GIPFA

This repository is the official implementation of GIPFA: Generating IPA Pronunciation from Audio.

Requirements

• Require Ubuntu 18.04, Python 3.6+.

• Download files on your machine

  • git clone https://github.com/marxav/gipfa.git

• Go to the gipfa main directory

  • cd gipfa

• Create a virtual environment

  • python3 -m venv gipfa

• Activate virtual environment

  • source gipfa/bin/activate

• Load the python librairies needed for GIPFA (e.g. numpy, pandas, torch...) from the requirements file

  • python -m pip install --upgrade pip
  • python -m pip install --upgrade setuptools
  • python -m pip install -r requirements.txt

• Download the dataset (size=1.2Go)

  • python3 download_dataset.py

Evaluation and Training

In order to run this code, you need to:

• Run the gipfa.ipynb in order to create the ANN model, train it, test it and display the results of the paper.

Results

GIPFA achieves the following performance:

Tested samples	Mean Accuracy
1000	75.0

The detailed performance is availaible in this CSV file (with '\t' as a separator)

License

GIPFA is released under the CC BY-SA 4.0.

Citation

The pipeline is described in the following paper:

@misc{marjou2020gipfa,
    title={GIPFA: Generating IPA Pronunciation from Audio},
    author={Xavier Marjou},
    year={2020},
    eprint={2006.07573},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}