It collects 2,568 local Chinese from Henan, Shanxi, Sichuan, Hunan and Fujian. It is mandarin speech data with heavy accent. The recorded content is a sentence that the speaker freely answers according to the guiding questions. Format：16kHz, 16bit, uncompressed wav, mono channel. Recording Environment：1,605 people complete the recording in relatively quiet indoor environment; and 963 in the normal environment with noise that does not affect the voice recognition Recording Content：smart car; smart home; speech assistant. Demographics：2,568 people; 53% are females; people aged from 21-30 account for 51%; people are from 28 provinces including Henan, Shaanxi, Sichuan, Hunan, Fujian, Heilongjiang, Guizhou. Device：Android mobile phone, iPhone. Language：mandarin with heavy local accent Application Scenarios：speech recognition; voiceprint recognition Accuracy：not lower than 98%.

Creator(s)

-

Distributor(s)

ELRA

Right Holder(s)

-