metadata
dataset_info:
features:
- name: 'Unnamed: 0'
dtype: int64
- name: path
dtype: audio
- name: label
dtype: string
splits:
- name: train
num_bytes: 2681260515. 414
num_examples: 84843
- name: valid
num_bytes: 312665460.466
num_examples: 9981
- name: test
num_bytes: 344947874.825
num_examples: 11005
- name: train_low
num_bytes: 26658119
num_examples: 849
download_size: 3024419370
dataset_size: 3365531969.705
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: valid
path: data/valid-*
- split: test
path: data/test-*
- split: train_low
path: data/train_low-*
Dataset Card for "beans_sc"
Dataset Description
Paper: https://doi.org/10.48550/arXiv.1804.03209
Dataset Summary
This dataset contains single-word utterances covering 35 English words including digits, commands, and others, spoken by multiple speakers with splits and preprocessing like described in BEANS. It is used for classification tasks and is an auxiliary dataset in BEANS.
Data Splits
| train | train_low | valid | test |
|---|---|---|---|
| 84843 | 849 | 9981 | 11005 |