噪声与振动控制2025，Vol.45Issue(2)：97-104,8.DOI:10.3969/j.issn.1006-1355.2025.02.016
语音声特征提取的总变分正则化流形学习方法
Total Variation Regularization Manifold Learning Method for Speech Feature Extraction
摘要
Abstract
The speech signal has significant time-frequency sparsity,time-variability and high-dimensional nonlineari-ty.In order to characterize and extract its acoustic features effectively,a total variational regularized manifold learning meth-od was proposed.Based on the local linear embedding algorithm,the quadratic Fourier transform was performed successive-ly for pre-processed speech signals.Then the long-term amplitude features were extracted through statistical analysis,and the sound feature vectors containing the short and long amplitude features were constructed to generate the high-dimensional feature matrix.Finally,a mathematical model of learning sound feature extraction based on the weight value energy minimi-zation constraint of the total variation regularization manifold was constructed.The optimal weight was obtained by convex optimization,and the low-dimensional manifold of speech sound features was analyzed.Through the analysis and compari-son of the methods,it was concluded that the proposed method not only defines the physical meaning of acoustic characteris-tic manifolds and avoids the distortion of manifolds,but also greatly reduces the amount of numerical calculation and im-proves the calculation speed,which provides a technical support for machine learning and pattern recognition of intelligent speech.关键词
声学/语音声信号/正则化流形/总变分/高维特征矩阵/k邻域/声特征提取Key words
acoustics/speech sound signal/regularized manifold/total variation/high dimensional feature matrix/k neighbors/acoustic feature extraction分类
信息技术与安全科学引用本文复制引用
张开业,赵化良,刘志红,徐希鑫,李建华..语音声特征提取的总变分正则化流形学习方法[J].噪声与振动控制,2025,45(2):97-104,8.基金项目
山东省自然科学基金资助项目(ZR2023MF018) （ZR2023MF018）
国家自然科学基金资助项目(61871447) （61871447）
