metadata
dataset_info:
features:
- name: mp3
dtype:
audio:
sampling_rate: 16000
- name: json
struct:
- name: caption
dtype: string
- name: duration
dtype: float64
- name: folder_name
dtype: string
- name: sample_id
dtype: string
- name: input_features
sequence:
sequence: float32
- name: input_length
dtype: float64
- name: labels
sequence: int64
- name: forced_ac_decoder_ids
sequence: int64
splits:
- name: train
num_bytes: 1029949155
num_examples: 811
- name: test
num_bytes: 59880522
num_examples: 47
- name: validation
num_bytes: 48864240
num_examples: 39
- name: val
num_bytes: 48864240
num_examples: 39
- name: train_mini
num_bytes: 9947087
num_examples: 8
- name: val_mini
num_bytes: 40216642
num_examples: 32
download_size: 611424049
dataset_size: 1237721886
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
- split: validation
path: data/validation-*
- split: val
path: data/val-*
- split: train_mini
path: data/train_mini-*
- split: val_mini
path: data/val_mini-*