README.md

metadata

language:
  - ta
task_categories:
  - automatic-speech-recognition
tags:
  - audio
  - speech
  - parquet

Audio Clips Dataset

AudioCaps-style parquet dataset generated from MG.

The dataset stores clipped audio directly inside parquet shards under data/.

Columns

• audio: embedded audio bytes and clip path
• audio_length: approximate number of samples at 16000 Hz
• caption: transcript text
• language, duration, upload_id, segment_id, collection, label, source_audio, start, end

Usage

from datasets import Audio, load_dataset

ds = load_dataset("007ask/MG", split="train", streaming=True)
ds = ds.cast_column("audio", Audio(sampling_rate=16000))

first = next(iter(ds))
print(first["audio"]["array"].shape, first["caption"])