metadata
dataset_info:
features:
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: transcription
dtype: string
- name: input_features
sequence:
sequence: float32
- name: input_length
dtype: int64
- name: reference
dtype: string
- name: prediction_whisper_tiny
dtype: string
- name: prediction_w2v_test
dtype: string
- name: prediction_whisper_large
dtype: string
splits:
- name: validation
num_bytes: 415429747
num_examples: 778
download_size: 409101896
dataset_size: 415429747
configs:
- config_name: default
data_files:
- split: validation
path: data/validation-*