AISHELL-2
A 1000-hour industrial-scale Mandarin ASR dataset covering 1,991 speakers with diverse accents
Duration
1000 hours
Languages
1
Sample Rate
16 kHz
Published
2018-08
Description
11,991 speakers from various accent regions across China
2Recorded through three parallel acoustic channels: high-fidelity microphone, Android phone, and iOS device
3Accent distribution: 1,293 northern accent speakers, 678 southern accent speakers, 20 other accent speakers
4Content covers 8 major topics: voice commands, IoT device control, points of interest, entertainment, finance, technology, sports, free conversation, etc.
5Freely available for academic research
Language Details
| Language | Duration |
|---|---|
| Mandarin Chinese | 1000 hours |
Publisher