license: cc-by-nc-nd-4.0
Description
The data were collected from 370 speakers, each with about 500 spoken sentences in Hangzhou dialect. Local people in Hangzhou participated in the recording and the accent was authentic. The recording content is colloquial and life-oriented, and reading is more natural and fluent. Hangzhou locals participate in QA and proofreading, and text transcription is more accurate. Equipped with mainstream Android and Apple phones.
For more details, please refer to the link: https://www.nexdata.ai/datasets/speechrecog/30?source=Huggingface
Format
16kHz, 16bit, uncompressed wav, mono channel
Recording environment
quiet indoor environment, without echo
Recording content (read speech)
oral language
Speaker
370 people from Hangzhou, 83% of which are female
Device
Android mobile phone, iPhone
Language
Hangzhou dialect
Transcription content
text, noise symbols
Accuracy rate
95% (the accuracy rate of noise symbols is not included)
Application scenarios
speech recognition, voiceprint recognition
Licensing Information
Commercial License