数据堂—50人中英混读客服平均音色合成库

50人中英混读客服平均音色合成库，由中文母语发音人录制，客服场景的录音文本，语料音素覆盖均衡，专业语音学家参与标注，精准匹配语音合成的研发需求。

A 50-speaker average-timbre speech synthesis corpus for mixed Mandarin-English customer service scenarios, recorded by native Mandarin speakers. The corpus uses texts tailored for customer service scenarios, has balanced phoneme coverage, is annotated by professional phoneticians, and accurately meets the R&D requirements of speech synthesis.

搜集汇总

背景与挑战

背景概述

该数据集是一个50人中英混读客服平均音色合成库，用于测试客服语音合成模型，包含50名20~30岁母语者录制的客服场景中英混读音频，覆盖平衡的音节和音素。数据格式为48kHz WAV，并进行了拼音和韵律标注。