec	ADJ
fcis-5510	82	6	algorithm	algorithm	NOUN
fcis-5510	82	7	the	the	DET
fcis-5510	82	8	video	video	NOUN
fcis-5510	82	9	voice	voice	NOUN
fcis-5510	82	10	text	text	NOUN
fcis-5510	82	11	audit	audit	NOUN
fcis-5510	82	12	scheme	scheme	NOUN
fcis-5510	82	13	corrected	correct	VERB
fcis-5510	82	14	by	by	ADP
fcis-5510	82	15	daec	daec	ADJ
fcis-5510	82	16	algorithm	algorithm	NOUN
fcis-5510	82	17	has	have	VERB
fcis-5510	82	18	a	a	DET
fcis-5510	82	19	high	high	ADJ
fcis-5510	82	20	error	error	NOUN
fcis-5510	82	21	correction	correction	NOUN
fcis-5510	82	22	rate	rate	NOUN
fcis-5510	82	23	.	.	PUNCT
fcis-5510	83	1	in	in	ADP
fcis-5510	83	2	20000	20000	NUM
fcis-5510	83	3	text	text	NOUN
fcis-5510	83	4	error	error	NOUN
fcis-5510	83	5	correction	correction	NOUN
fcis-5510	83	6	tests	test	VERB
fcis-5510	83	7	,	,	PUNCT
fcis-5510	83	8	the	the	DET
fcis-5510	83	9	correct	correct	ADJ
fcis-5510	83	10	rate	rate	NOUN
fcis-5510	83	11	of	of	ADP
fcis-5510	83	12	normal	normal	ADJ
fcis-5510	83	13	text	text	NOUN
fcis-5510	83	14	and	and	CCONJ
fcis-5510	83	15	spoken	spoken	ADJ
fcis-5510	83	16	text	text	NOUN
fcis-5510	83	17	of	of	ADP
fcis-5510	83	18	the	the	DET
fcis-5510	83	19	new	new	ADJ
fcis-5510	83	20	double	double	ADJ
fcis-5510	83	21	error	error	NOUN
fcis-5510	83	22	correction	correction	NOUN
fcis-5510	83	23	method	method	NOUN
fcis-5510	83	24	has	have	AUX
fcis-5510	83	25	increased	increase	VERB
fcis-5510	83	26	by	by	ADP
fcis-5510	83	27	10.8	10.8	NUM
fcis-5510	83	28	%	%	NOUN
fcis-5510	83	29	and	and	CCONJ
fcis-5510	83	30	15.3	15.3	NUM
fcis-5510	83	31	%	%	NOUN
fcis-5510	83	32	respectively	respectively	ADV
fcis-5510	83	33	compared	compare	VERB
fcis-5510	83	34	with	with	ADP
fcis-5510	83	35	bert	bert	NOUN
fcis-5510	83	36	,	,	PUNCT
fcis-5510	83	37	and	and	CCONJ
fcis-5510	83	38	by	by	ADP
fcis-5510	83	39	0.7	0.7	NUM
fcis-5510	83	40	%	%	NOUN
fcis-5510	83	41	and	and	CCONJ
fcis-5510	83	42	2.0	2.0	NUM
fcis-5510	83	43	%	%	NOUN
fcis-5510	83	44	respectively	respectively	ADV
fcis-5510	83	45	compared	compare	VERB
fcis-5510	83	46	with	with	ADP
fcis-5510	83	47	baidu	baidu	NOUN
fcis-5510	83	48	,	,	PUNCT
fcis-5510	83	49	as	as	SCONJ
fcis-5510	83	50	shown	show	VERB
fcis-5510	83	51	in	in	ADP
fcis-5510	83	52	table	table	NOUN
fcis-5510	83	53	2	2	NUM
fcis-5510	83	54	.	.	PUNCT
fcis-5510	83	55	table	table	NOUN
fcis-5510	83	56	2	2	NUM
fcis-5510	83	57	.	.	PUNCT
fcis-5510	83	58	comparison	comparison	NOUN
fcis-5510	83	59	of	of	ADP
fcis-5510	83	60	algorithms	algorithm	NOUN
fcis-5510	83	61	normal	normal	ADJ
fcis-5510	83	62	text	text	NOUN
fcis-5510	83	63	error	error	NOUN
fcis-5510	83	64	correction	correction	NOUN
fcis-5510	83	65	rate	rate	NOUN
fcis-5510	83	66	correctness	correctness	NOUN
fcis-5510	83	67	rate	rate	NOUN
fcis-5510	83	68	of	of	ADP
fcis-5510	83	69	more	more	ADJ
fcis-5510	83	70	spoken	spoken	ADJ
fcis-5510	83	71	texts	text	NOUN
fcis-5510	83	72	daec	daec	ADJ
fcis-5510	83	73	98.30	98.30	NUM
fcis-5510	83	74	%	%	NOUN
fcis-5510	83	75	96.60	96.60	NUM
fcis-5510	83	76	%	%	NOUN
fcis-5510	83	77	baidu	baidu	VERB
fcis-5510	83	78	97.60	97.60	NUM
fcis-5510	83	79	%	%	NOUN
fcis-5510	83	80	94.60	94.60	NUM
fcis-5510	83	81	%	%	NOUN
fcis-5510	83	82	bert	bert	NOUN
fcis-5510	83	83	77.50	77.50	NUM
fcis-5510	83	84	%	%	NOUN
fcis-5510	83	85	81.30	81.30	NUM
fcis-5510	83	86	%	%	NOUN
fcis-5510	83	87	3.3	3.3	NUM
fcis-5510	83	88	.	.	PUNCT
fcis-5510	84	1	experimental	experimental	ADJ
fcis-5510	84	2	results	result	NOUN
fcis-5510	84	3	of	of	ADP
fcis-5510	84	4	i	i	PROPN
fcis-5510	84	5	-	-	PUNCT
fcis-5510	84	6	zhrtvc	zhrtvc	PROPN
fcis-5510	84	7	based	base	VERB
fcis-5510	84	8	on	on	ADP
fcis-5510	84	9	the	the	DET
fcis-5510	84	10	innovative	innovative	ADJ
fcis-5510	84	11	zhrtvc	zhrtvc	NOUN
fcis-5510	84	12	algorithm	algorithm	NOUN
fcis-5510	84	13	,	,	PUNCT
fcis-5510	84	14	this	this	DET
fcis-5510	84	15	paer	paer	NOUN
fcis-5510	84	16	retrained	retrain	VERB
fcis-5510	84	17	with	with	ADP
fcis-5510	84	18	chinese	chinese	ADJ
fcis-5510	84	19	data	datum	NOUN
fcis-5510	84	20	from	from	ADP
fcis-5510	84	21	400	400	NUM
fcis-5510	84	22	people	people	NOUN
fcis-5510	84	23	in	in	ADP
fcis-5510	84	24	data_aishell	data_aishell	PROPN
fcis-5510	84	25	,	,	PUNCT
fcis-5510	84	26	and	and	CCONJ
fcis-5510	84	27	in	in	ADP
fcis-5510	84	28	vocoder	vocoder	NOUN
fcis-5510	84	29	,	,	PUNCT
fcis-5510	84	30	mel	mel	PROPN
fcis-5510	84	31	-	-	PUNCT
fcis-5510	84	32	spectrogram	spectrogram	NOUN
fcis-5510	84	33	generated	generate	VERB
fcis-5510	84	34	by	by	ADP
fcis-5510	84	35	synthesistrained	synthesistrained	ADJ
fcis-5510	84	36	models	model	NOUN
fcis-5510	84	37	was	be	AUX
fcis-5510	84	38	used	use	VERB
fcis-5510	84	39	instead	instead	ADV
fcis-5510	84	40	of	of	ADP
fcis-5510	84	41	the	the	DET
fcis-5510	84	42	original	original	ADJ
fcis-5510	84	43	griffin	griffin	NOUN
fcis-5510	84	44	-	-	PUNCT
fcis-5510	84	45	lim	lim	PROPN
fcis-5510	84	46	.	.	PUNCT
fcis-5510	85	1	in	in	ADP
fcis-5510	85	2	addition	addition	NOUN
fcis-5510	85	3	to	to	ADP
fcis-5510	85	4	the	the	DET
fcis-5510	85	5	realization	realization	NOUN
fcis-5510	85	6	of	of	ADP
fcis-5510	85	7	automatic	automatic	ADJ
fcis-5510	85	8	text	text	NOUN
fcis-5510	85	9	-	-	PUNCT
fcis-5510	85	10	to	to	ADP
fcis-5510	85	11	-	-	PUNCT
fcis-5510	85	12	speech	speech	NOUN
fcis-5510	85	13	(	(	PUNCT
fcis-5510	85	14	tts	tts	PROPN
fcis-5510	85	15	)	)	PUNCT
fcis-5510	85	16	,	,	PUNCT
fcis-5510	85	17	the	the	DET
fcis-5510	85	18	improved	improved	ADJ
fcis-5510	85	19	algorithm	algorithm	NOUN
fcis-5510	85	20	also	also	ADV
fcis-5510	85	21	improves	improve	VERB
fcis-5510	85	22	the	the	DET
fcis-5510	85	23	prosody	prosody	NOUN
fcis-5510	85	24	accuracy	accuracy	NOUN
fcis-5510	85	25	by	by	ADP
fcis-5510	85	26	15	15	NUM
fcis-5510	85	27	%	%	NOUN
fcis-5510	85	28	compared	compare	VERB
fcis-5510	85	29	with	with	ADP
fcis-5510	85	30	the	the	DET
fcis-5510	85	31	previous	previous	ADJ
fcis-5510	85	32	pyttsx	pyttsx	NOUN
fcis-5510	85	33	,	,	PUNCT
fcis-5510	85	34	and	and	CCONJ
fcis-5510	85	35	the	the	DET
fcis-5510	85	36	mos	mos	PROPN
fcis-5510	85	37	value	value	NOUN
fcis-5510	85	38	reaches	reach	VERB
fcis-5510	85	39	4.5	4.5	NUM
fcis-5510	85	40	.	.	PUNCT
fcis-5510	86	1	the	the	DET
fcis-5510	86	2	mos	mos	PROPN
fcis-5510	86	3	calculation	calculation	NOUN
fcis-5510	86	4	formula	formula	NOUN
fcis-5510	86	5	is	be	AUX
fcis-5510	86	6	as	as	SCONJ
fcis-5510	86	7	follows	follow	VERB
fcis-5510	86	8	:	:	PUNCT
fcis-5510	86	9	mos	mos	PROPN
fcis-5510	86	10	=	=	SYM
fcis-5510	86	11	∑	∑	PUNCT
fcis-5510	86	12	∑	∑	PROPN
fcis-5510	86	13	𝑆𝑖𝑗	𝑆𝑖𝑗	PROPN
fcis-5510	86	14	𝑁	𝑁	PROPN
fcis-5510	86	15	𝑖=1	𝑖=1	PROPN
fcis-5510	86	16	𝑀	𝑀	PROPN
fcis-5510	86	17	𝑗=1	𝑗=1	PROPN
fcis-5510	86	18	𝑀	𝑀	PROPN
fcis-5510	86	19	(	(	PUNCT
fcis-5510	86	20	11	11	NUM
fcis-5510	86	21	)	)	PUNCT
fcis-5510	86	22	by	by	ADP
fcis-5510	86	23	collecting	collect	VERB
fcis-5510	86	24	the	the	DET
fcis-5510	86	25	100	100	NUM
fcis-5510	86	26	use	use	NOUN
fcis-5510	86	27	cases	case	NOUN
fcis-5510	86	28	outlined	outline	VERB
fcis-5510	86	29	,	,	PUNCT
fcis-5510	86	30	compared	compare	VERB
fcis-5510	86	31	with	with	ADP
fcis-5510	86	32	the	the	DET
fcis-5510	86	33	traditional	traditional	ADJ
fcis-5510	86	34	pyttsx	pyttsx	NOUN
fcis-5510	86	35	speech	speech	NOUN
fcis-5510	86	36	-	-	PUNCT
fcis-5510	86	37	to	to	ADP
fcis-5510	86	38	-	-	PUNCT
fcis-5510	86	39	speech	speech	NOUN
fcis-5510	86	40	algorithm	algorithm	NOUN
fcis-5510	86	41	,	,	PUNCT
fcis-5510	86	42	the	the	DET
fcis-5510	86	43	newly	newly	ADV
fcis-5510	86	44	designed	design	VERB
fcis-5510	86	45	text	text	NOUN
fcis-5510	86	46	-	-	PUNCT
fcis-5510	86	47	to	to	ADP
fcis-5510	86	48	-	-	PUNCT
fcis-5510	86	49	speech	speech	NOUN
fcis-5510	86	50	pronunciation	pronunciation	NOUN
fcis-5510	86	51	accuracy	accuracy	NOUN
fcis-5510	86	52	can	can	AUX
fcis-5510	86	53	reach	reach	VERB
fcis-5510	86	54	95	95	NUM
fcis-5510	86	55	%	%	NOUN
fcis-5510	86	56	,	,	PUNCT
fcis-5510	86	57	most	most	ADJ
fcis-5510	86	58	of	of	ADP
fcis-5510	86	59	the	the	DET
fcis-5510	86	60	use	use	NOUN
fcis-5510	86	61	cases	case	NOUN
fcis-5510	86	62	pronunciation	pronunciation	NOUN
fcis-5510	86	63	standards	standard	NOUN
fcis-5510	86	64	are	be	AUX
fcis-5510	86	65	coordinated	coordinate	VERB
fcis-5510	86	66	with	with	ADP
fcis-5510	86	67	prosody	prosody	NOUN
fcis-5510	86	68	,	,	PUNCT
fcis-5510	86	69	a	a	DET
fcis-5510	86	70	small	small	ADJ
fcis-5510	86	71	number	number	NOUN
fcis-5510	86	72	of	of	ADP
fcis-5510	86	73	prosody	prosody	NOUN
fcis-5510	86	74	and	and	CCONJ
fcis-5510	86	75	pronunciation	pronunciation	NOUN
fcis-5510	86	76	errors	error	NOUN
fcis-5510	86	77	,	,	PUNCT
fcis-5510	86	78	and	and	CCONJ
fcis-5510	86	79	a	a	DET
fcis-5510	86	80	few	few	ADJ
fcis-5510	86	81	use	use	NOUN
fcis-5510	86	82	cases	case	NOUN
fcis-5510	86	83	have	have	VERB
fcis-5510	86	84	large	large	ADJ
fcis-5510	86	85	deviations	deviation	NOUN
fcis-5510	86	86	.	.	PUNCT
fcis-5510	87	1	based	base	VERB
fcis-5510	87	2	on	on	ADP
fcis-5510	87	3	the	the	DET
fcis-5510	87	4	traditional	traditional	ADJ
fcis-5510	87	5	pyttsx	pyttsx	NOUN
fcis-5510	87	6	to	to	ADP
fcis-5510	87	7	speech	speech	NOUN
fcis-5510	87	8	algorithm	algorithm	NOUN
fcis-5510	87	9	,	,	PUNCT
fcis-5510	87	10	the	the	DET
fcis-5510	87	11	algorithm	algorithm	NOUN
fcis-5510	87	12	has	have	VERB
fcis-5510	87	13	a	a	DET
fcis-5510	87	14	large	large	ADJ
fcis-5510	87	15	error	error	NOUN
fcis-5510	87	16	in	in	ADP
fcis-5510	87	17	the	the	DET
fcis-5510	87	18	prosody	prosody	NOUN
fcis-5510	87	19	results	result	NOUN
fcis-5510	87	20	of	of	ADP
fcis-5510	87	21	long	long	ADJ
fcis-5510	87	22	or	or	CCONJ
fcis-5510	87	23	complex	complex	ADJ
fcis-5510	87	24	text	text	NOUN
fcis-5510	87	25	-	-	PUNCT
fcis-5510	87	26	to	to	ADP
fcis-5510	87	27	-	-	PUNCT
fcis-5510	87	28	speech	speech	NOUN
fcis-5510	87	29	.	.	PUNCT
fcis-5510	88	1	as	as	SCONJ
fcis-5510	88	2	shown	show	VERB
fcis-5510	88	3	in	in	ADP
fcis-5510	88	4	table	table	NOUN
fcis-5510	88	5	3	3	NUM
fcis-5510	88	6	.	.	PUNCT
fcis-5510	88	7	table	table	NOUN
fcis-5510	88	8	3	3	NUM
fcis-5510	88	9	.	.	PUNCT
fcis-5510	89	1	i	i	PROPN
fcis-5510	89	2	-	-	PROPN
fcis-5510	89	3	zhrtvc	zhrtvc	PROPN
fcis-5510	89	4	vs.	vs.	ADP
fcis-5510	89	5	pyttsx	pyttsx	PROPN
fcis-5510	89	6	pronunciation	pronunciation	NOUN
fcis-5510	89	7	,	,	PUNCT
fcis-5510	89	8	prosody	prosody	NOUN
fcis-5510	89	9	,	,	PUNCT
fcis-5510	89	10	dictionary	dictionary	ADJ
fcis-5510	89	11	accuracy	accuracy	NOUN
fcis-5510	89	12	comparison	comparison	NOUN
fcis-5510	89	13	results	result	VERB
fcis-5510	89	14	total	total	ADJ
fcis-5510	89	15	number	number	NOUN
fcis-5510	89	16	of	of	ADP
fcis-5510	89	17	test	test	NOUN
fcis-5510	89	18	cases	case	NOUN
fcis-5510	89	19	pronunciation	pronunciation	NOUN
fcis-5510	89	20	accuracy	accuracy	NOUN
fcis-5510	89	21	prosodic	prosodic	ADJ
fcis-5510	89	22	accuracy	accuracy	NOUN
fcis-5510	89	23	dictionary	dictionary	ADJ
fcis-5510	89	24	accuracy	accuracy	NOUN
fcis-5510	90	1	i	i	PRON
fcis-5510	90	2	-	-	PROPN
fcis-5510	90	3	zhrtvc	zhrtvc	PROPN
fcis-5510	90	4	100	100	NUM
fcis-5510	90	5	95	95	NUM
fcis-5510	90	6	%	%	NOUN
fcis-5510	90	7	85	85	NUM
fcis-5510	90	8	%	%	NOUN
fcis-5510	90	9	83	83	NUM
fcis-5510	90	10	%	%	NOUN
fcis-5510	90	11	pyttsx	pyttsx	NOUN
fcis-5510	90	12	100	100	NUM
fcis-5510	90	13	87	87	NUM
fcis-5510	90	14	%	%	NOUN
fcis-5510	90	15	70	70	NUM
fcis-5510	90	16	%	%	NOUN
fcis-5510	90	17	80	80	NUM
fcis-5510	90	18	%	%	NOUN
fcis-5510	90	19	in	in	ADP
fcis-5510	90	20	addition	addition	NOUN
fcis-5510	90	21	,	,	PUNCT
fcis-5510	90	22	our	our	PRON
fcis-5510	90	23	team	team	NOUN
fcis-5510	90	24	's	's	PART
fcis-5510	90	25	i	i	PROPN
fcis-5510	90	26	-	-	PUNCT
fcis-5510	90	27	zhrtvc	zhrtvc	PROPN
fcis-5510	90	28	algorithm	algorithm	NOUN
fcis-5510	90	29	with	with	ADP
fcis-5510	90	30	reference	reference	NOUN
fcis-5510	90	31	to	to	ADP
fcis-5510	90	32	zhrtvc	zhrtvc	PROPN
fcis-5510	90	33	can	can	AUX
fcis-5510	90	34	achieve	achieve	VERB
fcis-5510	90	35	a	a	DET
fcis-5510	90	36	mos	mos	NOUN
fcis-5510	90	37	value	value	NOUN
fcis-5510	90	38	(	(	PUNCT
fcis-5510	90	39	voice	voice	NOUN
fcis-5510	90	40	quality	quality	NOUN
fcis-5510	90	41	evaluation	evaluation	NOUN
fcis-5510	90	42	index	index	NOUN
fcis-5510	90	43	)	)	PUNCT
fcis-5510	90	44	of	of	ADP
fcis-5510	90	45	4.5	4.5	NUM
fcis-5510	90	46	,	,	PUNCT
fcis-5510	90	47	as	as	SCONJ
fcis-5510	90	48	shown	show	VERB
fcis-5510	90	49	in	in	ADP
fcis-5510	90	50	table	table	NOUN
fcis-5510	90	51	4	4	NUM
fcis-5510	90	52	.	.	PUNCT
fcis-5510	90	53	by	by	ADP
fcis-5510	90	54	properly	properly	ADV
fcis-5510	90	55	training	train	VERB
fcis-5510	90	56	the	the	DET
fcis-5510	90	57	gmw	gmw	NOUN
fcis-5510	90	58	vocoder	vocoder	NOUN
fcis-5510	90	59	,	,	PUNCT
fcis-5510	90	60	it	it	PRON
fcis-5510	90	61	sounds	sound	VERB
fcis-5510	90	62	very	very	ADV
fcis-5510	90	63	natural	natural	ADJ
fcis-5510	90	64	,	,	PUNCT
fcis-5510	90	65	and	and	CCONJ
fcis-5510	90	66	the	the	DET
fcis-5510	90	67	dbplus	dbplus	ADJ
fcis-5510	90	68	126	126	NUM
fcis-5510	90	69	algorithm	algorithm	NOUN
fcis-5510	90	70	is	be	AUX
fcis-5510	90	71	used	use	VERB
fcis-5510	90	72	to	to	PART
fcis-5510	90	73	level	level	VERB
fcis-5510	90	74	the	the	DET
fcis-5510	90	75	average	average	ADJ
fcis-5510	90	76	db	db	PROPN
fcis-5510	90	77	of	of	ADP
fcis-5510	90	78	the	the	DET
fcis-5510	90	79	audio	audio	NOUN
fcis-5510	90	80	,	,	PUNCT
fcis-5510	90	81	and	and	CCONJ
fcis-5510	90	82	the	the	DET
fcis-5510	90	83	audio	audio	ADJ
fcis-5510	90	84	time	time	NOUN
fcis-5510	90	85	is	be	AUX
fcis-5510	90	86	used	use	VERB
fcis-5510	90	87	a.	a.	NOUN
fcis-5510	90	88	speed	speed	NOUN
fcis-5510	90	89	is	be	AUX
fcis-5510	90	90	the	the	DET
fcis-5510	90	91	same	same	ADJ
fcis-5510	90	92	duration	duration	NOUN
fcis-5510	90	93	as	as	ADP
fcis-5510	90	94	the	the	DET
fcis-5510	90	95	original	original	ADJ
fcis-5510	90	96	audio	audio	NOUN
fcis-5510	90	97	,	,	PUNCT
fcis-5510	90	98	sounds	sound	VERB
fcis-5510	90	99	complete	complete	ADJ
fcis-5510	90	100	without	without	ADP
fcis-5510	90	101	obvious	obvious	ADJ
fcis-5510	90	102	abnormal	abnormal	ADJ
fcis-5510	90	103	rhythmic	rhythmic	ADJ
fcis-5510	90	104	ups	up	NOUN
fcis-5510	90	105	and	and	CCONJ
fcis-5510	90	106	downs	down	NOUN
fcis-5510	90	107	,	,	PUNCT
fcis-5510	90	108	is	be	AUX
fcis-5510	90	109	relatively	relatively	ADV
fcis-5510	90	110	clear	clear	ADJ
fcis-5510	90	111	and	and	CCONJ
fcis-5510	90	112	smooth	smooth	ADJ
fcis-5510	90	113	,	,	PUNCT
fcis-5510	90	114	and	and	CCONJ
fcis-5510	90	115	is	be	AUX
fcis-5510	90	116	easier	easy	ADJ
fcis-5510	90	117	to	to	PART
fcis-5510	90	118	understand	understand	VERB
fcis-5510	90	119	,	,	PUNCT
fcis-5510	90	120	reaching	reach	VERB
fcis-5510	90	121	the	the	DET
fcis-5510	90	122	quality	quality	NOUN
fcis-5510	90	123	of	of	ADP
fcis-5510	90	124	people	people	NOUN
fcis-5510	90	125	's	's	PART
fcis-5510	90	126	ordinary	ordinary	ADJ
fcis-5510	90	127	conversations	conversation	NOUN
fcis-5510	90	128	,	,	PUNCT
fcis-5510	90	129	and	and	CCONJ
fcis-5510	90	130	listeners	listener	NOUN
fcis-5510	90	131	are	be	AUX
fcis-5510	90	132	willing	willing	ADJ
fcis-5510	90	133	to	to	PART
fcis-5510	90	134	accept	accept	VERB
fcis-5510	90	135	it	it	PRON
fcis-5510	90	136	.	.	PUNCT
fcis-5510	91	1	table	table	NOUN
fcis-5510	91	2	4	4	NUM
fcis-5510	91	3	.	.	PUNCT
fcis-5510	91	4	comparison	comparison	NOUN
fcis-5510	91	5	table	table	NOUN
fcis-5510	91	6	of	of	ADP
fcis-5510	91	7	mos	mos	ADJ
fcis-5510	91	8	values	value	NOUN
fcis-5510	91	9	of	of	ADP
fcis-5510	91	10	various	various	ADJ
fcis-5510	91	11	mainstream	mainstream	NOUN
fcis-5510	91	12	tts	tts	PROPN
fcis-5510	91	13	algorithms	algorithm	NOUN
fcis-5510	91	14	name	name	NOUN
fcis-5510	92	1	i	i	PROPN
fcis-5510	92	2	-	-	PUNCT
fcis-5510	92	3	zhrtvc	zhrtvc	PROPN
fcis-5510	92	4	gtts	gtts	PROPN
fcis-5510	92	5	pyttsx	pyttsx	PROPN
fcis-5510	92	6	win32com	win32com	PROPN
fcis-5510	92	7	mos	mos	PROPN
fcis-5510	92	8	4.5	4.5	NUM
fcis-5510	92	9	4.2	4.2	NUM
fcis-5510	92	10	3.8	3.8	NUM
fcis-5510	92	11	4.0	4.0	NUM
fcis-5510	92	12	4	4	NUM
fcis-5510	92	13	.	.	PUNCT
fcis-5510	93	1	conclusion	conclusion	VERB
fcis-5510	93	2	this	this	DET
fcis-5510	93	3	paper	paper	NOUN
fcis-5510	93	4	studies	study	NOUN
fcis-5510	93	5	video	video	NOUN
fcis-5510	93	6	analysis	analysis	NOUN
fcis-5510	93	7	based	base	VERB
fcis-5510	93	8	on	on	ADP
fcis-5510	93	9	text	text	NOUN
fcis-5510	93	10	error	error	NOUN
fcis-5510	93	11	correction	correction	NOUN
fcis-5510	93	12	from	from	ADP
fcis-5510	93	13	three	three	NUM
fcis-5510	93	14	aspects	aspect	NOUN
fcis-5510	93	15	:	:	PUNCT
fcis-5510	93	16	speech	speech	NOUN
fcis-5510	93	17	segmentation	segmentation	NOUN
fcis-5510	93	18	and	and	CCONJ
fcis-5510	93	19	speech	speech	NOUN
fcis-5510	93	20	to	to	ADP
fcis-5510	93	21	text	text	NOUN
fcis-5510	93	22	,	,	PUNCT
fcis-5510	93	23	text	text	NOUN
fcis-5510	93	24	recognition	recognition	NOUN
fcis-5510	93	25	and	and	CCONJ
fcis-5510	93	26	error	error	NOUN
fcis-5510	93	27	correction	correction	NOUN
fcis-5510	93	28	,	,	PUNCT
fcis-5510	93	29	text	text	NOUN
fcis-5510	93	30	to	to	ADP
fcis-5510	93	31	speech	speech	NOUN
fcis-5510	93	32	and	and	CCONJ
fcis-5510	93	33	video	video	NOUN
fcis-5510	93	34	speech	speech	NOUN
fcis-5510	93	35	synthesis	synthesis	NOUN
fcis-5510	93	36	.	.	PUNCT
fcis-5510	94	1	bayesian	bayesian	NOUN
fcis-5510	94	2	information	information	NOUN
fcis-5510	94	3	criterion	criterion	NOUN
fcis-5510	94	4	bic	bic	NOUN
fcis-5510	94	5	and	and	CCONJ
fcis-5510	94	6	statistical	statistical	ADJ
fcis-5510	94	7	mean	mean	NOUN
fcis-5510	94	8	euclidean	euclidean	ADJ
fcis-5510	94	9	distance	distance	NOUN
fcis-5510	94	10	will	will	AUX
fcis-5510	94	11	be	be	AUX
fcis-5510	94	12	used	use	VERB
fcis-5510	94	13	.	.	PUNCT
fcis-5510	95	1	medist	medist	PROPN
fcis-5510	95	2	's	's	PART
fcis-5510	95	3	phased	phase	VERB
fcis-5510	95	4	efficient	efficient	ADJ
fcis-5510	95	5	algorithm	algorithm	NOUN
fcis-5510	95	6	,	,	PUNCT
fcis-5510	95	7	daec	daec	ADJ
fcis-5510	95	8	algorithm	algorithm	NOUN
fcis-5510	95	9	and	and	CCONJ
fcis-5510	95	10	i	i	PROPN
fcis-5510	95	11	-	-	PUNCT
fcis-5510	95	12	zhrtvc	zhrtvc	PROPN
fcis-5510	95	13	algorithm	algorithm	NOUN
fcis-5510	95	14	are	be	AUX
fcis-5510	95	15	applied	apply	VERB
fcis-5510	95	16	to	to	ADP
fcis-5510	95	17	video	video	NOUN
fcis-5510	95	18	analysis	analysis	NOUN
fcis-5510	95	19	research	research	NOUN
fcis-5510	95	20	based	base	VERB
fcis-5510	95	21	on	on	ADP
fcis-5510	95	22	text	text	NOUN
fcis-5510	95	23	error	error	NOUN
fcis-5510	95	24	correction	correction	NOUN
fcis-5510	95	25	.	.	PUNCT
fcis-5510	96	1	experiments	experiment	NOUN
fcis-5510	96	2	show	show	VERB
fcis-5510	96	3	that	that	SCONJ
fcis-5510	96	4	the	the	DET
fcis-5510	96	5	video	video	NOUN
fcis-5510	96	6	analysis	analysis	NOUN
fcis-5510	96	7	scheme	scheme	NOUN
fcis-5510	96	8	based	base	VERB
fcis-5510	96	9	on	on	ADP
fcis-5510	96	10	text	text	NOUN
fcis-5510	96	11	error	error	NOUN
fcis-5510	96	12	correction	correction	NOUN
fcis-5510	96	13	proposed	propose	VERB
fcis-5510	96	14	and	and	CCONJ
fcis-5510	96	15	designed	design	VERB
fcis-5510	96	16	in	in	ADP
fcis-5510	96	17	this	this	DET
fcis-5510	96	18	paper	paper	NOUN
fcis-5510	96	19	has	have	VERB
fcis-5510	96	20	high	high	ADJ
fcis-5510	96	21	efficiency	efficiency	NOUN
fcis-5510	96	22	and	and	CCONJ
fcis-5510	96	23	accuracy	accuracy	NOUN
fcis-5510	96	24	.	.	PUNCT
fcis-5510	97	1	references	reference	NOUN
fcis-5510	97	2	[	[	X
fcis-5510	97	3	1	1	NUM
fcis-5510	97	4	]	]	PUNCT
fcis-5510	97	5	chen	chen	PROPN
fcis-5510	97	6	lijiang	lijiang	PROPN
fcis-5510	97	7	,	,	PUNCT
fcis-5510	97	8	ren	ren	PROPN
fcis-5510	97	9	jie	jie	PROPN
fcis-5510	97	10	,	,	PUNCT
fcis-5510	97	11	chen	chen	PROPN
fcis-5510	97	12	pengfei	pengfei	PROPN
fcis-5510	97	13	,	,	PUNCT
fcis-5510	97	14	mao	mao	PROPN
fcis-5510	97	15	xia	xia	PROPN
fcis-5510	97	16	,	,	PUNCT
fcis-5510	97	17	zhao	zhao	PROPN
fcis-5510	97	18	qi	qi	PROPN
fcis-5510	97	19	.	.	PROPN
fcis-5510	97	20	limited	limit	VERB
fcis-5510	97	21	text	text	NOUN
fcis-5510	97	22	speech	speech	NOUN
fcis-5510	97	23	synthesis	synthesis	NOUN
fcis-5510	97	24	with	with	ADP
fcis-5510	97	25	electroglottograph	electroglottograph	NOUN
fcis-5510	97	26	based	base	VERB
fcis-5510	97	27	on	on	ADP
fcis-5510	97	28	bilstm	bilstm	NOUN
fcis-5510	97	29	and	and	CCONJ
fcis-5510	97	30	modified	modify	VERB
fcis-5510	97	31	tacotron-2[j	tacotron-2[j	NOUN
fcis-5510	97	32	]	]	PUNCT
fcis-5510	97	33	.	.	PUNCT
fcis-5510	98	1	applied	apply	VERB
fcis-5510	98	2	intelligence,2022,52(13	intelligence,2022,52(13	PROPN
fcis-5510	98	3	)	)	PUNCT
fcis-5510	98	4	.	.	PUNCT
fcis-5510	99	1	[	[	X
fcis-5510	99	2	2	2	NUM
fcis-5510	99	3	]	]	X
fcis-5510	99	4	jin	jin	NOUN
fcis-5510	99	5	-	-	PUNCT
fcis-5510	99	6	song	song	NOUN
fcis-5510	99	7	zhang	zhang	PROPN
fcis-5510	99	8	&	&	CCONJ
fcis-5510	99	9	wen	wen	PROPN
fcis-5510	99	10	cao	cao	PROPN
fcis-5510	99	11	center	center	PROPN
fcis-5510	99	12	for	for	ADP
fcis-5510	99	13	studies	study	NOUN
fcis-5510	99	14	of	of	ADP
fcis-5510	99	15	chinese	chinese	ADJ
fcis-5510	99	16	as	as	ADP
fcis-5510	99	17	a	a	DET
fcis-5510	99	18	second	second	ADJ
fcis-5510	99	19	language	language	NOUN
fcis-5510	99	20	college	college	NOUN
fcis-5510	99	21	of	of	ADP
fcis-5510	99	22	information	information	NOUN
fcis-5510	99	23	sciences	sciences	PROPN
fcis-5510	99	24	beijing	beijing	PROPN
fcis-5510	99	25	language	language	PROPN
fcis-5510	99	26	university	university	PROPN
fcis-5510	99	27	no	no	NOUN
fcis-5510	99	28	.	.	PROPN
fcis-5510	99	29	15	15	NUM
fcis-5510	99	30	,	,	PUNCT
fcis-5510	99	31	road	road	NOUN
fcis-5510	99	32	xueyuan	xueyuan	PROPN
fcis-5510	99	33	,	,	PUNCT
fcis-5510	99	34	haidian	haidian	PROPN
fcis-5510	99	35	,	,	PUNCT
fcis-5510	99	36	beijing	beijing	PROPN
fcis-5510	99	37	100083	100083	NUM
fcis-5510	99	38	,	,	PUNCT
fcis-5510	99	39	p.	p.	PROPN
fcis-5510	99	40	r.	r.	PROPN
fcis-5510	99	41	china	china	PROPN
fcis-5510	99	42	.	.	PUNCT
fcis-5510	100	1	tone	tone	NOUN
fcis-5510	100	2	information	information	NOUN
fcis-5510	100	3	processing	processing	NOUN
fcis-5510	100	4	for	for	ADP
fcis-5510	100	5	chinese	chinese	ADJ
fcis-5510	100	6	automatic	automatic	ADJ
fcis-5510	100	7	speech	speech	NOUN
fcis-5510	100	8	recognition	recognition	NOUN
fcis-5510	100	9	and	and	CCONJ
fcis-5510	100	10	a	a	DET
fcis-5510	100	11	discussion	discussion	NOUN
fcis-5510	100	12	of	of	ADP
fcis-5510	100	13	its	its	PRON
fcis-5510	100	14	application	application	NOUN
fcis-5510	100	15	to	to	ADP
fcis-5510	100	16	computer	computer	NOUN
fcis-5510	100	17	aided	aid	VERB
fcis-5510	100	18	pronunciation	pronunciation	NOUN
fcis-5510	100	19	training	training	NOUN
fcis-5510	100	20	.	.	PUNCT
fcis-5510	101	1	[	[	X
fcis-5510	101	2	3	3	NUM
fcis-5510	101	3	]	]	PUNCT
fcis-5510	101	4	a.	a.	NOUN
fcis-5510	101	5	mouchtaris	mouchtaris	PROPN
fcis-5510	101	6	,	,	PUNCT
fcis-5510	101	7	s.s	s.s	PROPN
fcis-5510	101	8	.	.	PROPN
fcis-5510	101	9	narayanan	narayanan	PROPN
fcis-5510	101	10	,	,	PUNCT
fcis-5510	101	11	c.	c.	PROPN
fcis-5510	101	12	kyriakakis	kyriakakis	PROPN
fcis-5510	101	13	.	.	PUNCT
fcis-5510	102	1	multichannel	multichannel	PROPN
fcis-5510	102	2	audio	audio	ADJ
fcis-5510	102	3	synthesis	synthesis	NOUN
fcis-5510	102	4	by	by	ADP
fcis-5510	102	5	subband	subband	NOUN
fcis-5510	102	6	-	-	PUNCT
fcis-5510	102	7	based	base	VERB
fcis-5510	102	8	spectral	spectral	ADJ
fcis-5510	102	9	conversion	conversion	NOUN
fcis-5510	102	10	and	and	CCONJ
fcis-5510	102	11	parameter	parameter	NOUN
fcis-5510	102	12	adaptation[j	adaptation[j	PROPN
fcis-5510	102	13	]	]	PUNCT
fcis-5510	102	14	.	.	PUNCT
fcis-5510	103	1	ieee	ieee	PROPN
fcis-5510	103	2	/	/	SYM
fcis-5510	103	3	acm	acm	PROPN
fcis-5510	103	4	transactions	transaction	NOUN
fcis-5510	103	5	on	on	ADP
fcis-5510	103	6	audio	audio	ADJ
fcis-5510	103	7	speech	speech	NOUN
fcis-5510	103	8	and	and	CCONJ
fcis-5510	103	9	language	language	NOUN
fcis-5510	103	10	processing,2005,13(2	processing,2005,13(2	NOUN
fcis-5510	103	11	)	)	PUNCT
fcis-5510	103	12	.	.	PUNCT
fcis-5510	104	1	[	[	X
fcis-5510	104	2	4	4	NUM
fcis-5510	104	3	]	]	X
fcis-5510	104	4	mahmut	mahmut	NOUN
fcis-5510	104	5	emilian	emilian	PROPN
fcis-5510	104	6	-	-	PUNCT
fcis-5510	104	7	erman	erman	PROPN
fcis-5510	104	8	,	,	PUNCT
fcis-5510	104	9	nicola	nicola	PROPN
fcis-5510	104	10	stelian	stelian	PROPN
fcis-5510	104	11	,	,	PUNCT
fcis-5510	104	12	stoicu	stoicu	NOUN
fcis-5510	104	13	-	-	PUNCT
fcis-5510	104	14	tivadar	tivadar	PROPN
fcis-5510	104	15	vasile	vasile	NOUN
fcis-5510	104	16	.	.	PUNCT
fcis-5510	105	1	cross	cross	ADJ
fcis-5510	105	2	-	-	NOUN
fcis-5510	105	3	correlation	correlation	NOUN
fcis-5510	105	4	based	base	VERB
fcis-5510	105	5	automated	automate	VERB
fcis-5510	105	6	segmentation	segmentation	NOUN
fcis-5510	105	7	of	of	ADP
fcis-5510	105	8	audio	audio	ADJ
fcis-5510	105	9	samples.[j	samples.[j	PROPN
fcis-5510	105	10	]	]	PUNCT
fcis-5510	105	11	.	.	PUNCT
fcis-5510	106	1	studies	study	NOUN
fcis-5510	106	2	in	in	ADP
fcis-5510	106	3	health	health	NOUN
fcis-5510	106	4	technology	technology	NOUN
fcis-5510	106	5	and	and	CCONJ
fcis-5510	106	6	informatics,2020,272	informatics,2020,272	NOUN
fcis-5510	106	7	.	.	PUNCT
fcis-5510	107	1	[	[	X
fcis-5510	107	2	5	5	NUM
fcis-5510	107	3	]	]	PUNCT
fcis-5510	107	4	dabbabi	dabbabi	PROPN
fcis-5510	107	5	karim	karim	PROPN
fcis-5510	107	6	,	,	PUNCT
fcis-5510	107	7	cherif	cherif	PROPN
fcis-5510	107	8	adnen	adnen	PROPN
fcis-5510	107	9	,	,	PUNCT
fcis-5510	107	10	hajji	hajji	PROPN
fcis-5510	107	11	salah	salah	PROPN
fcis-5510	107	12	.	.	PUNCT
fcis-5510	108	1	an	an	DET
fcis-5510	108	2	optimization	optimization	NOUN
fcis-5510	108	3	of	of	ADP
fcis-5510	108	4	audio	audio	ADJ
fcis-5510	108	5	classification	classification	NOUN
fcis-5510	108	6	and	and	CCONJ
fcis-5510	108	7	segmentation	segmentation	NOUN
fcis-5510	108	8	using	use	VERB
fcis-5510	108	9	gasom	gasom	NOUN
fcis-5510	108	10	algorithm[j	algorithm[j	PROPN
fcis-5510	108	11	]	]	PUNCT
fcis-5510	108	12	.	.	PUNCT
fcis-5510	109	1	international	international	ADJ
fcis-5510	109	2	journal	journal	PROPN
fcis-5510	109	3	of	of	ADP
fcis-5510	109	4	advanced	advanced	ADJ
fcis-5510	109	5	computer	computer	NOUN
fcis-5510	109	6	science	science	NOUN
fcis-5510	109	7	and	and	CCONJ
fcis-5510	109	8	applications	application	NOUN
fcis-5510	109	9	(	(	PUNCT
fcis-5510	109	10	ijacsa),2018,9	ijacsa),2018,9	NOUN
fcis-5510	109	11	.	.	PUNCT