Noise robustness endpoint detection method based on likelihood ratio test

Abstract

The invention discloses a noise robustness endpoint detection method based on a likelihood ratio test. The improvement is achieved from the three aspects of signal to noise ratio estimation, threshold value robustness setting and trailing distortion elimination respectively, so that the suggested algorithm has a better detection property under a low signal to noise ratio environment, in particular under a non-stationary noise environment compared with the prior art. The method and a multi-observation likelihood ratio test algorithm based on harmonic wave features have similar voice boundary detection accuracy, however, the method can have better voice detection precision than the multi-observation likelihood ratio test algorithm based on the harmonic wave features, and therefore it can be proved that the method is more excellent in performance than a traditional method. Meanwhile, the method has the similar performance under the 15dB and 25dB signal to noise ratio, and it shows that the method has good robustness to noise. The noise robustness endpoint detection method can be used as an important and effective method for front end preprocessing of a voice recognition system or a voiceprint recognition system in an actual environment, and thus good application value can be achieved.

Description

technical field [0001] The invention discloses a noise robust endpoint detection method based on likelihood ratio test, which relates to the fields of speech processing and signal processing. Background technique [0002] Voice endpoint detection (VAD) is a very critical part of speech processing related technologies. It can be used not only for speech / non-speech detection in speech enhancement, but also for feature extraction and speech signal reverberation. The existing speech signal endpoint detection algorithms are mainly divided into three categories: endpoint detection methods based on time domain, endpoint detection methods based on frequency domain and endpoint detection methods based on model statistics. [0003] In practical applications, high-precision speech endpoint detection plays an extremely important role in subsequent speech enhancement, endpoint detection, speech recognition or voiceprint recognition. However, there are still some problems and deficiencie...

AI Technical Summary

Problems solved by technology

However, there are still some problems and deficiencies in the existing speech endpoint detection technology, especially in the actual channel environment, because the spectral characteristics of the unvoiced and fricative components of the speech signal are very similar to the noise, and most of the existing endpoint detection algorithms The distinction between speech and noise is based on the syllable characteristics of the speech itself, so in the process of detecting endpoints, the start or end of the speech may be lost, resulting in a truncation effect

At the same time, most algorithms cannot fully retain all speech information, and when the signal-to-noise ratio decreases, the detection performance will also decrease significantly

Images