A psychoacoustically motivated speech enhancement system

Abstract

A new method is introduced to perform enhancement of speech degraded by acoustic noise using the psychoacoustic property of masking. The goal of this algorithm is to preserve the natural quality of the noise while keeping the speech perceptually intact. Distortion masking principles based on prior work of Gustafsson are used to derive a hybrid gain function comprising a function minimizing speech distortion and another minimizing noise distortion. The system is implemented in floating-point software and was tested against several existing algorithms. In a forced choice listening test, the new system was preferred over the Enhanced Variable Rate Codec (EVRC) noise suppression algorithm in 88% of the cases. Informal listening tests showed preferable speech quality than Gustafsson's algorithm. As a front end to a vocoder, the new system was preferred over the other two by all the test subjects. Ideas on future work in speech enhancement are also explored.