metadata
dataset_info:
features:
- name: pair_id
dtype: float64
- name: image
dtype: int64
- name: sentence_id
dtype: int64
- name: image_id
dtype: string
- name: sentence
dtype: string
- name: translation
dtype: string
- name: speaker_id
dtype: int64
- name: gender
dtype: string
- name: duration
dtype: float64
- name: sample_rate
dtype: int64
- name: audio_format
dtype: string
- name: recording_lang
dtype: string
- name: native_lang
dtype: string
- name: audio
dtype:
audio:
sampling_rate: 16000
splits:
- name: train
num_bytes: 20268029847.548
num_examples: 82371
- name: valid
num_bytes: 669071385.274
num_examples: 2782
- name: test
num_bytes: 679122730.122
num_examples: 2779
download_size: 19804226281
dataset_size: 21616223962.944
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: valid
path: data/valid-*
- split: test
path: data/test-*