Voice end detection method based on energy and harmonic

Abstract

This invention relates to a phone end point test method based on energies and harmonic waves including the following steps: pre-processing digitalized sound signals and storing them in a section of buffer storage, shifting and updating the buffer storage and regulating the threshold value every time inputting a signal to judge the Tstart based on the energy, then searching for signals with sonant harmonic wave character from the buffer storage, if sonant is found, then the accurate Tstart is searched, on the contrary, the phone Tstart is searched based on the energy then to search for the phone terminal point according to the signal energy. Advantages: this invention can adjust accuracy based on noise strength so it is adaptive to the S / N ratio of input signals and energy test sphere is rather wide so phones with weak energy will not be omitted.

Description

technical field [0001] The invention relates to the field of automatic speech recognition, in particular to a speech endpoint detection method. Background technique [0002] The input signal of an automatic speech recognition system is usually speech with noise. In order to prevent the signal segment without speech from entering the recognizer, to ensure system performance and reduce computing overhead, it is necessary to detect the start and end of the user's speech in the signal. This The process is called endpoint detection. [0003] Commonly used endpoint detection algorithms can be divided into rule-based and model-based. Rule-based methods generally use signal energy, zero-crossing rate, cepstrum, long-term spectrum estimation and other features to calculate the distance, and determine whether the voice exists by comparing the distance with the threshold and logical operations. Model-based methods generally build models for the statistical characteristics of noise an...

AI Technical Summary

Problems solved by technology

For example, Sahar E. Bou-Ghazale et al. proposed in 2002 the method of using energy superposition relationship and cepstrum feature for endpoint detection. This method can work stably in a stable noise environment, but this method is very sensitive to noise energy mutations, and Since the cepstral feature of speech is affected by noise, it cannot be used in the case of low signal-to-noise ratio

Another example is the endpoint detection method proposed by Arnaud Martin in 2003 that combines signal energy and voiced sound features. The disadvantage of this method is that it is difficult to track the sudden change of noise energy. In addition, it needs to detect the fundamental frequency of the signal before determining the voiced sound, so It is easy to be interfered by the octave frequency of the pitch tone. At the same time, the algorithm uses a comb filter in the frequency domain, so it is difficult to search for voiced sounds in complex noise environments

Images