智能家居语音数据-中文

AISHELL-ASR0010 智能家居语音数据库共 1852 小时。录音语言，中文；录音地区，中国。录音文本包含主流家居场景智能控制、命令等 11 个分类。以中国北方口音区域为主邀请 646名发音人参与录制。录制过程在真实家居环境中，模拟智能家居电器及应用产品使用情况，设置 7 个录音位。其中 5 个为高保真麦克风(44.1kHz，16bit，每个麦克风录制时长在 217~284小时之间)、1 个 iOS 系统手机(16kHz，16bit，282 小时)与 1 个 Android 系统手机(16kHz，16bit，281 小时)进行录制。此数据库经过专业语音校对人员转写标注，并通过严格质量检验，文本正确率在 95%以上。

The AISHELL-ASR0010 Smart Home Speech Database has a total duration of 1852 hours. All recordings are in Chinese and collected in China. The transcribed texts cover 11 categories including mainstream smart home control scenarios, voice commands and other relevant contents. A total of 646 speakers, mainly from regions with northern Chinese accents, were invited for the recording sessions. The data was collected in real home environments, simulating the usage of smart home appliances and related application products, with 7 recording positions deployed. Five high-fidelity microphones (44.1 kHz, 16-bit, each with a recording duration ranging from 217 to 284 hours), one iOS smartphone (16 kHz, 16-bit, 282 hours) and one Android smartphone (16 kHz, 16-bit, 281 hours) were utilized for the recording. This database has undergone transcription and annotation by professional speech proofreaders, passed strict quality inspections, and achieved a text accuracy rate of over 95%.