v0.2.0

[erogol]

🐸 v0.2.0

🐞Bug Fixes

• Fix phoneme pre-compute issue.
• Fix multi-speaker setup in Tacotron models.
• Fix small issues in the Trainer regarding multi-optimizer training.

💾 Code updates

• W&B integration for model logging and experiment tracking, (👑 @AyushExel)
  Code uses the Tensorboard by default. For W&B, you need to set log_dashboard option in the config and define project_name and wandb_entity.
• Use ffsspec for model saving/loading (👑 @agrinh)
• Allow models to define their own symbol list with in-class make_symbols()
• Allow choosing after epoch or after step LR scheduler update with scheduler_after_epoch.
• Make converting spectrogram from amplitude to DB optional with do_amp_to_db_linear and do_amp_to_db_linear options.

🗒️ Docs updates

• Add GlowTTS and VITS docs.

🤖 Model implementations

• VITS implementation with pre-trained models (https://arxiv.org/abs/2106.06103)

🚀 Model releases

• vocoder_models--ja--kokoro--hifigan_v1 (👑 @kaiidams)

  HiFiGAN model trained on Kokoro dataset to complement the existing Japanese model.

  Try it out:

  tts --model_name tts_models/ja/kokoro/tacotron2-DDC --text "こんにちは、今日はいい天気ですか？"

• tts_models--en--ljspeech--tacotronDDC_ph

  TacotronDDC with phonemes trained on LJSpeech. It is to fix the pronunciation errors caused by the raw text
  in the released TacotronDDC model.

  Try it out:

  tts --model_name tts_models/en/ljspeech/tacotronDDC_ph --text "hello, how are you today?"

• tts_models--en--ljspeech--vits

  VITS model trained on LJSpeech.

  Try it out:

  tts --model_name tts_models/en/ljspeech/vits --text "hello, how are you today?"

• tts_models--en--vctk--vits

  VITS model trained on VCTK with multi-speaker support.

  Try it out:

  tts-server --model_name tts_models/en/vctk/vits     

• vocoder_models--en--ljspeech--univnet

  UnivNet model trained on LJSpeech to complement the TacotronDDC model above.

  Try it out:

  tts --model_name tts_models/en/ljspeech/tacotronDDC_ph --text "hello, how are you today?"

[godspirit00]

Thanks for the great work!
I have a question. I trained a TTS model in a previous version, and I noticed some pronunciation errors in the generated speech. Now that version 0.2.0 fixed it, do I have to retrain the model from start? Is it ok to continue training from the last checkpoint to fix the problem?
Thanks!

[erogol]

probably enough to finetune

[Sadam1195]

Just a suggestion, it would be nice to have some information on models released in every version. Like what dataset was used to train, number of steps, GPU resources used (time taken : a bit irrelevant but gives some idea about what to expect). Sometimes it is self explanatory with the model name but nonetheless, it would be good to have official statement on it.

[BillyBobQuebec]

I second this! Nvidia model catalog for example has very helpful details about what gpu(s) and batch size was used to train a model and what loss the model was at within different points of training, as well as inference speed on gpu vs cpu...

[erogol]

The info I wrote is from the paper. Depending on your local settings can things change. But I've not run apple to apple comparison personally.

Model cards are hard to maintain right now for a library in v0.2.0. But we plan to set it up when things are more stable.