131小时中国重口音普通话重口音语音数据_引导（手机）

中国重口音普通话重口音语音数据_朗读（手机），包含发音人根据引导问题自由作答的句子。此数据集标注了文本内容等多种属性，由2568名来自河南、陕西、四川、湖南、福建等省份的中国本土发音人参与录制，男女均衡，口音正宗，文本经过人工校对，准确率高，为语音识别相关研究及应用提供了丰富的资源，经多家AI公司验证：有助于模型面对真实世界的多样性时能够表现出色。我们严格遵循数据保护法规和隐私规定，确保数据采集、存储和使用的过程中维护用户的隐私和合法权益，所有数据均遵循GDPR, CCPA, PIPL。

Chinese Mandarin Speech Dataset with Regional Heavy Accents (Read Speech Recorded via Mobile Phones) consists of sentences spoken by speakers in response to guided prompts with free-form answers. This dataset is annotated with multiple attributes including text transcriptions. It was recorded by 2568 native Chinese speakers from provinces including Henan, Shaanxi, Sichuan, Hunan, Fujian and other regions, with a balanced gender ratio and authentic regional accents. All text transcripts have been manually proofread, ensuring high accuracy. This dataset provides abundant resources for speech recognition-related research and applications. Verified by multiple AI enterprises, it helps models deliver robust performance when facing real-world diversity. We strictly comply with data protection regulations and privacy norms, safeguarding user privacy and legitimate rights and interests throughout the entire process of data collection, storage and usage. All data adheres to GDPR, CCPA and PIPL.

搜集汇总

背景与挑战

背景概述

该数据集包含131小时中国重口音普通话语音数据，由2568名来自河南、陕西、四川等多省份的发音人通过手机录制，男女均衡、口音正宗，标注文本准确率达98%，适用于智能家居、车载等语音识别场景。数据格式为16kHz WAV单声道音频，涵盖安静和普通环境录音，旨在提升模型在真实世界中的多样性表现，并严格遵循数据隐私法规。