ivrit.ai
收藏资源简介:
ivrit.ai是由魏茨曼科学研究所创建的综合性希伯来语语音数据集,旨在解决希伯来语自动语音识别(ASR)技术资源不足的问题。该数据集包含超过3,300小时的希伯来语语音,来自1000多名不同背景的说话者,涵盖多种语境和话题。数据集提供三种形式:原始未处理音频、语音活动检测后的数据和部分转录数据,以满足不同研究需求。ivrit.ai的开放访问性质使其成为研究人员、开发者和商业实体的重要资源,广泛应用于提升希伯来语AI能力,如紧急响应系统、辅助工具和医疗转录服务等。
ivrit.ai is a comprehensive Hebrew speech dataset developed by the Weizmann Institute of Science, aiming to address the shortage of technical resources for Hebrew automatic speech recognition (ASR) technology. This dataset contains over 3,300 hours of Hebrew speech from more than 1,000 speakers with diverse backgrounds, covering various contexts and topics. It offers three formats: raw unprocessed audio, data processed via voice activity detection (VAD), and partially transcribed data to meet different research requirements. Benefiting from its open access nature, ivrit.ai has become a critical resource for researchers, developers and commercial entities, and is widely applied to enhance Hebrew AI capabilities such as emergency response systems, assistive tools and medical transcription services.

- 1ivrit.ai: A Comprehensive Dataset of Hebrew Speech for AI Research and Development魏茨曼科学研究所 · 2023年



