Training script for contentvec

This script makes the environment ready to start preprocessing the dataset, preprocess it, and start training afterwards. Tested on Ubuntu via WSL

Setting up the environment

1. Install the miniconda
2. Extract the file present in Releases to the home folder
3. Run the "setup.sh" script

Preprocess the dataset and start training

1. Convert the audios to 16kHz, in .wav
2. Place the audios in the "dataset" folder
3. Run the "process-dataset-and-start.sh" script and that's it, the training will start right after processing the dataset

Credits

• ContentVec by auspicious3000
• Dictionary creation script by freds0