摘要
针对传统谱减语音增强算法增强后的语音信号会残留明显的"音乐噪声"的问题,采用多频带谱减算法对其进行改进。改进算法的原理是将带噪的语音信号按照频率划分成不同的频带,并使这些频带之间互不交叠,根据频带内带有噪声的语音信号和噪声信号信噪比,利用自适应算法求得该频带的过减因子。仿真结果表明:改进多频带谱减算法的语音增强效果优于传统谱减法。
Aimed at addressing ＂music noise＂,as is evident in voice signals produced by conventional spectral subtraction speech enhancement algorithm,an improved multi-band spectral subtraction algorithm for a desired performance is proposed.The algorithm works by dividing noisy speech signal into different frequency bands according to frequency,preventing these bands from overlapping each other,and obtaining noise reduction factor for the band,depending on signal noise ratio between the speech signal with noise within the band and noise signal and using adaptive algorithm.The simulation results show that the improved multi-band spectral subtraction algorithm exhibits a better speech enhancement than conventional spectral subtraction speech algorithm.
出处
《电子设计工程》
2012年第7期144-147,共4页
Electronic Design Engineering
关键词
语音增强
谱减法
短时幅度谱
多频带
speech enhancement
spectral subtraction
short-time amplitude spectrum
multi-band
