CN-121983065-A - Touch screen voice awakening system based on voiceprint recognition

CN121983065ACN 121983065 ACN121983065 ACN 121983065ACN-121983065-A

Abstract

The invention discloses a voice-print recognition-based touch screen voice wake-up system, which relates to the technical field of voice data processing and comprises an intelligent control terminal for controlling data transmission and information interaction among various modules. According to the invention, various calculations are performed on awakening voice data input by a user in advance, the channel shape, acoustic structure characteristics and characteristic vectors which are unique to the awakening voice data of the user are determined, after the touch screen acquires surrounding voice data, the first voice data to be analyzed is subjected to characteristic emphasis so as to improve the characteristic intensity, finally, the channel shape, acoustic structure characteristics and characteristic vectors of the voice to be verified are verified through the channel shape, acoustic structure characteristics and characteristic vectors of the awakening voice data, whether the first voice data to be analyzed are sent by the user recording the awakening voice data or not is determined, if yes, the touch screen is awakened, if not, the touch screen is not awakened, and the privacy of the user is ensured not to be leaked.

Inventors

WU GUANGHUAI
FANG MINGHUI
SU YU
XU ZHEN

Assignees

河源市联腾物联科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260126

Claims (10)

1. A touch screen voice wake-up system based on voiceprint recognition, comprising: The intelligent control terminal is used for controlling the data transmission and information interaction among the modules, and is used for controlling the modules to perform characteristic emphasis processing, data calculation processing and calculation verification processing on the first voice data to be analyzed and determining whether the second voice data to be analyzed can wake up the touch screen; the database system is used for storing wake-up voice data of the touch screen; The voice acquisition module is used for carrying out audio data acquisition processing on the surrounding environment of the touch screen to acquire first voice data to be analyzed; The characteristic enhancement module is used for carrying out characteristic enhancement processing on the first voice data to be analyzed to obtain second voice data to be analyzed; the feature vector calculation module is used for calculating and processing the wake-up voice data to obtain feature vectors of the wake-up voice data; and the voice verification module performs calculation verification processing on the second voice data to be analyzed according to the feature vector of the awakening voice data, and determines whether the second voice data to be analyzed can awaken the touch screen or not.
2. The touch screen voice wake-up system based on voiceprint recognition of claim 1, wherein the feature enhancement module is configured to perform feature enhancement processing on the first voice data to be analyzed, and the step of obtaining the second voice data to be analyzed specifically includes the following steps: Based on the intelligent control terminal, performing data cutting processing on the first voice data to be analyzed to obtain a plurality of groups of segmented voice data; Based on the intelligent control terminal, performing data windowing on the plurality of groups of segmented voice data to obtain a plurality of groups of windowed voice data; based on the intelligent control terminal, carrying out energy analysis processing on a plurality of groups of windowed voice data, and determining the weak energy position of the voice data and the windowed voice data without enhancing the characteristics; based on the characteristic strengthening module, carrying out characteristic aggravating treatment on the weak energy position of the voice data to obtain windowed voice data with strengthened characteristics; and based on the intelligent control terminal, combining the windowed voice data without the enhancement features and the windowed voice data with the enhancement features to obtain second analyzed voice data.
3. The voice wakeup system of touch screen based on voiceprint recognition according to claim 2, wherein the step of performing energy analysis processing on a plurality of groups of windowed voice data based on the intelligent control terminal to determine a weak energy position of the voice data and windowed voice data without enhancing features specifically comprises the following steps: based on the intelligent control terminal, performing data calculation processing on each frame of voice signals in the plurality of groups of windowed voice data to acquire an energy value of each frame of voice signals; based on the intelligent control terminal, carrying out normalization processing on the energy value of each frame of voice signal to obtain the normalized energy value of each frame of voice signal; based on the intelligent control terminal, judging and processing the normalized energy value of each frame of voice signal and the set normalized energy value threshold; if the normalized energy value of each frame of voice signal contains data smaller than the set normalized energy value threshold, setting the position of the data smaller than the set normalized energy value threshold as a weak position of the voice data energy based on the intelligent control terminal; and setting the data which is larger than or equal to the set normalized energy value threshold value as windowed voice data without enhancing the characteristic based on the intelligent control terminal.
4. The voice-print recognition-based touch screen voice wake-up system of claim 1, wherein the feature vector calculation module is configured to perform calculation processing on wake-up voice data, and obtain feature vectors of the wake-up voice data specifically includes the following steps: Based on the intelligent control terminal, carrying out data frequency domain calculation processing on the wake-up voice data, and determining the sound channel shape of the wake-up voice; based on the intelligent control terminal, carrying out data calculation processing on the wake-up voice data, and determining acoustic structural characteristics of the wake-up voice; And carrying out data calculation processing on the wake-up voice data based on the feature vector calculation module to acquire the feature vector of the wake-up voice data.
5. The voice-print recognition-based touch screen voice wake-up system of claim 4, wherein the data frequency domain calculation processing is performed on wake-up voice data based on the intelligent control terminal, and determining the channel shape of the wake-up voice specifically comprises the following steps: performing frequency domain conversion processing on the wake-up voice data based on a short-time Fourier transform algorithm to acquire frequency domain representation of the wake-up voice data; Based on the intelligent control terminal, carrying out Mel frequency conversion on the frequency domain representation of the wake-up voice data, and designing a Mel filter bank; based on the intelligent control terminal, carrying out logarithmic transformation processing on the Mel filter bank to obtain logarithmic energy of the Mel filter bank; based on discrete cosine transform algorithm, the logarithmic energy of the Mel filter bank is processed in a decorrelation way to determine the channel shape of the wake-up voice.
6. The voice-print recognition-based touch screen voice wake-up system of claim 4, wherein the data calculation processing is performed on wake-up voice data based on the intelligent control terminal, and determining the acoustic structural characteristics of the wake-up voice specifically comprises the following steps: Based on the intelligent control terminal, carrying out data calculation processing on the wake-up voice data, and determining an autocorrelation function of the wake-up voice data; Based on the intelligent control terminal, carrying out data searching processing on the autocorrelation function of the wake-up voice data, and determining the peak value of the autocorrelation function; And based on the intelligent control terminal, comparing, analyzing and processing the peak value of the autocorrelation function, and determining the fundamental frequency of the autocorrelation function.
7. The voice wake-up system of claim 6, wherein the comparing and analyzing the peak value of the autocorrelation function based on the intelligent control terminal, and determining the fundamental frequency of the autocorrelation function specifically comprises the following steps: judging the peak value of the autocorrelation function and setting a first peak value threshold value based on the intelligent control terminal; if the peak value of the autocorrelation function is greater than a set first peak threshold value; Judging the peak value of the autocorrelation function and setting a second peak value threshold value based on the intelligent control terminal; If the peak value of the autocorrelation function is larger than the set second peak value threshold value, setting the frequency of the position of the peak value larger than the set second peak value threshold value as the fundamental frequency of the autocorrelation function based on the intelligent control terminal; if the peak value of the autocorrelation function is smaller than or equal to the set second peak value threshold value, setting the frequency at the position where the set second peak value threshold value is smaller than or equal to the frequency multiplication of the autocorrelation function based on the intelligent control terminal; if the peak value of the autocorrelation function is smaller than or equal to the set first peak value threshold, the peak value of the autocorrelation function is a false peak value.
8. The voice-print recognition-based touch screen voice wake-up system of claim 4, wherein the feature vector calculation module performs data calculation processing on wake-up voice data, and the feature vector obtaining of the wake-up voice data specifically comprises the following steps: performing dimension data extraction processing on the wake-up voice data based on a convolutional neural network algorithm to obtain Gao Weisheng line features of the wake-up voice data; Based on the intelligent control terminal, carrying out data normalization processing on Gao Weisheng line features of wake-up voice data to obtain normalized data of Gao Weisheng line features; based on the intelligent control terminal, carrying out data calculation processing on the normalized data of Gao Weisheng lines of features to obtain feature vectors of wake-up voice data.
9. The voice-print recognition-based touch screen voice wake-up system of claim 1, wherein the voice verification module performs calculation verification processing on the second voice data to be analyzed according to the feature vector of the wake-up voice data, and determines whether the second voice data to be analyzed can wake up the touch screen, and the method specifically comprises the following steps: Based on the feature vector calculation module, calculating the second voice data to be analyzed to obtain the sound channel shape, the acoustic structural feature and the feature vector of the voice to be verified; and based on the intelligent control terminal, performing distance comparison processing on the feature vector of the wake-up voice data and the feature vector of the voice to be verified, and determining whether the second voice data to be analyzed can wake up the touch screen.
10. The voice-print recognition-based touch screen voice wake-up system of claim 9, wherein the intelligent control terminal-based distance comparison processing is performed on feature vectors of wake-up voice data and feature vectors of voice to be verified, and determining whether the second voice data to be analyzed can wake up the touch screen specifically comprises the following steps: Based on the intelligent control terminal, carrying out vector distance calculation processing on the feature vector of the wake-up voice data and the feature vector of the voice to be verified, and obtaining the cosine distance of the voice data; based on the voice verification module, judging the cosine distance of voice data and the set cosine distance range; If the cosine distance of the voice data is within the set cosine distance range and the sound channel shape of the voice to be verified, the acoustic structural feature of the voice to be verified, the sound channel shape of the wake-up voice and the acoustic structural feature of the wake-up voice are identical, the second voice data to be analyzed can wake up the touch screen; If the cosine distance of the voice data is not in the set cosine distance range, or the sound channel shape of the voice to be verified, the acoustic structural feature of the voice to be verified, the sound channel shape of the wake-up voice and the acoustic structural feature of the wake-up voice are not identical, the second voice data to be analyzed cannot wake up the touch screen.

Description

Touch screen voice awakening system based on voiceprint recognition Technical Field The invention relates to the technical field of voice data processing, in particular to a touch screen voice awakening system based on voiceprint recognition. Background The touch screen is an induction type display device capable of receiving input signals such as a contact, a user can directly operate on the screen through fingers or a touch pen to realize man-machine interaction, and the touch screen combines a display technology and an input technology and is widely applied to various electronic devices such as smart phones, tablet computers, ATM machines, industrial control devices, self-service terminals and the like. Some users can wake up the touch screen by using specific voice keywords, if feature analysis is not performed on the wake-up voice of the user, after other users mistakenly speak the voice keywords, the touch screen is woken up, so that other users can use the touch screen, and the privacy leakage of the user occurs. Disclosure of Invention In order to solve the technical problems, the touch screen voice awakening system based on voiceprint recognition is provided, and the technical scheme solves the problems that if the awakening voice of the user is not subjected to feature analysis, after other users mistakenly speak voice keywords, the touch screen is awakened, so that other users can use the touch screen, and the privacy of the users is leaked. In order to achieve the above purpose, the invention adopts the following technical scheme: A touch screen voice wake-up system based on voiceprint recognition, comprising: The intelligent control terminal is used for controlling the data transmission and information interaction among the modules, and is used for controlling the modules to perform characteristic emphasis processing, data calculation processing and calculation verification processing on the first voice data to be analyzed and determining whether the second voice data to be analyzed can wake up the touch screen; the database system is used for storing wake-up voice data of the touch screen; The voice acquisition module is used for carrying out audio data acquisition processing on the surrounding environment of the touch screen to acquire first voice data to be analyzed; The characteristic enhancement module is used for carrying out characteristic enhancement processing on the first voice data to be analyzed to obtain second voice data to be analyzed; the feature vector calculation module is used for calculating and processing the wake-up voice data to obtain feature vectors of the wake-up voice data; and the voice verification module performs calculation verification processing on the second voice data to be analyzed according to the feature vector of the awakening voice data, and determines whether the second voice data to be analyzed can awaken the touch screen or not. Preferably, the feature enhancing module is configured to perform feature enhancing processing on the first voice data to be analyzed, and the obtaining the second voice data to be analyzed specifically includes the following steps: Based on the intelligent control terminal, performing data cutting processing on the first voice data to be analyzed to obtain a plurality of groups of segmented voice data; Based on the intelligent control terminal, performing data windowing on the plurality of groups of segmented voice data to obtain a plurality of groups of windowed voice data; based on the intelligent control terminal, carrying out energy analysis processing on a plurality of groups of windowed voice data, and determining the weak energy position of the voice data and the windowed voice data without enhancing the characteristics; based on the characteristic strengthening module, carrying out characteristic aggravating treatment on the weak energy position of the voice data to obtain windowed voice data with strengthened characteristics; and based on the intelligent control terminal, combining the windowed voice data without the enhancement features and the windowed voice data with the enhancement features to obtain second analyzed voice data. Preferably, the step of performing energy analysis processing on a plurality of groups of windowed voice data based on the intelligent control terminal, and determining the position of weak energy of the voice data and the windowed voice data without enhancing features specifically includes the following steps: based on the intelligent control terminal, performing data calculation processing on each frame of voice signals in the plurality of groups of windowed voice data to acquire an energy value of each frame of voice signals; based on the intelligent control terminal, carrying out normalization processing on the energy value of each frame of voice signal to obtain the normalized energy value of each frame of voice signal; based on the intelligent control terminal, judging and processing