数据堂—100人中文通用平均音色合成库

100人中文通用平均音色合成库，由中文母语发音人录制。涵盖新闻、日常口语、有声读物、诗词、广告、消息播送、娱乐等类别，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。共计74.5小时

This is a 100-speaker Mandarin general-purpose average voice synthesis corpus recorded by native Mandarin speakers. It covers categories including news, daily spoken language, audiobooks, poetry, advertisements, news broadcasting, entertainment and others. The speech corpus has balanced phoneme coverage, and is annotated by professional phoneticians, which precisely meets the R&D requirements of speech synthesis. The total duration of the corpus is 74.5 hours.

搜集汇总

背景与挑战

背景概述

该数据集为数据堂发布的100人中文通用平均音色合成库，专用于中文通用平均音色合成模型的测试任务。它包含100名不同年龄和性别的说话者录制的新闻、日常口语、有声书等多种内容，音频格式为48kHz WAV，并采用专业语音标注以支持语音合成研发。