polaris-audio/street-speech-12h
Speech recognition9.2k rows1.9 GBaudio+jsonLicense: cc-by-4.0ende#noisy#transcribed
Updated Sep 24, 2026
Part of the Open Pace seed catalog. Viewer rows below are illustrative samples built from the schema; the files stream at launch.
Summary
Twelve hours of consented recordings in noisy places, each with a checked transcript.
About 9.2k rows, 1.9 GB on disk, stored as audio+json and released under cc-by-4.0.
Schema
audio— stringtext— stringnoise_level— labelseconds— float
Intended use
- Training and evaluating speech recognition systems.
- Benchmarks, as long as the validation split stays out of training.
Considerations
- Check the license of the dataset and of any linked source before redistribution.
- Report problems with a signed note on this page so the authors can see them.