数据堂—344人美式英语手机采集语音数据_引导

344人美式英语手机采集语音数据_引导通过采集344名美式英语发音人，均为美国本地人员。每人50句，有效时长9.7小时，录音环境安静，录音内容丰富，美式英语手机采集语音数据_引导可应用于车载场景、智能家居、语音助手等多个领域

Mobile-collected American English speech dataset: This dataset is constructed via guided data collection with 344 native American English speakers, all of whom are local residents of the United States. Each participant recorded 50 utterances, resulting in a total effective recording duration of 9.7 hours. All recordings were conducted in quiet environments with rich and diverse content. This dataset can be applied to multiple fields including in-vehicle scenarios, smart home systems, voice assistants and other relevant domains.

搜集汇总

背景与挑战

背景概述

该数据集包含344名美国本土英语母语者通过手机采集的9.7小时语音数据，录制于安静室内环境，内容多样，适用于车载、智能家居等场景的英语语音识别模型测试。数据格式为16kHz WAV文件，附带文本转录和元数据，采用Apache 2.0许可证并支持商业用途。