数据堂—496人印尼语手机采集语音数据

印尼语手机采集语音数据_引导是由496位印尼本地人员参与采集。录音环境安静，录音内容包括车载、家居、语音助手等多个领域。每人约50句，有效数据时长10.5小时，所有文本由人工转写，准确率高

Guided Indonesian speech dataset collected via mobile devices involved 496 local Indonesian participants. All recordings were conducted in quiet environments, covering multiple domains such as in-vehicle scenarios, home daily scenarios and voice assistant interactions. Each participant contributed approximately 50 utterances, resulting in a total valid duration of 10.5 hours. All transcriptions were manually performed with high accuracy.

搜集汇总

背景与挑战

背景概述

该数据集专用于印尼语语音识别模型的测试，由496名本地参与者通过手机在安静环境中采集，包含约10.5小时的有效语音数据，格式为16kHz单声道wav文件，内容覆盖娱乐、新闻等多个领域，且所有文本均经过高精度人工转录。