Dataset · huuuyeah
US city council meeting audio recordings and transcripts
huuuyeah/MeetingBank_Audio
Benchmark featuring long city council meeting recordings, transcripts, and protocols for summarization.
What's inside
It contains 1,366 recordings of city council meetings from six major US cities with a total duration of over 3,579 hours, along with transcripts, agendas, and PDF protocols. On average, a single meeting lasts 2.6 hours, and its transcript exceeds 28,000 tokens. The modality in the repository is audio, and the volume is between 1,000 and 10,000 records.
What you can do with it
- Evaluating and fine-tuning models for summarizing long meetings and multi-hour dialogues
- Training systems to extract structure and key decisions from meeting records
- Testing ASR systems on long multi-speaker speeches
- Size
- 1K–10K rows
- Downloads this month
- 2.8K
- Likes
- 24
How to load it
from datasets import load_dataset
ds = load_dataset("huuuyeah/MeetingBank_Audio")