以下のようなデータ構造の配列になっています
{
"audio_filename": "callhome_jpn_0.wav",
"sampling_rate": 16000,
"duration": 606.53,
"utterances": [
{
"start_time": 0.0,
"end_time": 3.4799999999999898,
"speaker": "B"
},
{
"start_time": 1.5999999999999943… See the full description on the dataset page:
https://huggingface.co/datasets/ayousanz/callhome-ja-plus.
