Frequency-domain blind deconvolution method for voice signal

Abstract

The invention discloses a frequency-domain blind deconvolution method for a voice signal, comprising the following steps of: converting a time-domain convolution mixed voice signal to a frequency domain and then performing blind separation; converting and transforming the time-domain convolution mixed voice signal to a frequency-domain linear instantaneous mixture model via windowed Fourier transform according to the short-time stability of the voice signal; after performing pre-processing such as filtering and whitening in the frequency domain, realizing segmented blind separation for the voice signal by adopting a method of the approximate joint diagonalization of correlation matrices under different time delays; and after solving the problem of the fuzziness of the blind separation for the signal, performing segmented recombination for the separated signals in the time domain via inverse Fourier transform. Via the frequency-domain blind deconvolution method disclosed by the invention, a good separation effect is realized for 2*2 real-time recoded mixed voice signal, and the recognition accuracy of the voice signal of a human-computer interaction system in an environment with the speech interference of other people can be efficiently improved.

Description

technical field [0001] The invention belongs to the field of voice signal extraction and recognition in multimedia information processing, and in particular relates to a voice signal frequency domain blind deconvolution method, which can be applied to human-computer interaction scenes to improve the interactive recognition rate. Background technique [0002] After more than 60 years of development of automatic speech recognition technology, the recognition rate has exceeded 95% in a noise-free or interference-free environment. But in the actual application environment, especially when two or more speakers speak at the same time, the speech recognition rate drops suddenly, which greatly limits the application of this technology in human-machine interaction (Human-Machine Interaction, HMI). The human auditory system can obtain information of interest in a noisy environment, but it is difficult for a robot in a human-computer interaction environment to have this ability. Blind...

AI Technical Summary

Problems solved by technology

[0006] 1) Most of the algorithms are obtained under certain limited conditions, the separation effect is not ideal, the signal cross interference is large after separation, and the robustness is not high

[0007] 2) In the process of human-computer interaction in the real environment, the recognition accuracy rate is not high

[0008] 3) The existing algorithm has slow search speed and poor real-time performance, and cannot be well applied to real-time human-computer interaction scenarios

Images