A real-time decomposition/synthesis method of digital speech based on auditory perception characteristics

Abstract

The invention discloses a digital speech real-time decomposition / synthesis method based on auditory perception characteristics, and relates to the field of voice signal processing. The method comprises the following steps: forming an N-order Gammatone filter through N-stage-cascaded second-order band-pass filters, and then, constructing an arbitrary-order Gammatone digital filter model and parameters thereof; in the speech decomposition stage, decomposing an input speech into M paths of signals by adopting a floating-point algorithm or a fixed-point algorithm and through M paths of Gammatone filters; and in the speech synthesis stage, introducing time delay in a Gammatone filterbank to accord with characteristics of the human ear better, human ear basilar membrane time delay being inversely proportional to frequency, and finally, carrying out speech synthesis operation. Through reference to equiloudness curve characteristics of the human ear, the speech decomposition / synthesis method is improved, and thus the final speech synthesis effect is allowed to be close to the effect of an ideal band-pass filter. The method can be applied to speech equipment of a mobile phone, an artificial cochlea and a hearing aid and the like.

Description

technical field [0001] The invention belongs to the field of digital voice signal processing, and in particular relates to a real-time decomposition / synthesis method of digital voice based on auditory perception characteristics. Background technique [0002] In daily life, there are various noises. The performance of devices such as speech enhancement and speech recognition will deteriorate significantly in noisy environments, limiting their application scenarios. Because the human ear can still work normally in a noisy environment, and has strong sensitivity and anti-interference ability to sound. Therefore, it is urgent to realize the auditory perception characteristics of the human ear, especially the basilar membrane, in the speech signal processing system. The sensory properties of the basilar membrane of the human ear are: [0003] 1. Frequency selection characteristics: Different frequencies have corresponding resonance points on the basilar membrane. Higher freque...

AI Technical Summary

Problems solved by technology

[0037] 1. This method limits the order of the gamma-pass function to N=4, which is only a special case of the gamma-pass filter, and does not give the implementation method of the gamma-pass filter when N is other values

[0038] 2. Some key parameters of this method are obtained through simulation and lack theoretical calculation basis, mainly including parameter b and normalization parameter g n and channel group delay D m , which reduces method operability and reproducibility

This eventually leads to the suppression of speech at some frequencies of the synthesized speech

Images