数据堂—1,297小时录音笔采集场景噪音数据

录音笔采集场景噪音数据，时长1,297小时，采集涵盖地铁、超市、餐厅、马路等多场景噪音数据。音频均使用专业录音笔录制，采用高采样率、双声道格式收集，对非纯噪音部分标注时间点和类型。1,297小时录音笔采集场景噪音数据可应用于噪声建模

This dataset contains scene noise data collected using professional digital voice recorders, with a total duration of 1,297 hours. The collected noise data covers various scenarios including subway, supermarket, restaurant, street, and other typical environments. All audio recordings were captured with professional voice recorders, adopting high sampling rate and stereo channel formats. Non-pure noise segments are annotated with their timestamps and corresponding types. This 1,297-hour scene noise dataset collected via professional voice recorders can be applied to noise modeling.

搜集汇总

背景与挑战

背景概述

该数据集包含1,297小时由录音笔采集的场景噪音数据，覆盖地铁、超市、餐厅等多种嘈杂环境，音频格式为44.1kHz 16bit立体声WAV，并标注了非纯噪声段的时间和类型，适用于噪声建模和噪声语音识别模型的测试任务。