数据堂—240小时印地语手机采集语音数据_朗读

240小时印地语手机采集语音数据由401名印度人参与录制；录音涵盖安静和噪音的不同环境，更贴合语音识别实际应用场景；录音内容丰富，覆盖经济，娱乐，新闻，口语等多个领域，所有文本由人工转写，准确率高。240小时印地语手机采集语音数据可应用于语音识别、机器翻译、声纹识别

This 240-hour Hindi speech dataset collected via mobile phones was recorded by 401 Indian participants. The recordings cover various environments with both quiet and noisy conditions, which better aligns with real-world application scenarios for speech recognition. The recorded content is diverse, covering multiple domains such as economy, entertainment, news, and daily conversational language. All transcriptions are manually conducted with high accuracy. This dataset can be applied to tasks including speech recognition, machine translation, and speaker verification.

搜集汇总

背景与挑战

背景概述

该数据集包含240小时的印地语手机采集语音数据，由401名印度参与者在安静和嘈杂环境中录制，内容涵盖经济、娱乐、新闻等多个领域。所有音频均经过高精度人工转录，适用于语音识别、机器翻译和说话人验证等任务，数据格式为16kHz、16bit的WAV文件。