100小时阿拉伯语实体标注手机采集朗读语音数据

阿拉伯语实体标注语音数据_朗读（手机），基于给定的文本朗读并模拟录制，涵盖人名、电话号码、地址、数字字母序列、Email、产品型号、产品序列号、金钱等多个实体类型，内容丰富。此数据集标注了文本内容等多种属性，准确性高，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。

Arabic Entity-Annotated Reading Speech Dataset (Collected via Mobile Phones): Recorded by reading aloud preset texts in a simulated manner, this dataset covers diverse entity types including personal names, phone numbers, addresses, alphanumeric sequences, Email addresses, product models, product serial numbers, monetary amounts and more, with rich content. It is annotated with multiple attributes such as text content, boasts high accuracy, and provides abundant resources for research and applications related to speech recognition. Verified by multiple AI companies, this dataset helps models deliver excellent performance when facing the diversity of real-world scenarios. We strictly comply with data protection regulations and privacy rules to ensure user privacy and legitimate rights and interests are protected during data collection, storage and use. 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100小时通过手机采集的阿拉伯语朗读语音，规模为10GB，标注了人名、电话号码、地址等多种实体类型。它适用于阿拉伯语语音合成和实体识别研究，需授权使用且遵循数据保护法规。