VITS TTS
The end-to-end TTS architecture that combines a variational autoencoder, normalizing flows, and adversarial training.
Fi àkọlé rẹ pamọ́ sí àwọn àmì-ìwé SSML fún ìdáràn:
<speak><prosody rate="slow">Slow speech</prosody></speak>
Àwọn Àmì-ìwé tí àwọn ìṣàmúlò-ètò tí a yàn gbọ́ - tẹ̀ láti fi ọkan sínú àkọ́lé rẹ̀ nínú àwọn ààyè-iṣẹ́ tí o bá jẹ́:
Àwọn àwọn àkọlé àwọn ààyè-iṣẹ́ àwọn àwọn àmì-ìwé àwọn àmì-ìwé àwọn àwọn àmì-ìwé àwọn à
Àwọn àwọn ìṣàfarawé àwọn àwọn ìṣàfarawé àwọn (ọrọ = ìṣàfàlì):
Ààyè-iṣẹ́ VITS
VITS — Variational Inference with adversarial learning for end-to-end Text-to-Speech — was introduced by Jaehyeon Kim and collaborators in 2021 and became a foundational architecture for modern neural speech. Rather than the older two-stage pipeline, it synthesizes audio in a single parallel end-to-end pass, pairing a variational autoencoder with normalizing flows and a GAN-style adversarial training process to lift naturalness. At about 25M parameters and trained on ~585 hours, it produces natural prosody at fast inference speeds and supports multiple speakers. It serves as a solid general-purpose, free baseline and underpins many later models such as Piper and MeloTTS.
Tí o dara jù fún: General-purpose text-to-speech with natural prosody
Wá Gbogbo àwòrán VITS Àwọn àwòránNínú àwọn ìṣàfarawé
- Àwọn Àkọlé
- Jaehyeon Kim et al.
- Àwọn Ààyè-iṣẹ́
- MIT
- Àwọn àwọn ààyè-iṣẹ́
- free
- Ìjánu-ìṣàmúlò-ètò
- fast
- Ìṣàfarawé àwọn àmì-ìwé
- Àwọn àwọn àgbéwọlé
- Àwọn
- English, German, Spanish, French, Portuguese, Dutch, Finnish, Hungarian, Bulgarian, Japanese, Polish
- Àwọn àyọkà ìpele
- 2000