metadata
dataset_info:
features:
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: binary_mask
sequence: int64
- name: chunk_id
dtype: string
- name: sample_name
dtype: string
- name: entity_token_count
dtype: int64
- name: mask_stats
struct:
- name: entity_percentage
dtype: float64
- name: entity_tokens
dtype: int64
- name: non_entity_tokens
dtype: int64
splits:
- name: train
num_bytes: 3518098799.86
num_examples: 3619
- name: validation
num_bytes: 391763995
num_examples: 403
download_size: 3754738147
dataset_size: 3909862794.86
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*