Apply FastSpeech 2 model to Vietnamese TTS

Dataset

• Infore: a single speaker Vietnamese dataset with 14935 short audio clips of a female speaker
• Download and extract files into ./raw_data/infore/

Montreal Forced Aligner

• Recommended version: 2.0.6

Preprocess data and train model

• Do step by step according to scripts included in ./scripts/infore/
• Alignment of the dataset Infore at here Infore's alignment : download and extract into ./proprocessed_data/infore/

Pretrained model

• Download and extract pretrained model into ./output/ckpt/infore/

Inferrence

python3 synthesize.py --text "YOUR_DESIRED_TEXT" --restore_step 100000 --mode single -p config/infore/preprocess.yaml -m config/infore/model.yaml -t config/infore/train.yaml

Output samples

• Ánh nắng cuối cùng luyến tiếc rời bỏ ngàn lá xanh bên kia bờ, vạn vật trở nên buồn buồn trong bóng hoàng hôn.

sample1.mov

• Trên không, vài con cò về tổ trễ đập nhanh đôi cánh trắng phau rồi khuất trong lùm cây rậm lá. Những đám mây trắng đá ngả màu ngà, bầu trời xanh cũng đã ngả sang màu sậm đưa đến màu đen. Đâu đó có tiếng chim lẻ bạn, tiếng dơi muỗi lào xào lẫn trong tiếng gió nhẹ lay cành. Dưới bến sông, con nước ròng lên đầy mé đã đứng lại không lùa được những đợt lục bình lờ lững giữa dòng ra sông cái. Dòng nước xanh chìm đi trong màu xám sậm và những bóng cây bên bờ kia ngả xuống dòng càng lúc càng hiện rõ lù lù thành hàng trong bóng nước.

sample2.mov

• Mặt trăng tròn vành vạnh từ từ nhô lên sau luỹ tre. Trăng đêm nay sáng quá! Bầu trời điểm xuyết một vài ngôi sao lấp lánh như những con đom đóm nhỏ. Ánh trăng tàng dịu mát tỏa xuống, chảy loang lổ trên mặt đất, trên các cành cây ngọn cỏ. Không gian mới yên tĩnh làm sao! Chỉ còn tiếng sương đêm rơi lốp bốp trên lá cây và tiếng côn trùng ra rả trong đất ẩm . Chị gió chuyên cần nhẹ nhàng bay làm rung mấy ngọn xà cừ ven đường. Thoang thoảng đâu đây mùi hoa thiên lí dịu dàng lan tỏa. Đêm trăng thật đẹp và êm đềm.

sample3.mov

Training loss

FastSpeech 2 - PyTorch Implementation

This is a PyTorch implementation of Microsoft's text-to-speech system FastSpeech 2: Fast and High-Quality End-to-End Text to Speech. This project is based on xcmyz's implementation of FastSpeech. Feel free to use/modify the code.

There are several versions of FastSpeech 2. This implementation is more similar to version 1, which uses F0 values as the pitch features. On the other hand, pitch spectrograms extracted by continuous wavelet transform are used as the pitch features in the later versions.

Updates

• 2021/7/8: Release the checkpoint and audio samples of a multi-speaker English TTS model trained on LibriTTS
• 2021/2/26: Support English and Mandarin TTS
• 2021/2/26: Support multi-speaker TTS (AISHELL-3 and LibriTTS)
• 2021/2/26: Support MelGAN and HiFi-GAN vocoder

Audio Samples

Audio samples generated by this implementation can be found here.

Quickstart

Dependencies

You can install the Python dependencies with

pip3 install -r requirements.txt

Inference

You have to download the pretrained models and put them in output/ckpt/LJSpeech/, output/ckpt/AISHELL3, or output/ckpt/LibriTTS/.

For English single-speaker TTS, run

python3 synthesize.py --text "YOUR_DESIRED_TEXT" --restore_step 900000 --mode single -p config/LJSpeech/preprocess.yaml -m config/LJSpeech/model.yaml -t config/LJSpeech/train.yaml

For Mandarin multi-speaker TTS, try

python3 synthesize.py --text "大家好" --speaker_id SPEAKER_ID --restore_step 600000 --mode single -p config/AISHELL3/preprocess.yaml -m config/AISHELL3/model.yaml -t config/AISHELL3/train.yaml

For English multi-speaker TTS, run

python3 synthesize.py --text "YOUR_DESIRED_TEXT"  --speaker_id SPEAKER_ID --restore_step 800000 --mode single -p config/LibriTTS/preprocess.yaml -m config/LibriTTS/model.yaml -t config/LibriTTS/train.yaml

For Vietnamese single-speaker TTS, run

python3 synthesize.py --text "YOUR_DESIRED_TEXT" --restore_step 100000 --mode single -p config/infore/preprocess.yaml -m config/infore/model.yaml -t config/infore/train.yaml

The generated utterances will be put in output/result/.

Here is an example of synthesized mel-spectrogram of the sentence "Printing, in the only sense with which we are at present concerned, differs from most if not from all the arts and crafts represented in the Exhibition", with the English single-speaker TTS model.

Batch Inference

Batch inference is also supported, try

python3 synthesize.py --source preprocessed_data/LJSpeech/val.txt --restore_step 900000 --mode batch -p config/LJSpeech/preprocess.yaml -m config/LJSpeech/model.yaml -t config/LJSpeech/train.yaml

to synthesize all utterances in preprocessed_data/LJSpeech/val.txt

Controllability

The pitch/volume/speaking rate of the synthesized utterances can be controlled by specifying the desired pitch/energy/duration ratios. For example, one can increase the speaking rate by 20 % and decrease the volume by 20 % by

python3 synthesize.py --text "YOUR_DESIRED_TEXT" --restore_step 900000 --mode single -p config/LJSpeech/preprocess.yaml -m config/LJSpeech/model.yaml -t config/LJSpeech/train.yaml --duration_control 0.8 --energy_control 0.8

Training

Datasets

The supported datasets are

• LJSpeech: a single-speaker English dataset consists of 13100 short audio clips of a female speaker reading passages from 7 non-fiction books, approximately 24 hours in total.
• AISHELL-3: a Mandarin TTS dataset with 218 male and female speakers, roughly 85 hours in total.
• LibriTTS: a multi-speaker English dataset containing 585 hours of speech by 2456 speakers.
• Infore: a single speaker Vietnamese dataset with 14935 short audio clips of a female speaker

We take LJSpeech as an example hereafter.

Preprocessing

First, run

python3 prepare_align.py config/LJSpeech/preprocess.yaml

for some preparations.

As described in the paper, Montreal Forced Aligner (MFA) is used to obtain the alignments between the utterances and the phoneme sequences. Alignments of the supported datasets are provided here. You have to unzip the files in preprocessed_data/LJSpeech/TextGrid/.

After that, run the preprocessing script by

python3 preprocess.py config/LJSpeech/preprocess.yaml

Alternately, you can align the corpus by yourself. Download the official MFA package and run

./montreal-forced-aligner/bin/mfa_align raw_data/LJSpeech/ lexicon/librispeech-lexicon.txt english preprocessed_data/LJSpeech

or

./montreal-forced-aligner/bin/mfa_train_and_align raw_data/LJSpeech/ lexicon/librispeech-lexicon.txt preprocessed_data/LJSpeech

to align the corpus and then run the preprocessing script.

python3 preprocess.py config/LJSpeech/preprocess.yaml

Training

Train your model with

python3 train.py -p config/LJSpeech/preprocess.yaml -m config/LJSpeech/model.yaml -t config/LJSpeech/train.yaml

The model takes less than 10k steps (less than 1 hour on my GTX1080Ti GPU) of training to generate audio samples with acceptable quality, which is much more efficient than the autoregressive models such as Tacotron2.

TensorBoard

Use

tensorboard --logdir output/log/LJSpeech

to serve TensorBoard on your localhost. The loss curves, synthesized mel-spectrograms, and audios are shown.

Implementation Issues

• Following xcmyz's implementation, I use an additional Tacotron-2-styled Post-Net after the decoder, which is not used in the original FastSpeech 2.
• Gradient clipping is used in the training.
• In my experience, using phoneme-level pitch and energy prediction instead of frame-level prediction results in much better prosody, and normalizing the pitch and energy features also helps. Please refer to config/README.md for more details.

Please inform me if you find any mistakes in this repo, or any useful tips to train the FastSpeech 2 model.

References

• FastSpeech 2: Fast and High-Quality End-to-End Text to Speech, Y. Ren, et al.
• xcmyz's FastSpeech implementation
• TensorSpeech's FastSpeech 2 implementation
• rishikksh20's FastSpeech 2 implementation

Citation

@INPROCEEDINGS{chien2021investigating,
  author={Chien, Chung-Ming and Lin, Jheng-Hao and Huang, Chien-yu and Hsu, Po-chun and Lee, Hung-yi},
  booktitle={ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)}, 
  title={Investigating on Incorporating Pretrained and Learnable Speaker Representations for Multi-Speaker Multi-Style Text-to-Speech}, 
  year={2021},
  volume={},
  number={},
  pages={8588-8592},
  doi={10.1109/ICASSP39728.2021.9413880}}