VITS TTS
The end-to-end TTS architecture that combines a variational autoencoder, normalizing flows, and adversarial training.
Ajusta el text a les etiquetes SSML per al control precís:
<speak><prosody rate="slow">Slow speech</prosody></speak>
Etiquetes del model seleccionat entenen el clic show clic per a deixar- ne un al text a on succeeix:
Aquest model llegeix text pla, així que les etiquetes inserides s' ignoren. Per a emocions basades en etiquetes, canvieu a un model expressiu com Orfeus o Bark.
Defineix pronúncies personalitzades (word = pronunciació):
Quant a 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.
Millor per: General-purpose text-to-speech with natural prosody
Navega- ho tot VITS veusEn una mirada
- Desenvolupador
- Jaehyeon Kim et al.
- Llicència
- MIT
- TierCity name (optional, probably does not need a translation)
- free
- Velocitat
- fast
- clonació de veu
- No
- Idiomes
- English, German, Spanish, French, Portuguese, Dutch, Finnish, Hungarian, Bulgarian, Japanese, Polish
- Nombre màxim de caràcters
- 2000