metadata
dataset_info:
features:
- name: sample_id
dtype: int64
- name: dirty_audio
dtype:
audio:
sampling_rate: 44100
- name: clean_audio
dtype:
audio:
sampling_rate: 44100
- name: predicted_audio
dtype:
audio:
sampling_rate: 44100
- name: dirty_tokens
dtype: string
- name: clean_tokens
dtype: string
- name: predicted_tokens
dtype: string
splits:
- name: train
num_bytes: 44578954
num_examples: 58
download_size: 38382771
dataset_size: 44578954
configs:
- config_name: default
data_files:
- split: train
path: data/train-*