Speech Sources — Segmented and Chunked
Collection
Reusable segmented or chunked speech datasets. • 10 items • Updated
Datasets:
Pods Chunked VAD Transcribed
This public, manually gated dataset contains VAD-produced speech chunks paired with transcripts.
Important: this release is not Sidonized yet. It has only completed VAD/chunk preparation and transcription. It has not gone through the later restoration, alignment, profiling, or final filtering stages.
Contents
- Rows: 17966
- Audio duration: 299.42 hours
- Language: Uzbek
- Access: public repository with manual gated access
Sources
- pods_chunked: 17966 rows
Notes
The dataset is an intermediate raw release for inspection and downstream processing. It should not be treated as the final restored or aligned dataset.
- Downloads last month
- 7