freococo/9000hours_voa_burmese_audio

Overview

VOA Burmese radio news archive covering Morning (နံနက် ၅:၃၀ – ၆:၃၀) and Evening (ညပိုင်း ၉:၀၀ – ၁၀:၀၀) programmes for every calendar day from 2012-09-16 → 2025-06-09.

Metric	Value
Hours / rows	9 159
Files per day	2 (morning, evening)
Typical file size	15 – 50 MB
Licence	Public-domain (VOA staff recordings, U.S. 17 U.S.C. § 105)

This dataset upgrades Burmese from low-resource to mid-resource status for speech research, enabling self-supervised pre-training, keyword spotting, and diarisation work without manual transcripts.

Quick-start

from datasets import load_dataset

# Stream the dataset (no full download required)
ds = load_dataset("freococo/9000hours_voa_burmese_audio", streaming=True)

# Inspect one example
ex = next(iter(ds))
print(ex["date"], ex["slot"], ex["audio"]["array"].shape)

Use random 5–30 s crops (or VAD) when feeding SSL / CTC training loops.

Dataset details

• Columns: date, slot, title, article_url, audio (remote MP3)
• Missing audio rows: 28 (broadcast skipped or file removed)
• Exact-URL duplicates removed: final count = 9 159 rows
• Total size if mirrored: ≈ 139 GB @ 15 MB/track or ≈ 465 GB @ 50 MB/track
• Licence caveat: VOA staff speech is public-domain; some segments may contain agency clips or music beds—verify before commercial re-use.
• Suggested uses: self-supervised ASR pre-training, keyword spotting, speaker diarisation, Burmese TTS timbre modelling

Citation

If you use this dataset, please cite it as:

@dataset{freococo_voa_burmese_2025,
  author       = {freococo},
  title        = {VOA Burmese – 9 159 h Audio Index},
  year         = {2025},
  url          = {https://huggingface.co/datasets/freococo/9000hours_voa_burmese_audio},
  note         = {Public-domain U.S. Government material}
}