计算机与数字工程2025，Vol.53Issue(3)：713-717,754,6.DOI:10.3969/j.issn.1672-9722.2025.03.018
联合序列多特征均值的方言分类方法
Dialect Classification Method for Multi-feature Means of Joint Sequences
摘要
Abstract
In view of the problem that it is difficult to fully capture useful information from a single phonological feature and the waste of data resources caused by fixed length phonological training,this paper proposes to use the method of multi feature fu-sion to preliminarily extract the features of the phonological signal,so that the model can receive comprehensive audio information and effectively solve the problem of under fitting caused by a single feature.The sequence length of different speech features is uni-fied to 1 by using the mean value of sequence features,it not only solves the problem that the indefinite length speech cannot be trained,but also solves the disadvantage of data waste caused by the fixed length phonological training.Dialect classification is car-ried out by using a model structure similar to WaveNet,which further improves the feature extraction ability of dialect classification.The experimental results show that compared with other methods and models,the classification accuracy of the proposed method is significantly improved under the data used in this paper.关键词
多特征/均值/不定长/WaveNet/准确率Key words
multi-feature/mean value/variable length/WaveNet/accuracy rate分类
数理科学引用本文复制引用
宋欢,陈雪..联合序列多特征均值的方言分类方法[J].计算机与数字工程,2025,53(3):713-717,754,6.基金项目
国家重点研发计划项目(编号:2017YFB1400704)资助. （编号:2017YFB1400704）
