VITS TTS
The end-to-end TTS architecture that combines a variational autoencoder, normalizing flows, and adversarial training.
Envolva o seu texto em tags SSML para controle preciso:
<speak><prosody rate="slow">Slow speech</prosody></speak>
Etiquetas o modelo selecionado entende — clique para soltar um para o seu texto onde acontece:
Este modelo lê texto simples, por isso as etiquetas inline são ignoradas. Para emoção baseada em tags, mude para um modelo expressivo como Orpheus ou Bark.
Definir pronúncias personalizadas (palavra = pronúncia):
Sobre 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.
Melhor para: General-purpose text-to-speech with natural prosody
Procurar todos VITS vozesDe uma olhada
- Desenvolvedor
- Jaehyeon Kim et al.
- Licença
- MIT
- Tier
- free
- Velocidade
- fast
- Clonagem de voz
- Não
- Línguas
- English, German, Spanish, French, Portuguese, Dutch, Finnish, Hungarian, Bulgarian, Japanese, Polish
- Número máximo de caracteres
- 2000