海天瑞声-新加坡英语识别语音库-对话（手机）

真实场景采集，包含手机和网络电话的双通道录音，所有对话内容进行了精准的句意切分。话题包含：电销客服、金融消费、通用生活口语、社会热点、旅游购物、体育娱乐、教育学习、医疗新冠、科技数码游戏等领域。30%的英文对话内容与电销客服和金融消费相关

This dataset is collected from real-world scenarios, featuring dual-channel recordings of both mobile phone calls and Voice over Internet Protocol (VoIP) calls. All dialogue content has been precisely segmented based on sentence meaning. The covered topics span various domains including telemarketing customer service, financial consumption, general colloquial daily speech, social hot topics, travel and shopping, sports and entertainment, education and learning, COVID-19-related medical scenarios, as well as technology, digital products and gaming. Notably, 30% of the English dialogue content is related to telemarketing customer service and financial consumption.

搜集汇总

背景与挑战

背景概述

该数据集由海天瑞声发布，是一个针对手机场景的新加坡英语对话语音识别库，包含400小时的真实场景双通道录音，内容按句意精确切分，覆盖日常对话、电话销售客服等多个话题。数据格式为16KHz、16位WAV，涉及200名参与者，主要用于大型语音模型的测试和训练，且版权允许商业使用。