/Cross-Lingual-Voice-Cloning

Tacotron 2 - PyTorch implementation with faster-than-realtime inference modified to enable cross lingual voice cloning.

Primary LanguageJupyter NotebookBSD 3-Clause "New" or "Revised" LicenseBSD-3-Clause

Tacotron 2 (without wavenet)

DISCLAIMER :- The following code base has been modified according to the paper Learning to Speak Fluently in a Foreign Language:Multilingual Speech Synthesis and Cross-Language Voice Cloning

Dataset Format

The model needs to be provided 2 text files 1 for the purpose of training and 1 for validation. Each line of the txt file should follow the following format :-

<path-to-wav-file>|<text-corresponding-to-speech-in-wav>|<speaker-no>|<lang-no>

<speaker-no> goes from 0 to n-1, where n is the number of speakers.

<lang-no> goes from 0 to m-1 , where m is the number of languages.

Hparams

hparams.training_files, hparams.validation_files need to be set to the path to the txt files of previous section.

hparams.n_speakers, hparams.dim_yo need to be changed to the number of speakers.

hparams.n_langs must be set to number of languages.

To change the languages, add/remove unicode characters in _letters variable of text/symbols.py .

TODO

See Projects

Old Readme.md

PyTorch implementation of Natural TTS Synthesis By Conditioning Wavenet On Mel Spectrogram Predictions.

This implementation includes distributed and automatic mixed precision support and uses the LJSpeech dataset.

Distributed and Automatic Mixed Precision support relies on NVIDIA's Apex and AMP.

Visit our website for audio samples using our published Tacotron 2 and WaveGlow models.

Alignment, Predicted Mel Spectrogram, Target Mel Spectrogram

Pre-requisites

  1. NVIDIA GPU + CUDA cuDNN

Setup

  1. Download and extract the LJ Speech dataset
  2. Clone this repo: git clone https://github.com/NVIDIA/tacotron2.git
  3. CD into this repo: cd tacotron2
  4. Initialize submodule: git submodule init; git submodule update
  5. Update .wav paths: sed -i -- 's,DUMMY,ljs_dataset_folder/wavs,g' filelists/*.txt
    • Alternatively, set load_mel_from_disk=True in hparams.py and update mel-spectrogram paths
  6. Install PyTorch 1.0
  7. Install Apex
  8. Install python requirements or build docker image
    • Install python requirements: pip install -r requirements.txt

Training

  1. python train.py --output_directory=outdir --log_directory=logdir
  2. (OPTIONAL) tensorboard --logdir=outdir/logdir

Training using a pre-trained model

Training using a pre-trained model can lead to faster convergence
By default, the dataset dependent text embedding layers are ignored

  1. Download our published Tacotron 2 model
  2. python train.py --output_directory=outdir --log_directory=logdir -c tacotron2_statedict.pt --warm_start

Multi-GPU (distributed) and Automatic Mixed Precision Training

  1. python -m multiproc train.py --output_directory=outdir --log_directory=logdir --hparams=distributed_run=True,fp16_run=True

Inference demo

  1. Download our published Tacotron 2 model
  2. Download our published WaveGlow model
  3. jupyter notebook --ip=127.0.0.1 --port=31337
  4. Load inference.ipynb

N.b. When performing Mel-Spectrogram to Audio synthesis, make sure Tacotron 2 and the Mel decoder were trained on the same mel-spectrogram representation.

Related repos

WaveGlow Faster than real time Flow-based Generative Network for Speech Synthesis

nv-wavenet Faster than real time WaveNet.

Acknowledgements

This implementation uses code from the following repos: Keith Ito, Prem Seetharaman as described in our code.

We are inspired by Ryuchi Yamamoto's Tacotron PyTorch implementation.

We are thankful to the Tacotron 2 paper authors, specially Jonathan Shen, Yuxuan Wang and Zongheng Yang.