ba-dls-deepspeech

Train your own CTC model!

Table of Contents

1. Dependencies
2. Data
3. Running an example

Dependencies

You will need the following packages installed before you can train a model using this code. You may have to change PYTHONPATH to include the directories of your new packages.

1. theano
   The underlying deep learning Python library. We suggest using the bleeding edge version through
   git clone https://github.com/Theano/Theano
   Follow the instructions on: http://deeplearning.net/software/theano/install.html#bleeding-edge-install-instructions
   Or, simply: cd Theano; python setup.py install

2. keras
   This is a wrapper over Theano that provides nice functions for building networks. Once again, we suggest using the bleeding edge version. Make sure you install it with support for hdf5 - we make use of that to save models.
   git clone https://github.com/fchollet/keras
   Follow the installation instructions on https://github.com/fchollet/keras
   Or, simply: cd keras; python setup.py install

3. warp-ctc
   This contains the main implementation of the CTC cost function.
   git clone https://github.com/baidu-research/warp-ctc
   To install it, follow the instructions on https://github.com/baidu-research/warp-ctc

4. theano-warp-ctc
   This is a theano wrapper over warp-ctc.
   git clone https://github.com/sherjilozair/ctc
   Follow the instructions on https://github.com/sherjilozair/ctc for installation.

5. Others
   You may require some additional packages. Install Python requirements through pip as:
   pip install soundfile
   On Ubuntu, avconv (used here for audio format conversions) requires libav-tools.
   sudo apt-get install libav-tools

Data

We will make use of the LibriSpeech ASR corpus to train our models. Use the download.sh script to download this corpus (~65GB). Use flac_to_wav.sh to convert any flac files to wav.
We make use of a JSON file that aggregates all data for training, validation and testing. Once you have a corpus, create a description file that is a json-line file in the following format:

{"duration": 15.685, "text": "spoken text label", "key": "/home/username/LibriSpeech/train-clean-360/5672/88367/5672-88367-0031.wav"}
{"duration": 14.32, "text": "ground truth text", "key": "/home/username/LibriSpeech/train-other-500/8678/280914/8678-280914-0009.wav"}

Each line is a JSON. We will make use of the durations to construct a curriculum in the first epoch (shorter utterances are easier).
You can query the duration of a file using: soxi -D filename. By default, we split this data as 80%: training, 10%: validation and 10%: testing. You can play around with these in data_generator.py

Running an example

Finally, let's train a model!
python train.py corpus.json ./save_my_model_here
This will checkpoint a model every few iterations into the directory you specify.