VITS TTS
The end-to-end TTS architecture that combines a variational autoencoder, normalizing flows, and adversarial training.
Wrap uw tekst in SSML-tags voor nauwkeurige controle:
<speak><prosody rate="slow">Slow speech</prosody></speak>
Tags het geselecteerde model begrijpt
Dit model leest platte tekst, dus inline tags worden genegeerd. Voor emotie op basis van tags, schakel naar een expressief model zoals Orpheus of Bark.
Definieer aangepaste uitspraaken (woord = uitspraak):
Info 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.
Beste voor: General-purpose text-to-speech with natural prosody
Alles doorbladeren VITS stemmenIn een oogopslag
- Ontwikkelaar
- Jaehyeon Kim et al.
- Licentie
- MIT
- Niveau
- free
- Snelheid
- fast
- Klonen van stemmen
- Nee
- Talen
- English, German, Spanish, French, Portuguese, Dutch, Finnish, Hungarian, Bulgarian, Japanese, Polish
- Max. tekens
- 2000