基于大规模语音样本声调分析软件的设计与实现

Design and Implementation of Tone Analysis Software Based on Large-Scale Speech Samples

摘要: 基于大规模语音样本的声调分析软件在基础语音分析功能的同时，创造性地设计并实现了批量处理通过大容量词表调查获取的大量单音节词语音样本数据的程序，分析流程包括：音频载入、提取基频、确定声调段、筛选F0数据、数据预处理、数据归一化、调值估计、数据聚类，整个过程全自动运行。该方法的应用价值在于能够与广泛使用的小样本分析方法形成参照、有效地挖掘出声调系统细微的变异与演变信息、为音系整理工作提供些许便利。

Abstract: Besides the basic speech analysis function, the tone analysis software based on large-scale speech samples creatively designs and implements the program for batch processing of a large number of voice sample data of monosyllabic words obtained through a large volume of word list. The analysis process includes: audio loading, F0 extraction, tone segment determination, F0 data screening, data preprocessing, data normalization, tone value estimation, data clustering. The whole process is fully automatic operation. The application value of this method lies in that it can form a reference with the widely used small sample analysis method, effectively dig out the subtle variation and evolution information of the tone system, and provide some convenience for phonological system arrangement.