This script makes the environment ready to start preprocessing the dataset, preprocess it, and start training afterwards. Tested on Ubuntu via WSL
- Install the miniconda
- Extract the file present in Releases to the home folder
- Run the "setup.sh" script
- Convert the audios to 16kHz, in .wav
- Place the audios in the "dataset" folder
- Run the "process-dataset-and-start.sh" script and that's it, the training will start right after processing the dataset
- ContentVec by auspicious3000
- Dictionary creation script by freds0