Voice activation detection method and device
A voice activation detection and voice technology, applied in voice analysis, voice recognition, instruments, etc., can solve the problems of difficult threshold and difficult activation of voice recognition equipment, and achieve the effect of reducing missed activation, ensuring performance, and high detection rate
Active Publication Date: 2018-04-17
YUTOU TECH HANGZHOU
View PDF7 Cites 6 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
In fact, it is only necessary to perform acoustic scoring on the activated speech, and calculate the acoustic score on the set activation words. At the same time, the threshold of acceptance and rejection can be set according to the score; however, the threshold is very difficult to control, because the threshold Setting too low will bring a lot of false activations, and setting the threshold too high will make it difficult for voice recognition devices to be activated
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0041] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but not as a limitation of the present invention.
[0042] Such as figure 1 As shown, the present embodiment relates to a voice activation detection method, which is applied to voice detection when a voice recognition device provided with an activation word is activated. The method mainly includes the following steps:
[0043] Step S1, performing endpoint detection on the voice data to be tested, so as to obtain voice data including voice signals.
[0044] The reason why the endpoint detection step is placed first in the method flow is that if the acoustic calculation of the speech data to be tested (continuous speech signal) is continuously performed, a large amount of resources will be wasted, and the subsequent acoustic calculation after the endpoint detection is only Performed on speech data containing speech signals, which can save computin...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention relates to the technical field of voice recognition, and particularly relates to a voice activation detection method and device. The activation detection method is applied for two times,the voice segment taken from the first time of activation judgment is inputted to a BLSTM recurrent neural network and all the frames of the whole voice are processed so that whether voice recognition equipment is activated is finally judged. In two times of activation confirmation, the threshold of the first time of activation can be set to be appropriately loose so as to guarantee high detection rate; and the second time of activation confirmation is relatively more accurate because of the known starting point, and false activation and leak activation can be simultaneously reduced by two times of activation detection, i.e. the equal error rate of activation can be effectively reduced so that the performance of activation can be more effectively guaranteed.
Description
technical field [0001] The invention relates to the field of voice recognition, in particular to a voice activation detection method and device for secondary confirmation by using a bidirectional cyclic neural network. Background technique [0002] Many speech recognition devices need to be activated by voice before they can pick up the sound and then recognize it. After the voice recognition device is activated, the voice is picked up firstly to reduce the power consumption of the device, and secondly to reduce the voice that does not need to be recognized into the voice recognition process, which in turn brings unnecessary responses. In the near-field speech recognition environment, such as the speech recognition on the mobile phone, you can enter the sound pickup state by touching or pressing buttons. In far-field recognition, or when near-field recognition is inconvenient to operate by hand, it becomes indispensable to use voice to activate the device to make the device...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More IPC IPC(8): G10L15/14G10L15/16G10L25/51
Inventor 范利春朱磊
Owner YUTOU TECH HANGZHOU