pre_generation

pre_generation

This folder gathers scripts to download the files needed to generate the dataset of our paper.

We also provide a script to download files from Freesound, with their API.

Download LibriMix

bash download_librispeech.sh

Download then noise files from Zenodo

bash download_noises_from_zenodo.sh

Download some sounds from Freesound

We provide a script to download freesound files either from queries or from IDs. An example file is given to show how to use the main script.

To get an access token, follow these instructions. We will assume that our token is AbcD12eF.

Example 1: Serial download

python download_freesound_queries AbcD12eF config.yaml

Example 2: Parallel downloads

python download_freesound_queries.py \
       --num_jobs 4 \
       --save_dir ../../../dataset/freesound/data/train \
       AbcD12eF \
       config.yaml

For more information about the program, type

python download_freesound_queries.py --help