tts_impl

pytorch implementation of speech synthesis methods

Models

implemented: ✅
work in progress: 🚧

• Vocoders
  • HiFi-GAN ✅
  • NSF-HiFi-GAN ✅
  • ISTFTNet
  • WaveNeXt 🚧
  • Vocos 🚧
  • DDSP (Subtractive / Additive)
  • SoundStream
• TTS: Text to speech
  • VITS ✅
  • VITS2 🚧
  • JETS 🚧
  • FastSpeech
  • Diffusion TTS Models
  • Flow Matching TTS Modelss
• Aligner:
  • ForcedAlign ✅

Installation

pip install git+https://github.com/uthree/tts_impl.git

or clone this repository,

pip install -e .

type check / lint / test

• pysen run format : automatic formatting
• pysen run lint : linter
• pytest : test
• codespell : detect typos

References

• VITS (Paper)
• VITS2 (Paper)
• HiFi-GAN (Paper)
• Super Monotonic Alignment Search (Paper)
• JETS (ESPNet)