id	author	title	date	pages	extension	mime	words	sentence	flesch	summary	cache	txt
bitcs-10314	Imani, Aina Avrilia; Rahman, Aviv Yuniar; Nurdiyansyah, Firman	Development of a Deep Learning-Based Text-To-Speech System for the Malang Walikan Language Using the Pre-Trained SpeechT5 and Hifi-GAN Models: Author's Country: Indonesia	2025	10	.pdf	application/pdf	3952	182	52	However, the use of pre-trained models directly (zero-shot) in low-resource regional languages, particularly Malang Walikan, is still rarely researched. Test Sentence WER SpeechT5 + HiFi-GAN WER Voice Female WER Voice Male info lokasi dong nawak 0.5 0.25 0.25 sesuai dengan gambar di bis halokes 0.5 0.17 0.33 wah umak lihai juga berbahasa arudam 0.67 0.5 0.83 rame ilakes area kayutangan di malam minggu 0.71 0.29 0.71 sekali nade tetep nade 0.75 0.75 0.75 agomes lancar rejeki hari ini 0.8 0.6 0.4 nakam lah mbah 1.0 0.33 1.0 umak ngalam 1.0 0.5 1.0 jenenge ae ongis nade yo nade temenan 1.0 0.57 0.71 Table 2 shows examples of WER values for the three audio types.	cache/bitcs-10314.pdf	txt/bitcs-10314.txt