9,283 recorded hours of audio in MP3 format paired with corresponding text files across 60 different languages. The collection includes 7,335 validated hours and features demographic metadata such as age, sex, and accent for a subset of the recordings.
Use Cases
- Train speech-to-text models using the MP3 audio and corresponding text files
- Analyze speech patterns across different demographics using the age, sex, and accent metadata
- Develop language-specific acoustic models for any of the 60 supported languages
Strengths
- 9,283 total recorded hours of audio data
- 7,335 validated hours across 60 distinct languages
- Includes demographic metadata fields for age, sex, and accent
- Data format consists of MP3 audio files paired with text transcriptions