Automatic Speech Recognition (ASR) with PyTorch

About • Installation • How To Use • Credits • License

WanDB report

About

This repo is my own implementation of DeepSpeech2 ASR model

Installation

Follow these steps to install the project:

0. Create and activate new environment using conda.

   a. conda version:

   # create env
   conda create -n asr python=3.11
   
   # activate env
   conda activate asr

1. Install all required packages

   cd deepspeech2
   
   pip install -r requirements.txt

2. Install pre-commit:

   pre-commit install

3. Dowload model weights:

   cd deepspeech2
   
   python download.py

How To Use

To train a model, run the following command:

python3 train.py -cn=CONFIG_NAME HYDRA_CONFIG_ARGUMENTS

Where CONFIG_NAME is a config from src/configs and HYDRA_CONFIG_ARGUMENTS are optional arguments.

To run inference (evaluate the model or save predictions):

python3 inference.py HYDRA_CONFIG_ARGUMENTS

To evaluate WER and CER on your predictions, simply run

python wer_cer_calculation.py --dir_path_gt <Path to ground truth folder with .txt> --dir_path_pred <Path to dir with predicions>

Credits

This repository is based on a PyTorch Project Template.

License