相似语言环境下目标语音提取研究  

Research on target speech extraction in similar language environment

作　　者：王智 Wang Zhi(College of Electronic Information,Guangxi Minzu University,Nanning 530006,China)

机构地区：[1]广西民族大学电子信息学院,南宁530006

出　　处：《现代计算机》2024年第14期84-88,共5页Modern Computer

摘　　要：针对实际应用中常见的相似语音分离问题,构建了一个新的数据库P-C以模拟相似语言环境。该数据库结合了自制的重庆方言数据集CQSpeech和公开的中文数据集THCH30,以此来研究普通话与重庆方言混合语音的分离问题。另外,为了充分利用语音特征,在CRN网络中嵌入说话人特征。首先通过模型训练大量数据以获取说话人特征,然后将说话人特征与分离模型中的特征进行融合,这样能够有效地提高相似语言语音分离的清晰度和准确性。根据实验验证,该模型在P-C数据库上展示了良好的分离效果。Aiming at the common similar speech separation problem in practical applications,a new database,P‑C,is con‑structed to simulate similar language environments.This database combines the self‑made Chongqing dialect dataset CQSpeech and the publicly available Chinese dataset THCH30 as a way to study the separation problem of mixed speech between Mandarin and Chongqing dialects.In addition,in order to fully utilize the speech features,speaker features are embedded in the CRN net‑work.A large amount of data is first trained by the model to obtain speaker features,and then the speaker features are fused with the features in the separation model.This can effectively improve the clarity and accuracy of similar language speech separation.According to the experimental verification,the model demonstrates good separation effect on P‑C database.

关 键 词：相似语言 语音分离 数据集 

分 类 号：TN912.3[电子电信—通信与信息系统] TP18[电子电信—信息与通信工程]