基于多任务自适应知识蒸馏的语音增强  

Speech Enhancement Based on Multi-Task Adaptive Knowledge Distillation

作　　者：张刚敏 李雅荣 贾海蓉[1] 王鲜霞 段淑斐[1] ZHANG Gangmin;LI Yarong;JIA Hairong;WANG Xianxia;DUAN Shufei(College of Electronic Information and Optical Engineering,Taiyuan University of Technology,Jinzhong 030600,China;College of Mathematics,Taiyuan University of Technology,Jinzhong 030600,China)

机构地区：[1]太原理工大学电子信息与光学工程学院,山西晋中030600 [2]太原理工大学数学学院,山西晋中030600

出　　处：《太原理工大学学报》2024年第4期720-726,共7页Journal of Taiyuan University of Technology

基　　金：国家自然科学基金资助项目(12004275);Shanxi Scholarship Council of China(2020-042);山西省自然科学基金资助项目(20210302123186)。

摘　　要：【目的】提出一种多任务自适应知识蒸馏的语音增强算法,旨在解决复杂模型在时间和硬件等计算成本方面带来的问题,同时提高语音增强算法的性能。【方法】首先,采用知识蒸馏的思想来解决现有的语音增强模型过于庞大、参数多造成计算成本上升问题;其次,充分考虑不同时频单元之间的差异,引入加权因子来优化传统损失函数提升学生网络性能;为了避免教师网络预测的不确定性影响学生网络的性能,构建多任务自适应学习的知识蒸馏网络,可以更好地利用不同任务之间的关联性优化模型。【结果】实验仿真结果表明,所提出的算法在减少参数量、缩短计算时间的同时,还能有效提高语音增强模型的性能。【Purposes】In order to solve the computational cost problem of complex model in time and hardware,and improve the performance of speech enhancement algorithm,a speech enhancement algorithm using multi-task adaptive knowledge distillation is proposed.【Methods】First,the idea of knowledge distillation is adopted to solve the problems that the existing speech enhancement model is too large,has many parameters,and has high calculation cost.Second,the differences between different time-frequency units are fully considered,and the weighting factor is introduced to optimize the traditional loss function to improve the network performance of students.In order to avoid the uncertainty of teacher network prediction affecting the performance of student network,the knowledge distillation network of multi-task adaptive learning is built to better utilize the correlation between different tasks to optimize the model.【Findings】The simulation results show that the proposed algorithm can effectively improve the performance of speech enhancement model while reducing the number of parameters and shortening the calculation time.

关 键 词：语音增强 知识蒸馏 多任务自适应学习 加权损失函数 

分 类 号：TN912.3[电子电信—通信与信息系统]