Robust playback speech detection method
A voice detection and robust technology, which is applied in the field of robust playback voice detection, can solve the problems that cannot be taken into account at the same time, the system robustness is not good, and the accuracy is affected, so as to improve the detection accuracy and remove the channel influence , to avoid the effect of abnormal interference
Inactive Publication Date: 2019-04-09
NINGBO UNIV
View PDF5 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0004] However, the current detection methods can only express low-frequency or high-frequency information alone, and cannot take into account both at the same time, resulting in poor system robustness.
More importantly, these algorithms cannot fully consider the impact of feature variability, and most current detection methods focus on improving back-end modeling or developing new features while ignoring feature variability, especially the impact of playback channel variability
In actual scenarios, attackers use a variety of performance parameters such as recording devices and playback devices, which cause the playback channel to change continuously with the change of devices, and the channel variability is the most influential to the detection of replay attacks. Removing variable channel features will seriously affect the accuracy of detection
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
[0062] Embodiments of the present invention are described in detail below, examples of which are shown in the drawings, wherein the same or similar reference numerals designate the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the figures are exemplary only for explaining the present invention and should not be construed as limiting the present invention.
[0063] 1-6 show schematic diagrams corresponding to various operation stages corresponding to the preferred embodiment of the robust playback voice detection method of the present application. This method first analyzes the difference between the real speech and the playback speech on the frequency sub-band, then extracts the cepstrum feature for the sub-band with difference, and finally uses the normalization method to post-process the cepstrum to eliminate the influence of the channel.
[0064] Specifically, the method includes,
[00...
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 a robust playback speech detection method. The method comprises the following steps: 1, analyzing the difference between frequency subbands of real speech and playback speech;2, selecting a stopband filter according to an analysis result, and extracting cepstrum features of differential subbands after filtering speech signals to be detected by the stopband filter, to obtain the cepstrum features of stopband frequency; 3, removing channel influence in the cepstrum features of the stopband frequency by subtracting a mean value, and performing normalization; and 4, training the cepstrum features obtained in the step 3 by using a Gaussian mixture model, calculating a likelihood ratio, comparing the likelihood ratio with a threshold value, and judging that the speech signals to be detected are the playback speech or the real speech. The robust playback speech detection method has the advantages of high detection accuracy and high robustness.
Description
technical field [0001] The invention relates to the field of intelligent control, in particular to a robust playback voice detection method. Background technique [0002] The automatic speaker verification system (Automatic Speaker Verification, ASV) is widely used in the fields of life and finance because of its advantages of high security, convenient acquisition and remote access. While the technology continues to develop, the threat of various spoofed voices to the ASV system is also becoming more and more serious. Among them, the most deceptive and the most convenient to operate is the playback voice. Its generation process is shown in Figure 1(b), and Figure 1(a) shows the real speech generation process. It can be seen that the real voice is the voice obtained when the target speaker authenticates the ASV system, and the playback voice is the voice that the attacker secretly recorded the target speaker's voice and played back in front of the ASV system. [0003] With...
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 Patent Type & Authority Applications(China)
IPC IPC(8): G10L17/20G10L25/24G10L25/27G10L25/51
CPCG10L17/20G10L25/24G10L25/27G10L25/51
Inventor 王让定林朗严迪群
Owner NINGBO UNIV