The dataset is currently empty. Upload or create new data files. Then, you will be able to explore them in the Dataset Viewer.
Indonesian Voice Note Dataset
Dataset plan for Indonesian WhatsApp-style voice note intelligence.
Intended Fields
{
"audio_id": "",
"audio_path": "",
"language": "id",
"accent_region": "",
"transcript": "",
"summary": "",
"intent": "",
"urgency": "low|medium|high",
"action_items": [],
"deadline_mentions": [],
"reply_style": "",
"consent": true
}
Collection Rules
- Use only consented audio.
- Do not include private third-party voice notes.
- Remove phone numbers, addresses, account numbers, and sensitive identifiers.
- Mark dialect, code-mixing, and noisy audio where relevant.
First Milestone
Collect 50-100 short consented samples covering casual, work, student, customer-service, and UMKM scenarios.
- Downloads last month
- 10