数据堂—397人印地语手机采集语音数据_引导

印地语手机采集语音数据_引导是由397名印度人录制，每人约50句，8.6小时；口音正宗，转写精确；录音内容涉及车载场景，智能家居，智能语音助手；印地语手机采集语音数据_引导可用于机器翻译的语料库建设以及用于声纹识别的模型训练和算法研究

Guided Hindi Mobile-collected Speech Dataset was recorded by 397 Indian participants, with approximately 50 utterances per individual, totaling 8.6 hours of audio duration. It features authentic Hindi accents and highly accurate transcriptions. The recorded content covers three main scenarios: in-vehicle environments, smart home scenarios, and intelligent voice assistant scenarios. This dataset can be utilized for the construction of machine translation corpora, as well as model training and algorithmic research for speaker verification.

搜集汇总

背景与挑战

背景概述

该数据集包含397名印度人通过手机录制的印地语语音，总时长约8.6小时，采样率为16kHz，适用于印地语语音识别模型的测试任务。录音内容涵盖车载场景、智能家居和语音助手，数据格式为未压缩的WAV文件，版权归数据堂所有。