DataoceanAI/Chinese_Mandarin_Speech_Recognition_Corpus

该数据集具有高采样率，包含车载语料库和室内安静环境下的录音，覆盖多种场景（如车辆控制、音乐、通用、地图和闲聊场景），适用于车载和其他常见的语音识别场景。音频参数包括16k采样率下1人0.5小时的录音，44.1k采样率下148人74.9小时的录音，以及48k采样率下2463人1354.8小时的录音。数据集总时长为1430.2小时，包含1092名男性和1520名女性的录音。

This dataset is a high-quality speech library for speech recognition, particularly suitable for in-vehicle environments and other common speech recognition scenarios. It includes speech data from multiple scenarios such as vehicle control, music, map, and casual chat. The audio parameters cover speech data at different sampling rates, totaling 1430.2 hours, with 1092 males and 1520 females.