Voice detection method and voice detection device for portable terminal

Abstract

The invention provides a voice detection method and a voice detection device for a portable terminal. The voice detection method includes: detecting the motion of the portable terminal; and setting a voice energy threshold for voice recognition according to the detected change of the motion relative to a voice sound source. By detecting the movement of the portable terminal and according to the change of the movement relative to the sound source of the speech, dynamically set the speech energy threshold for speech recognition; based on the dynamically set speech energy threshold, the start point and end of the user's speech can be recognized more accurately points to improve the accuracy of speech recognition.

Description

technical field [0001] The invention relates to the technical field of voice recognition, in particular to a voice detection method and device for a portable terminal. Background technique [0002] In technologies related to speech detection such as speech recognition, it is necessary to accurately detect the start point and end point of the speech, obtain valid speech data and perform corresponding processing (for example, record and upload the recorded data to the server). [0003] In the prior art, the detection of the start point and the end point of the speech needs to refer to the preset speech energy threshold, and the time for the energy of the detected speech to change from below the speech energy threshold to above the speech energy threshold The point is considered to be the starting point of the user's voice (speech); the energy of the detected voice changes from above the voice energy threshold to below the voice energy threshold and remains unchanged for a peri...

AI Technical Summary

Problems solved by technology

It can be seen that in the speech detection process, due to the low setting of the speech energy threshold, the background sound (noise) is considered as valid data, which causes the starting point of speech recognition to be identified too early; and if the speech energy threshold is set higher, then in such as image 3 In the case shown in the case, the start point of the speech may not be detected

[0008] In addition, when portable terminals such as smartphones and tablet computers use the voice recognition function in a moving state, the received voice energy and the volume of the recording will be unstable, thereby affecting the accuracy of voice data recognition and the accuracy of the user's voice. Detection of start and end points

Images