data/README.md

Download Datasets

WenetSpeech

1. Fill out the Google Form to request access to the WenetSpeech dataset.

2. Obtain your Hugging Face API token from Hugging Face Account Settings.

3. Run the following command in your terminal to log in:

cd data
huggingface-cli login

4. Run the following command in your terminal to download dataset:

python download_benchmark.py --output-dir ~/downloaded_datasets --dataset WenetSpeech

Others

1. Run the following command in your terminal to download all dataset(If you only want to download a single dataset, the dataset name is as follows: VoiceBench, OpenAudioBench, LibriSpeech, WenetSpeech, Fleurs, AISHELL-1, AISHELL-2, MMAU, ClothoAQA, VocalSound, Nonspeech7k, MELD, TUT2017, CochlScene)

python download_benchmark.py --output-dir ~/downloaded_datasets --dataset all