Method and device for voice signal processing according to frequency domain energy

Abstract

Provided are a method and apparatus for processing speech signal according to frequency domain energy. The method for processing speech signal according to frequency domain energy comprises: receiving an original speech signal comprising a first speech frame and a second speech frame that are adjacent (101); performing Fourier transform on the first speech frame and the second speech frame to obtain a first frequency domain signal and a second frequency domain signal respectively (102); obtaining frequency domain energy distributions of the first speech frame and the second speech frame (103); obtaining a frequency domain energy relevance coefficient of the first speech frame and the second speech frame (104); and segmenting the original speech signal according to the frequency domain energy relevance coefficient (105). A problem of insufficiently high accuracy of a segmentation result of a speech signal caused by the influence of phonemic features of the speech signal or relatively strong noise during fine segmentation of the speech signal can be solved.

Description

technical field [0001] Embodiments of the present invention relate to speech signal processing technologies, and in particular to a method and device for processing speech signals according to frequency domain energy. Background technique [0002] When evaluating the quality of a speech signal or performing speech recognition, it is often necessary to finely segment the speech signal. [0003] In the prior art, the segmentation of the voice signal is mainly to analyze the sudden change of the time-domain energy in the voice signal, and segment the voice signal according to the time change point of the energy mutation; if there is no change, the voice signal is not to segment. [0004] However, when the speech signal changes, due to the characteristics of the phoneme itself or the influence of strong noise, the energy in the time domain does not necessarily change abruptly. Therefore, the accuracy of the voice signal segmentation result in the prior art is not high. Conte...

AI Technical Summary

Problems solved by technology

[0005] Embodiments of the present invention provide a method and device for processing speech signals according to frequency-domain energy, so as to solve the problems caused by the characteristics of the speech signal phoneme itself or the influence of strong noise when the speech signal is finely segmented. The problem of low accuracy of signal segmentation results

Images