VITS TTS
The end-to-end TTS architecture that combines a variational autoencoder, normalizing flows, and adversarial training.
Ampidiro anatin'ny tag SSML ny lahabolana mba hahazoana fifehezana mazava tsara:
<speak><prosody rate="slow">Slow speech</prosody></speak>
Tag fantatry ny modely voafaritra — tsindrio mba hametrahana iray ao anatin'ny lahatsoratrao izay misy azy:
Mamakiana lahabolana tsotra io modely io, ka tsy raharahaina ny tag anatin'ny andalana. Raha mila fihetseham-po mifototra amin'ny tag ianao, dia miova ho modely maneho fihetseham-po toy ny Orpheus na Bark.
Mamaritra ny fanononana safidy (teny = fanononana):
Mombamomba 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.
Tsara indrindra ho an'ny: General-purpose text-to-speech with natural prosody
Jereo izy rehetra VITS feoAmin'ny fijery fohy
- Mpamorona
- Jaehyeon Kim et al.
- Lisansa
- MIT
- Taona
- free
- Hafainganan'ny fanovana
- fast
- Fandraisana feo
- Tsy misy
- Teny
- English, German, Spanish, French, Portuguese, Dutch, Finnish, Hungarian, Bulgarian, Japanese, Polish
- Marika betsaka indrindra
- 2000