AISHELL-3
High-fidelity multi-speaker Mandarin TTS dataset with 218 speakers and ~85 hours of recordings
Duration
85 hours
Languages
1
Sample Rate
44.1 kHz
Published
2020-10
Description
1218 native Mandarin speakers with 88,035 utterances
2Emotionally neutral recordings suitable for multi-speaker TTS research
3Provides auxiliary attribute annotations including gender, age group, and native accent
4Over 98% pronunciation accuracy with professional annotation and strict quality control
Language Details
| Language | Duration |
|---|---|
| Mandarin Chinese | 85 hours |
Publisher