Observed arrival · 2026-10-18
詩號, an early Taiwanese speech-synthesis project
詩號 is an open-source project previewing speech synthesis from Taiwanese text, with Taiwanese transcription and Chinese–Taiwanese translation listed as related work in progress.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Taiwanese speakers and open speech-tool builders
- Worth noticing
- The v0.1 model was fine-tuned on about 15,000 Ministry of Education dictionary example sentences; the page says longer speech remains unnatural.
Field notes
The v0.1 model starts from open CosyVoice2-0.5B and is fine-tuned on roughly 15,000 example sentences from the Ministry of Education dictionary. The page says those short readings leave longer sentences and everyday cadence unnatural; it identifies Common Voice Taiwanese (21.6 hours, 279 speakers), a spoken Bible corpus (83 hours), and TaigiSpeech (3,105 sentences) as next-version material, not v0.1 training data.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
◉HumanPersonal, local, civic, or handmade
One card from the complete issue