数据堂—794小时四川方言手机采集语音数据

中国四川方言语音数据_朗读（手机），基于给定的脚本朗读并模拟录制，由2,507名四川发音人，在安静的室内环境下的录音数据。录音内容广泛，覆盖日常短信及多领域客户咨询，句子平均重复次数1.3次，平均句长12.5字。文本经过人工校对，准确率高，匹配主流安卓、苹果系统手机，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

Mobile-recorded Sichuan Dialect Reading Aloud Speech Dataset. This dataset consists of audio recordings collected from 2,507 Sichuan dialect speakers, who read aloud pre-defined scripts in quiet indoor environments, with the recordings simulating real-world daily usage scenarios. The recorded content covers a wide range of scenarios including daily text messages and customer inquiries across multiple domains. The average number of repetitions per sentence is 1.3, with an average sentence length of 12.5 Chinese characters. All transcripts have been manually proofread with high accuracy. The dataset is compatible with mainstream Android and iOS mobile devices, providing valuable resources for speech recognition-related research and practical applications. Verified by multiple AI companies, this dataset enables speech recognition models to deliver robust performance when confronted with real-world diversity. We strictly comply with international data protection regulations and privacy standards to safeguard users' privacy and legitimate rights and interests throughout the entire process of data collection, storage and utilization. All dataset contents are in compliance with GDPR, CCPA and PIPL.

搜集汇总

背景与挑战

背景概述

该数据集包含794小时的四川方言手机采集语音数据，由2507名四川人在安静室内环境下录制，内容覆盖日常文本和客户咨询，音频格式为16kHz、16bit单声道WAV，适用于中文语音识别模型的测试任务。数据集经过质量检查和本地校对，兼容主流安卓和苹果手机系统。