metadata
dataset_info:
features:
- name: audio
dtype: audio
- name: filename
dtype: string
splits:
- name: train
num_bytes: 128746364164.175
num_examples: 848135
download_size: 104451357229
dataset_size: 128746364164.175
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
Unfortunately I cannot share access to this dataset publicly at this time due to various reasons. Please email me me @ mrfake . name or reach me on Discord realmrfakename for any questions.
GSpeech
A lot of the data is pretty noisy - needs to be cleaned using a model such as SAM-Audio, etc. Though great range of emotions.