TTS

Introduction

This is a public domain speech dataset consisting of 13,100 short audio clips of a single speaker reading passages from 7 non-fiction books. A transcription is provided for each clip. Clips vary in length from 1 to 10 seconds and have a total length of approximately 24 hours.

The texts were published between 1884 and 1964, and are in the public domain. The audio was recorded in 2016-17 by the LibriVox project and is also in the public domain.

The above information is from the LJSpeech website.

VITS

This recipe provides a VITS model trained on the LJSpeech dataset.

Pretrained model can be found here.

For tutorial and more details, please refer to the VITS documentation.

The training command is given below:

export CUDA_VISIBLE_DEVICES=0,1,2,3
./vits/train.py \
  --world-size 4 \
  --num-epochs 1000 \
  --start-epoch 1 \
  --use-fp16 1 \
  --exp-dir vits/exp \
  --max-duration 500

To inference, use:

./vits/infer.py \
  --exp-dir vits/exp \
  --epoch 1000 \
  --tokens data/tokens.txt

Quality vs speed

If you feel that the trained model is slow at runtime, you can specify the argument --model-type during training. Possible values are:

• low, means low quality. The resulting model is very small in file size and runs very fast. The following is a wave file generatd by a low quality model

  low.mp4

  The text is Ask not what your country can do for you; ask what you can do for your country.

  The exported onnx model has a file size of 26.8 MB (float32).

• medium, means medium quality. The following is a wave file generatd by a medium quality model

  medium.mp4

  The text is Ask not what your country can do for you; ask what you can do for your country.

  The exported onnx model has a file size of 70.9 MB (float32).

• high, means high quality. This is the default value.

  The following is a wave file generatd by a high quality model

  high.mp4

  The text is Ask not what your country can do for you; ask what you can do for your country.

  The exported onnx model has a file size of 113 MB (float32).

A pre-trained low model trained using 4xV100 32GB GPU with the following command can be found at https://huggingface.co/csukuangfj/icefall-tts-ljspeech-vits-low-2024-03-12

export CUDA_VISIBLE_DEVICES=0,1,2,3
./vits/train.py \
  --world-size 4 \
  --num-epochs 1601 \
  --start-epoch 1 \
  --use-fp16 1 \
  --exp-dir vits/exp \
  --model-type low \
  --max-duration 800

A pre-trained medium model trained using 4xV100 32GB GPU with the following command can be found at https://huggingface.co/csukuangfj/icefall-tts-ljspeech-vits-medium-2024-03-12

export CUDA_VISIBLE_DEVICES=4,5,6,7
./vits/train.py \
  --world-size 4 \
  --num-epochs 1000 \
  --start-epoch 1 \
  --use-fp16 1 \
  --exp-dir vits/exp-medium \
  --model-type medium \
  --max-duration 500

# (Note it is killed after `epoch-820.pt`)