Low-complexity silence detection method

Abstract

The invention discloses a low-complexity silence detection method. The method comprises the following steps of (1) sampling and processing an analog audio signal to obtain a sampled signal X[n]; (2) performing low-pass, band elimination and high-pass filtering processing on the audio sampled signal X[n] respectively at the same time to obtain sub-signals X1[n], X2[n] and X3[n]; (3) performing real-time framing processing on the three sub-signals obtained in the step (2) at the same time, and calculating the energy values of the framed sub-signals in each frame; (4) calculating the VAD (Voice Activity Detection) decision variant Dk of X[n] in each frame through the energy values of the sub-signals obtained in the step (3) in each frame; (5) setting first X frames of X[n] as mute frames after startup by default, inputting the VAD decision variant Dk into a decision device for deciding to obtain a VAD initial decision value Gk of X[n] in each frame; and (6) smoothing Gk to obtain a VAD finial decision result GGk of X[n] in each frame. The detection method has the advantages of low complexity, high accuracy, high adaptability and the like.

Description

technical field [0001] The invention relates to an audio signal detection method, in particular to a low-complexity silence detection method. Background technique [0002] The frequency band of short-wave communication is 3MHz-30MHz. Due to the increasing number of users using this frequency band, the phenomenon of co-channel interference is becoming more and more serious. Often contains noise signals with strong power, which greatly affects the answering effect. A large number of experiments have shown that the background noise type of the shortwave receiver is additive white Gaussian noise, and its power level will change slowly (second level) over time. During communication, periods of sound and periods of silence often appear alternately. Especially in the process of short-wave voice communication, unlike the working mode of mobile phones, the short-wave receiver will always be in the power-on state. If it is not processed, the sound player will keep playing very noisy...

AI Technical Summary

Problems solved by technology

The composite features of audio data include zero-crossing rate, short-term energy value, and Mel-scale cepstral coefficients based on variable-resolution spectrum. The extraction of composite features of audio data requires complex calculations, so it is not suitable for low-end products.

Images