A Speech Recognition Method Based on Power Spectrum Gabor Eigen Sequence Recursive Model

Abstract

The invention discloses a speech recognition method based on the power spectrum Gabor feature sequence recursive model. The basic steps of the method include: 1. Preprocessing the speech input signal; 2. Extracting the power spectrum features and dynamic spectrum Delta features respectively; 3. .Use the spatiotemporal Gabor filter to filter the spectral features, and obtain the speech feature sequence through PCA dimensionality reduction processing; 4. Construct a recursive graph based on the speech feature sequence; 5. Complete speech recognition by performing distance detection on the speech recursive model. The present invention preprocesses the speech signal, obtains the speech feature sequence through feature extraction, and then converts the speech feature sequence into a recursive model for similarity detection, which effectively solves the problem of unstable noise and low signal noise in the current automatic speech recognition system. This improves the robustness of the speech recognition algorithm because the recognition rate is not ideal and the performance is prone to deterioration in complex situations.

Description

Technical field [0001] The invention belongs to the technical field of speech recognition, and relates to a speech recognition method under a complicated background, and in particular to a speech recognition method based on a recursive model of a power spectrum Gabor characteristic sequence. Background technique [0002] Voice, as the most natural and convenient way of communication, has always been one of the most important researches in the field of human-computer communication and interaction, and automatic speech recognition (ASR) is a particularly critical technology for human-computer interaction. After years of research, ASR has come into our lives. Voice transcription, automatic translation, and mobile phone assistants are all typical representatives. But these systems are mostly dependent on the acoustic environment they are in, and their robustness is not strong. [0003] The existing speech recognition consists of two stages: one is the study of speech feature extractio...

AI Technical Summary

Problems solved by technology

[0004] For the problems existing in the above-mentioned prior art, the object of the present invention is to provide a speech recognition method based on the power spectrum Gabor feature sequence recursive model, by preprocessing the speech signal, obtaining the speech feature sequence through feature extraction, and then The speech feature sequence is converted into a recursive graph for similarity detection, which effectively solves the problems of the current automatic speech recognition system in complex situations such as unsteady noise and low signal-to-noise ratio. The recognition rate is not ideal and the performance is easy to deteriorate, thereby improving speech recognition. Algorithm Robustness

Images