数据堂—200,955条中文文本韵律标注语料数据

中文文本韵律标注语料数据包含了精心挑选20万条中文文本包含新闻类和口语化句子进行四级韵律标注，句子长度合理，句式多样化。本套数据可作为TTS前端韵律预测训练数据集

Chinese Text Prosodic Annotation Corpus consists of 200,000 carefully selected Chinese texts, including news articles and colloquial sentences, which are annotated with four-level prosodic annotation. The corpus features reasonable sentence lengths and diverse syntactic structures, and can serve as a training dataset for prosody prediction in the front-end module of Text-to-Speech (TTS) systems.

搜集汇总

背景与挑战

背景概述

该数据集包含200,955条中文文本，源自新闻和日常聊天内容，并进行了四级韵律标注。它专为TTS前端韵律预测训练而设计，适用于相关测试任务。