metadata
dataset_info:
features:
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: transcription
dtype: string
- name: duration
dtype: int32
splits:
- name: train_with_synthetic
num_bytes: 31951379121.666
num_examples: 109682
- name: train_without_synthetic
num_bytes: 34888443975.04
num_examples: 69368
- name: validation
num_bytes: 4226085661.72
num_examples: 9520
- name: test_mls
num_bytes: 430870413
num_examples: 871
- name: test_fleurs
num_bytes: 373310465
num_examples: 919
- name: test_cv
num_bytes: 1453742074.336
num_examples: 9362
- name: test_bracarense
num_bytes: 596697965
num_examples: 867
download_size: 72087977166
dataset_size: 73920529675.762
configs:
- config_name: default
data_files:
- split: train_with_synthetic
path: data/train_with_synthetic-*
- split: train_without_synthetic
path: data/train_without_synthetic-*
- split: validation
path: data/validation-*
- split: test_mls
path: data/test_mls-*
- split: test_fleurs
path: data/test_fleurs-*
- split: test_cv
path: data/test_cv-*
- split: test_bracarense
path: data/test_bracarense-*
Data used to reproduce all of the experiments in the paper ; https://ieeexplore.ieee.org/document/10720758/