摘要
研究了一种基于深度置信网络的语音增强算法。选取在噪声环境下传统语音增强算法中较好的LOGMMSE与OMLSA算法和基于深度置信网络的语音增强算法进行了分析比较,结果证明深度置信网络的语音增强算法在3种算法中体现出了卓越的性能,尤其对增强后的语音质量的提升远远超过前两种算法。
DBN(Deep Belief Network)was studied based on speech enhancement algorithm.We chose Log-Spectral Minimum Mean Square Error(LOGMMSE)algorithm and optimally-modified log-spectral amplitude(OM-LSA)speech estimator,which are the best traditional speech enhancement algorithms,to compare with DBN-based speech enhancement algorithm.The results indicated that the DBN-based speech enhancement algorithm demonstrated superior performance among three algorithms,especially the improvement of speech quality.
作者
阴法明
唐於烽
YIN Faming;TANG Yufeng(School of Communications Engineering,Nanjing College of Information Technology,Nanjing 210023,China;School of Information Science and Engineering,Southeast University,Nanjing 210096,China)
出处
《电子器件》
CAS
北大核心
2018年第5期1325-1329,共5页
Chinese Journal of Electron Devices
基金
国家自然科学基金项目(61673108)
江苏高校品牌专业建设工程项目(PPZY2015A092)
南京信息职业技术学院科技创新团队项目
