数据堂—998人远场家居手机麦克风阵列采集语音数据

中国普通话远场家居语音数据（麦克风阵列+手机），由998人参与语音采集，男女比例1;1；录制文本涵盖通用类语句、家居场景指令、功能性助手、唤醒词、数字等多种应用场景。为语音识别相关研究及应用提供了丰富的资源，经多家AI公司验证：有助于模型面对真实世界的多样性时能够表现出色。我们严格遵循数据保护法规和隐私规定，确保数据采集、存储和使用的过程中维护用户的隐私和合法权益，所有数据均遵循GDPR, CCPA, PIPLFar-field Mandarin Chinese home speech dataset (microphone array + mobile phone) was collected from 998 participants, with a 1:1 male-to-female gender ratio. The recorded speech texts cover a wide range of application scenarios, including general sentences, home scene commands, functional assistant utterances, wake-up words, numerals and more. This dataset serves as a rich resource for speech recognition-related research and applications, and has been validated by multiple AI companies to help models deliver excellent performance when facing real-world diversities. We strictly comply with data protection regulations and privacy rules, ensuring that user privacy and legitimate rights and interests are fully protected throughout the entire process of data collection, storage and usage. All datasets comply with GDPR, CCPA and PIPL.

搜集汇总

背景与挑战

背景概述

该数据集包含998名参与者的远场家居手机麦克风阵列采集的语音数据，男女比例均衡，录音内容覆盖家庭场景命令、唤醒词等多种应用。数据适用于家庭环境中的语音增强和语音识别任务，采用48kHz采样率，麦克风阵列配置包括多种圆形和线性结构。