A PyTorch implementation of Speech Transformer [1][2][3], an end-to-end automatic speech recognition with Transformer [4] network, which directly converts acoustic features to character sequence using a single nueral network.

• Python3 (recommend Anaconda)
• PyTorch 0.4.1+
• Kaldi (just for feature extraction)
• pip install -r requirements.txt
• cd tools; make KALDI=/path/to/kaldi
• If you want to run egs/aishell/run.sh, download aishell dataset for free.

$ cd egs/aishell
# Modify aishell data path to your path in the begining of run.sh 
$ bash run.sh

That's all!

You can change parameter by $ bash run.sh --parameter_name parameter_value, egs, $ bash run.sh --stage 3. See parameter name in egs/aishell/run.sh before . utils/parse_options.sh.

Workflow of egs/aishell/run.sh:

• Stage 0: Data Preparation
• Stage 1: Feature Generation
• Stage 2: Dictionary and Json Data Preparation
• Stage 3: Network Training
• Stage 4: Decoding

egs/aishell/run.sh provide example usage.

# Set PATH and PYTHONPATH
$ cd egs/aishell/; . ./path.sh
# Train
$ train.py -h
# Decode
$ recognize.py -h

If you want to visualize your loss, you can use visdom to do that:

1. Open a new terminal in your remote server (recommend tmux) and run $ visdom.
2. Open a new terminal and run $ bash run.sh --visdom 1 --visdom_id "<any-string>" or $ train.py ... --visdom 1 --vidsdom_id "<any-string>".
3. Open your browser and type <your-remote-server-ip>:8097, egs, 127.0.0.1:8097.
4. In visdom website, chose <any-string> in Environment to see your loss.

$ bash run.sh --continue_from <model-path>

When happened in training, try to reduce batch_size. $ bash run.sh --batch_size <lower-value>.

• [1] Linhao Dong, Shuang Xu,and Bo Xu. “Speech-transformer:A no-recurrence sequence-to-sequence model for speech recognition” in ICASSP 2018
• [2] Shiyu Zhou, Linhao Dong, et al. “Syllable-based sequence-to-sequence speech recognition with the transformer in mandarin chinese” in Interspeech 2018
• [3] Shiyu Zhou, Linhao Dong, et al. “A comparison of modeling units in sequence-to-sequence speech recognition with the transformer on mandarin chinese” arXiv preprint arXiv:1805.06239
• [4] Ashish Vaswani, Noam Shazeer, et al. “Attention is all you need” in NIPS 2017