metadata
dataset_info:
features:
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: speaker_id
dtype: string
- name: hidden_states
sequence: float16
- name: probabilities
sequence: float16
- name: sorted_probabilities
sequence: int64
- name: top_accents
sequence: string
- name: most_frequent_first_accent
dtype: string
- name: entropy_score
dtype: float64
- name: weighted_entropy_accent
dtype: string
- name: borda_accent
dtype: string
splits:
- name: train
num_bytes: 9747001278.316
num_examples: 88172
download_size: 9680238077
dataset_size: 9747001278.316
configs:
- config_name: default
data_files:
- split: train
path: data/train-*