Time unifying method and device for audio data and reference signal

Abstract

The invention discloses a time unifying method and a time unifying device for audio data and a reference signal and belongs to the communication field. The method comprises the steps as follows: obtaining the audio data which is collected currently and obtaining the reference signal corresponding to the audio data from a reference queue; calculating a first time delay estimation expected value according to a correlation value between the audio data and the reference signal; detecting whether there is beating effect between the audio data and the reference signal in the reference queue according to a second time delay estimation expected value and the first time delay estimation expected value, wherein the second time delay estimation expected value is a time delay estimation expected value which is obtained by last calculation before current time; unifying a time relation between the audio data and the reference signal in the reference queue if the beating effect is existed. The device comprises an obtaining module, a calculating module, a detecting module and a unifying module. The time unifying method and time unifying device of the invention improve robustness of echo cancellation.

Description

technical field [0001] The invention relates to the communication field, in particular to a method and device for time alignment of audio data and reference signals. Background technique [0002] Voice call software is an application capable of making Internet calls or Internet audio conferences, and is currently installed in mobile terminals by more and more users. At the same time, as the physical size of mobile terminals develops toward miniaturization, the acoustic coupling problem between the speaker and the microphone becomes more prominent, so that the sound collected by the microphone includes the sound played by the speaker, so that the sound heard by the other party feels There is an echo of one's own speech, which affects the call effect of both parties. [0003] In order to avoid affecting the call effect of both parties, it is necessary to eliminate the echo contained in the audio data collected by the microphone. The specific process is as follows: the mobile ...

AI Technical Summary

Problems solved by technology

[0004] Among them, it should be noted that: the voice call software works on the operating system, and the echo in the audio data is eliminated by the application processor; when the load of the application processor is large, a part of the received audio data will be discarded and not input Go to the reference queue or discard some of the audio data collected by the microphone without echo cancellation, so it may appear that after the microphone collects a frame of audio data, there is no reference signal corresponding to the frame of audio data in the reference queue, that is, a beat phenomenon occurs; and The reference signal most relevant to the frame of audio data found by the mobile terminal from the reference queue is not the reference signal corresponding to the frame of audio data, and the echo in the frame of audio data cannot be eliminated by using the reference signal

[0005] After the beating phenomenon occurs, because the mobile terminal buffers the audio data as a reference signal in the reference queue every time it receives the audio data, and the reference signal with the longest buffer time in the reference queue is removed from the reference queue, so that when there is a gap After shooting, there may be no reference signal corresponding to the audio data collected by the microphone each time in the reference queue, causing the echo to continue forever, reducing the robustness of echo cancellation

Images