会议数据-1

AISHELL-ASR0055 会议对话语音数据库共 639 场会议，共 381 有效小时。录音语言，中文；录音地区，中国。会议内容覆盖商务、生活、工作等。以中国北方口音区域为主邀请 162 名发音人参与录制。录制过程在真实会议环境中，录制设备包括头戴式麦克风、1 个真实会议语音记录设备、高保真麦克风、Android系统平板、iOS 手机、16 麦面阵和 16 麦圆型麦克风阵列。音频存储格式为 16kHz，16bit。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。

AISHELL-ASR0055 is a meeting dialogue speech database comprising 639 meetings with a total of 381 valid hours. The recordings are in Mandarin Chinese and collected across China. The meeting contents cover diverse scenarios including business, daily life and work, and the majority of speakers have northern Chinese accents. A total of 162 speakers were invited to participate in the recordings, which were conducted in real meeting environments. The recording devices used include head-mounted microphones, one dedicated professional meeting voice recorder, high-fidelity microphones, Android tablets, iOS smartphones, a 16-element planar microphone array, and a 16-element circular microphone array. All audio files are stored in the format of 16 kHz sampling rate and 16-bit bit depth. This database has been transcribed and annotated by professional speech proofreaders and underwent strict quality inspection, with a text transcription accuracy rate of over 95%.

搜集汇总

背景与挑战

背景概述

该数据集包含639场会议录音，总计381有效小时，覆盖商务、生活、工作等内容，主要为中国北方口音。录音在真实会议环境中进行，使用多种设备录制，音频格式为16kHz、16bit，转写标注文本正确率95%以上。