Unlock instant, AI-driven research and patent intelligence for your innovation.

Voice activation detection method and device

A voice activation detection, voice signal technology, applied in the field of communication, can solve the problem of insufficient robustness

Active Publication Date: 2016-09-07
SPREADTRUM COMM (SHANGHAI) CO LTD
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

None of the above algorithms can make full use of the characteristics of speech signals, and their robustness in noisy environments is not high enough

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 more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Voice activation detection method and device
  • Voice activation detection method and device
  • Voice activation detection method and device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0183] Embodiment 1: Monotone component X A [k] Corresponding frequency T f [k] is kf s / N, where N is the window function length used to obtain the audio signal spectrum for spectrum analysis, f s is the sampling frequency of the audio signal. This is an approximate representation.

Embodiment 2

[0184] Embodiment 2: Monotone component X A [k] Corresponding frequency T f [k] for k f f s / N. Here, k f corresponds to X A [k-1], X A [k] and X A [k+1] Do second-order polynomial fitting to obtain its highest point. The second-order polynomial fitting is to make the second-order polynomial curve ax 2 +bx+c=y through three points {k-1, x A [k-1]},{k,x A [k]},{k+1,X A [k+1]}, the maximum value of the curve will appear at

[0185] k f = x = - b 2 a = k + X A [ k - 1 ] - X A [ k + 1 ] ...

Embodiment 3

[0188] Embodiment 3: Monotone component X A [k] Corresponding frequency T f [k] for k f f s / N. Here, k f Corresponds to using X A [k-1], X A [k] and X A [k+1] is the highest point obtained by isosceles triangle matching.

[0189] Isosceles triangle matching is to make three points {k-1, X A [k-1]}, {k, x A [k]}, {k+1, X A [k+1]} On the two symmetrical sides of the isosceles triangle, the base of the isosceles triangle is parallel to the axis corresponding to the index. Optionally, the magnitude value X for isosceles triangle matching A [k-1], X A [k] and X A [k+1] can be replaced by their log domain values. Figure 4 A schematic diagram of an embodiment of the isosceles triangle matching of the present invention, wherein X A [k-1]A [k+1].

[0190] If X A [k-1]A [k+1], then the vertices of the isosceles triangle appear at

[0191] k f = k + 1 2 - ...

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

No PUM Login to View More

Abstract

The invention discloses a voice activation detection method and device. Wherein in the voice activation detection method, detect the monotone component in the audio signal, and place the monotone component in the monotone component set; calculate the harmony criterion of the continuous harmonic segment in the monotone component set; use the maximum harmony criterion as A detection criterion, if the detection criterion is greater than a discrimination threshold, it is determined that the audio signal is a speech signal. By detecting continuous homophonic segments in the audio signal, estimating the harmony of each continuous homophonic segment, and using the harmony criterion to judge whether there is a speech signal, thereby improving the accuracy and robustness of voice activation detection in non-stationary noise environments sex.

Description

technical field [0001] The invention relates to the communication field, in particular to a voice activation detection method and device. Background technique [0002] Voice Active Detection (VAD for short) is the basis of digital voice processing technology, which provides a judgment on whether there is a voice signal in an audio signal. Voice activation detection is widely used in the fields of speech coding, speech enhancement and denoising, speech recognition, etc. To improve the efficiency of coding; for speech enhancement and denoising, speech activation detection makes it possible to estimate the noise of speech gaps and the signal-to-noise ratio of speech segments; good speech activation detection can greatly improve the performance of speech recognition. Accuracy. [0003] Although voice activation detection is so basic and important, and its implementation algorithms are diverse, its accuracy, robustness and real-time performance are still extremely difficult pro...

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
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G10L25/78
Inventor 吴晟林福辉徐晶明蒋斌
Owner SPREADTRUM COMM (SHANGHAI) CO LTD