A Speech Enhancement System Based On The Auditory Characteristics And The Speech Spectrum Characteristics

The speech enhancement is an important practical technology in speech signal processing. Presently, the speech enhancement technology has applied in many fields such as speech processing, communication, and multimedia technology.This paper studies the speech enhancement technology, and proposes an improved spectral subtraction method based on the auditory characteristics and the speech spectrum characteristics. It reduces the"music noise"using the human auditory characteristics, and enhances the hearing quality using the speech spectrum distribution characteristics in the time-frequency dimension.According to the characteristics of the human pronunciation and the speech spectrum distribution in the time-frequency dimension, the paper finds out that there is a shortcoming of the speech enhancement system which is based on the masking properties of human auditory. The human ear is very sensitive to the speech in the frequency band from 2kHz to 4kHz, but the speech power spectrum in this frequency band is very small. So, the subtraction parameter will increase and will destroy the speech information of this frequency band. According to the speech spectrum distribution characteristics in the time-frequency dimension, this paper proposes the posterior Bark signal-to-noise ratio to estimate the speech content, designs the subtraction parameter again in the frequency band from 2kHz to 4kHz.The simulations show that the method in this paper can increase the output...