
info@juzhikan.asia
西京学院,电子信息学院,710123;
摘要:病理语音自动识别可为嗓音疾病筛查和辅助评估提供非侵入式手段。针对人工声学参数依赖经验、单尺度网络难以兼顾局部扰动与整体变化的问题,本文构建基于HIVIT层次化架构的病理语音分类方法。首先对SVD与AVFAD语音进行规范化、预加重、降噪、分帧加窗和Mel谱变换;随后以二维Mel频谱为输入,通过局部、中层和全局三级特征提取逐步扩大时频感受范围,并利用移位窗口注意力实现跨窗口交互;最后融合三级特征并由Softmax输出类别。分析表明,层次化表示能够兼顾短时扰动、周期级动态和整体频谱模式,为病理语音多分类提供较完整的声学表征。
关键词:病理语音;HIVIT;层次化特征;Mel频谱;移位窗口注意力
参考文献
[1] 刘思维, 邵骏. 嗓音声学分析方法的研究现状[J]. 临床耳鼻咽喉头颈外科杂志, 2022, 36(12): 966-970, 976.
[2] 郭乐乐, 曹辉, 李涛. 有效特征参数分类正常与病理语音[J]. 声学技术, 2019, 38(5): 554-559.
[3] 朱欣程, 伍远博, 赵登煌, 等. 融合MPEG-7和声门特征的病理嗓音识别方法研究[J]. 电子器件, 2022, 45(3): 587-592.
[4] 于海畅, 许涛, 钟林堃, 等. 基于特征融合与SMOTE过采样的病理语音检测[J]. 声学技术, 2025, 44(6): 956-962.
[5] HEGDE S, SHETTY S, RAI S, et al. A survey on machine learning approaches for automatic detection of voice disorders[J]. Journal of Voice, 2019, 33(6): 947.e11-947.e33.
[6] TIRRONEN S, KADIRI S R, ALKU P. Hierarchical multi-class classification of voice disorders using self-supervised models and glottal features[J]. IEEE Open Journal of Signal Processing, 2023, 4: 80-88.
[7] LIU Z, LIN Y, CAO Y, et al. Swin Transformer: Hierarchical vision transformer using shifted windows[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision. 2021: 10012-10022.
[8] JESUS L M T, BELO I, MACHADO J, et al. The Advanced Voice Function Assessment Databases (AVFAD): Tools for voice clinicians and speech research[M]//Advances in Speech-language Pathology. InTech, 2017.
[9] ABDULMAJEED N Q, AL-KHATEEB B, MOHAMMED M A. A review on voice pathology: Taxonomy, diagnosis, medical procedures and detection techniques, open challenges, limitations, and recommendations for future directions[J]. Journal of Intelligent Systems, 2022, 31(1): 855-875.