metadata
dataset_info:
features:
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: transcription
dtype: string
- name: input_ids
sequence: int32
- name: labels
sequence:
sequence: float32
- name: input_length
dtype: int64
- name: speaker_embeddings
sequence: float32
splits:
- name: train
num_bytes: 952142456.4864544
num_examples: 5054
download_size: 979797232
dataset_size: 952142456.4864544
configs:
- config_name: default
data_files:
- split: train
path: data/train-*