Shanghai Dialect Speech Data by Mobile Phone - 1,030 Hours

It collects 2.956 speakers from Shanghai and is recorded in quiet indoor environment. The recorded content includes multi-domain customer consultation, short messages, numbers, Shanghai POI, etc. The corpus has no repetition and the average sentence length is 12.68 words. Recording devices are mainstream Android phones and iPhones.Format：16kHz, 16bit, uncompressed wav, mono channelRecording Environment：quiet indoor environment, without echoRecording Content (read speech)：generic category, human-machine interaction category, numbers, Shanghai POI.Speaker：2,956 people, 1,921 females, accounting for 65%Device：Android mobile phone and iPhoneLanguage：Shanghai dialectTranscription content：text, 4 noise symbols, special identifiersAccuracy rate：95% (the accuracy rate of noise symbols and other identifiers is not included)Application scenarios：speech recognition, voiceprint recognition

本数据集汇集了来自上海的2,956位说话者，录音环境为安静室内，内容涵盖多领域客户咨询、短信、数字、上海POI等。数据集无重复，平均句长为12.68词。录音设备为主流的安卓手机和iPhone。格式为16kHz，16位，未压缩的wav格式，单声道。录音环境为无回声的安静室内环境。录音内容包括通用类别、人机交互类别、数字、上海POI等。说话者包括2,956人，其中女性1,921人，占比65%。设备为安卓手机和iPhone。语言为上海方言。转录内容包含文本、4个噪声符号、特殊标识符。准确率为95%（不包括噪声符号和其他标识符的准确率）。应用场景包括语音识别、声纹识别。