听障语音数据库

AISHELL-ASR0086 听障语音数据库，邀请母语为中文的听障人群录制，采样率 16000Hz，比特率 16bit。 数据库经过专业语音校对人员转写标注，并通过严格质量检验，文本正确率在 97%以上，主要用于听障语音识别模型训练与调优，改善发音不标准带来的识别偏差，支撑语音转写、实时字幕等无障碍工具研发。 上。AISHELL-ASR0086 Hearing-impaired Speech Database invites native Chinese-speaking hearing-impaired individuals to record speech data, with a sampling rate of 16000 Hz and a bit rate of 16 bit. The database has been transcribed and annotated by professional speech proofreaders, undergone strict quality inspection, and has a text accuracy rate of over 97%. It is primarily used for the training and fine-tuning of hearing-impaired speech recognition models, to mitigate recognition deviations caused by non-standard pronunciation, and to support the development of accessibility tools such as speech transcription and real-time subtitles.

搜集汇总

背景与挑战

背景概述

该数据集为AISHELL-ASR0086听障语音数据库，包含由母语为中文的听障人群录制的语音数据，采样率为16000Hz、比特率为16bit。经过专业转写标注和质量检验，文本正确率超过97%，主要用于训练听障语音识别模型，以改善发音不标准带来的识别偏差，并支持语音转写和实时字幕等无障碍工具研发。