metadata
dataset_info:
config_name: serbia_second_push
features:
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: key
dtype: string
- name: country
dtype: string
- name: language
dtype: string
- name: video_id
dtype: string
- name: transcript_id
dtype: string
- name: start_seconds
dtype: float32
- name: end_seconds
dtype: float32
- name: duration_seconds
dtype: float32
- name: asr_transcript
dtype: string
- name: human_transcript
dtype: string
- name: cer
dtype: float32
- name: wer
dtype: float32
- name: original_transcript_start_idx
dtype: int32
- name: original_transcript_end_idx
dtype: int32
splits:
- name: train
num_bytes: 152335904055.752
num_examples: 321032
- name: validation
num_bytes: 1559771277.745
num_examples: 3215
- name: test
num_bytes: 1592447690.64
num_examples: 3332
download_size: 158129235349
dataset_size: 155488123024.13702
configs:
- config_name: serbia_second_push
data_files:
- split: train
path: serbia_second_push/train-*
- split: validation
path: serbia_second_push/validation-*
- split: test
path: serbia_second_push/test-*