Lawyer talk room voice transcription device with privacy protection function
Patent Information
- Application Number
- CN202610902851.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-06-23
- Publication Date
- 2026-10-09
AI Technical Summary
[0003]现有的语音转写系统在律师与嫌疑人交谈时，若隔音结构设计的过于封闭的话，会转写准确率的降低，若采用开放式结构，则门外人员可听清内容，导致律师和嫌疑人谈话的隐私容易被窃听
本发明通过数据采集模块采集语音信号数据、振动信号数据以及干扰信号数据，通过数据处理模块基于窃听入侵检测算法对振动信号特征以及干扰信号特征进行窃听处理计算得到窃听入侵效果系数，基于语音信号特征进行声源定位辅助隐私判定处理，得到声源辅助定位系数，通过隐私分析模块基于窃听入侵效果系数和声源辅助定位系数的大小确定语音是否进行转写，基于对应的语音是否进行转写执行语音转写的执行或暂停，从而能够根据判定是否存在窃听后对语音转写的执行进行处理。
Smart Images
Figure CN122889008A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of semantic recognition technology, specifically a speech-to-text device for a lawyer's interview room with privacy protection features. Background Technology
[0002] Currently, the speech-to-text systems used in lawyer meeting rooms typically consist of an omnidirectional microphone array plus a cloud-based or local ASR engine for automatic speech recognition.
[0003] Existing speech-to-text systems suffer from reduced accuracy when lawyers and suspects are conversing if the soundproofing is too enclosed, while an open structure allows outsiders to hear the conversation, making it easy for the privacy of the lawyer-suspect conversation to be compromised. Summary of the Invention
[0004] To address the shortcomings mentioned in the background section, the present invention aims to provide a speech transcription device for a lawyer's interview room with privacy protection features.
[0005] Firstly, the objective of this invention can be achieved through the following technical solution: a speech-to-text device for a lawyer's consultation room with privacy protection function, comprising: The data acquisition module is used to collect voice signal data, vibration signal data, and interference signal data, and send them to the data processing module. The data processing module is used to extract features from voice signal data, vibration signal data, and interference signal data to obtain voice signal features, vibration signal features, and interference signal features. Based on the eavesdropping intrusion detection algorithm, the vibration signal features and interference signal features are processed for eavesdropping processing calculation to obtain the eavesdropping intrusion effect coefficient. Based on the voice signal features, sound source localization is performed to assist privacy judgment processing to obtain the sound source auxiliary localization coefficient. The eavesdropping intrusion effect coefficient and the sound source auxiliary localization coefficient are sent to the privacy analysis module. The privacy analysis module is used to determine whether speech should be transcribed based on the magnitude of the eavesdropping intrusion effect coefficient and the sound source assisted localization coefficient. The speech transcription module executes or pauses speech transcription based on whether the corresponding speech should be transcribed.
[0006] In conjunction with the first aspect, in some implementations of the first aspect, the system further includes: the calculation of eavesdropping processing on vibration signal features and interference signal features based on the eavesdropping intrusion detection algorithm, and the calculation based on obtaining vibration signal risk factors and interference signal risk factors through vibration signal features and interference signal features.
[0007] In conjunction with the first aspect, in some implementations of the first aspect, the system further includes: the calculation formula for the eavesdropping processing calculation is as follows: Where K is the eavesdropping and intrusion effectiveness coefficient, Z risk G is a vibration signal risk factor. risk α represents the risk factor for interference signals, β represents the vibration confidence weighting coefficient, and β represents the interference confidence weighting coefficient.
[0008] In conjunction with the first aspect, in some implementations of the first aspect, the system further includes: the vibration signal risk factor is calculated based on the characteristics of the vibration signal, as shown in the following formula: In the formula, min() is the minimum value function, Zi is the vibration signal characteristic, T is the preset fixed duration, Z0 is the preset standard vibration characteristic coefficient, and w1 is the vibration signal environmental influence coefficient. i is the feature number of the vibration signal feature and the interference signal feature, and i = 1, 2, 3, ..., n, where n is the total number of vibration signal features and interference signal features.
[0009] In conjunction with the first aspect, in some implementations of the first aspect, the system further includes: the interference signal risk factor is calculated based on the characteristics of the interference signal, as shown in the following formula: In the formula, Gi represents the characteristics of the interference signal, G0 represents the preset standard interference characteristic coefficient, w2 represents the environmental influence coefficient of the interference signal, and γ represents the preset correlation factor between interference and speech.
[0010] In conjunction with the first aspect, in some implementations of the first aspect, the system further includes: the calculation formula for the sound source localization-assisted privacy determination processing based on speech signal features to obtain the sound source-assisted localization coefficient is as follows: In the formula, F is the sound source auxiliary localization coefficient, max() is the maximum value function, Si is the speech signal feature, S0 is the preset standard speech feature coefficient, and w3 is the speech signal environmental influence coefficient.
[0011] In conjunction with the first aspect, in some implementations of the first aspect, the system further includes: the range of values for the eavesdropping intrusion effect coefficient K and the sound source auxiliary positioning coefficient F is 0-1.
[0012] In conjunction with the first aspect, in some implementations of the first aspect, the system further includes: the analysis process of the privacy analysis module includes the following steps: When K is less than 0.3 and F is less than 0.7, there is no eavesdropping, and a transcription execution signal is sent to the speech transcription module. When K is less than 0.3 and F is less than 0.7, the equipment is abnormal and should be inspected and repaired. When K is greater than 0.3 and F is greater than 0.7, check the source of the vibration signal and send a pause transcription signal to the speech transcription module. When K is greater than 0.3 and F is less than 0.7, this checks the source of the interference signal and sends a pause transcription signal to the speech transcription module.
[0013] The beneficial effects of this invention are: This invention collects voice signal data, vibration signal data, and interference signal data through a data acquisition module. A data processing module performs eavesdropping and intrusion detection algorithms on the vibration and interference signal features to calculate the eavesdropping and intrusion effect coefficient. Based on the voice signal features, it performs sound source localization-assisted privacy determination processing to obtain the sound source auxiliary localization coefficient. A privacy analysis module determines whether speech should be transcribed based on the magnitude of the eavesdropping and intrusion effect coefficient and the sound source auxiliary localization coefficient. Based on whether speech should be transcribed, the invention executes or pauses speech transcription, thus enabling the processing of speech transcription based on whether eavesdropping has occurred. Attached Figure Description
[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Figure 1 This is a schematic diagram of the device structure of the present invention. Detailed Implementation
[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0016] Example 1: like Figure 1 As shown, a speech-to-text device for a lawyer's consultation room with privacy protection features includes: The module includes a data acquisition module, a data processing module, a privacy analysis module, and a speech-to-text module. The data acquisition module is used to collect voice signal data, vibration signal data and interference signal data, and send the collected voice signal data, vibration signal data and interference signal data to the data processing module for processing; Specifically, the data acquisition process of the data acquisition module includes the following steps: The process of acquiring voice signal data is as follows: The microphones are placed in the lawyer's interview room and enter the room through the sound pickup holes set in the outer shell of the device. They are received by the microphone array located in the Faraday cage shielded cavity. The audio signal content of the original words spoken by the lawyer and the suspect during the conversation is obtained. In this embodiment, the number of microphones can be selected as 2-4, and the sampling frequency is set to 16kHz-32kHz. The process of acquiring vibration signal data is as follows: Based on the installation of a vibration sensor within a device housing, which is then embedded into the wall of the interview room, the sensor's outer end is exposed outside the housing for signal acquisition, or it can be extended via a wired connection to the surface of the interview room door (this embodiment uses the embedded method, making data acquisition simpler and more convenient). The system collects mechanical vibration signals from the environment during the lawyer's conversation with the suspect, using these signals as vibration data. In this embodiment, the vibration sensor's sampling frequency is 1Hz-500Hz. The process of acquiring interference signal data is as follows: A power ripple acquisition unit is placed on the power line for transmitting messages in the lawyer's interview room. The AC ripple component on the power supply line is extracted by a high-pass filter, and then the ripple component is converted into a digital signal by a high-precision analog-to-digital converter. The digital signal is then converted into a spectral feature signal by a spectrum analyzer through Fourier transform. The spectral feature signal is compared with a preset standard spectral feature. If an abnormal ripple signal is found in the comparison, it is determined that there is an interference signal. The interference signal is extracted and used as interference signal data. The sampling frequency of the power ripple acquisition unit is set to 20kHz-60kHz. After receiving voice signal data, vibration signal data, and interference signal data from the data acquisition module, the data processing module performs data processing. Specifically, the data processing module's processing procedure includes the following steps: Feature extraction was performed on the speech signal data, vibration signal data, and interference signal data respectively to obtain speech signal features, vibration signal features, and interference signal features; Based on the eavesdropping intrusion detection algorithm, the characteristics of vibration signal and interference signal are processed and calculated to obtain the eavesdropping intrusion effect coefficient. Specifically, in this embodiment, the calculation formula of the data processing module is as follows: Where K is the eavesdropping and intrusion effectiveness coefficient, Z risk G is a vibration signal risk factor. risk Here, α represents the risk factor for interference signals, β represents the vibration confidence weighting coefficient, and β represents the interference confidence weighting coefficient. Among them, the vibration confidence weight coefficient and the interference confidence weight coefficient are derived by dividing the eavesdropping impact ratio generated by combining the historical vibration signal risk and interference signal risk during the process of processing vibration signal characteristics and interference signal characteristics in this application; It should be further explained that the vibration signal risk factor is calculated based on the characteristics of the vibration signal, as shown in the following formula: In the formula, min() is the minimum value function, Zi is the vibration signal characteristic, T is the preset fixed duration, Z0 is the preset standard vibration characteristic coefficient, and w1 is the vibration signal environmental influence coefficient. The interference signal risk factor is calculated based on the characteristics of the interference signal, as shown in the following formula: In the formula, Gi is the interference signal characteristic, G0 is the preset standard interference characteristic coefficient, w2 is the interference signal environmental influence coefficient, and γ is the preset correlation factor between interference and speech. Where i is the feature number of the vibration signal feature and the interference signal feature, and i = 1, 2, 3, ..., n, where n is the total number of vibration signal features and interference signal features; The preset correlation factor between interference and speech is used to determine the impact of the generated interference signal on semantic relevance. The range is 0-1, with the closer to 1, the greater the impact and the greater the suspicion.
[0017] Furthermore, in the specific implementation process, the preset standard vibration characteristic coefficient and the preset standard interference characteristic coefficient are obtained by acquiring and extracting vibration signal characteristics and interference signal characteristics in daily life, and then by performing multiple simulation calculations and taking the average value of the data. In this embodiment, the environmental impact coefficient of vibration signal and the environmental impact coefficient of interference signal are calculated by comprehensively evaluating the influence of external factors when routinely acquiring and extracting vibration signal features and interference signal features, including human factors, machine detection, and environmental factors, etc.; human factors are those caused by improper human operation or scanning. Based on speech signal features, a sound source localization-assisted privacy assessment process is performed to obtain sound source localization coefficients. Specifically, the calculation formula for the data processing module is as follows: In the formula, F is the sound source auxiliary localization coefficient, max() is the maximum value function, Si is the speech signal feature, S0 is the preset standard speech feature coefficient, and w3 is the speech signal environmental influence coefficient. Among them, the magnitude of the sound source auxiliary localization coefficient is used to help determine the specific relevant area of voice propagation localization. If it is located in the area close to the eavesdropping intrusion effect coefficient, it indicates that there is a huge risk of being eavesdropped at this time, and the conversation should be stopped. The range of values for the eavesdropping intrusion effect coefficient K and the sound source-assisted localization coefficient F is 0-1; The data processing module sends the eavesdropping intrusion effect coefficient and the sound source-assisted localization coefficient to the privacy analysis module for analysis; The privacy analysis module is used to analyze the privacy situation after receiving the eavesdropping intrusion effect coefficient and sound source assisted localization coefficient sent by the data processing module. Specifically, the analysis process of the privacy analysis module includes the following steps: When K is less than 0.3 and F is less than 0.7, it is determined that the lawyer and the suspect are having a normal conversation and there is no eavesdropping. A transcription execution signal is then sent to the speech transcription module. When K is less than 0.3 and F is less than 0.7, it is determined that the data acquisition device is abnormal and needs to be inspected and repaired. When K is greater than 0.3 and F is greater than 0.7, it is determined that there is eavesdropping during the conversation between the lawyer and the suspect. The source of the vibration signal is then checked, and a pause transcription signal is sent to the speech transcription module. When K is greater than 0.3 and F is less than 0.7, it is determined that the voice circuit is being interfered with when the lawyer and the suspect are talking. The source of the interference signal is then checked, and a pause transcription signal is sent to the voice transcription module. When the speech transcription module receives the transcription execution signal sent by the privacy analysis module, it receives the speech signal data from the data acquisition module and performs transcription. When it receives the pause transcription signal sent by the privacy analysis module, it pauses transcription.
[0018] Based on the same inventive concept, this invention also provides a computer device, comprising: one or more processors, and a memory for storing one or more computer programs; the programs include program instructions, and the processor executes the program instructions stored in the memory. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, used to implement one or more instructions, specifically for loading and executing one or more instructions stored in a computer storage medium to implement the above-described method.
[0019] It should be further explained that, based on the same inventive concept, the present invention also provides a computer storage medium storing a computer program, which, when executed by a processor, performs the above-described method. This storage medium can be any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In the present invention, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.
[0020] The above formulas are all numerical calculations after removing dimensions. The formulas are obtained by software simulation based on a large amount of data and are closest to the real situation. The preset parameters and preset thresholds in the formulas are set by those skilled in the art according to the actual situation or obtained by simulation based on a large amount of data.
[0021] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this disclosure. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.
[0022] The foregoing has shown and described the basic principles, main features, and advantages of this disclosure. Those skilled in the art should understand that this disclosure is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of this disclosure. Various changes and modifications can be made to this disclosure without departing from its spirit and scope, and all such changes and modifications fall within the scope of this disclosure as claimed.
Claims
1. A speech-to-text device for a lawyer's consultation room with privacy protection features, characterized in that, include: The data acquisition module is used to collect voice signal data, vibration signal data, and interference signal data, and send them to the data processing module. The data processing module is used to extract features from voice signal data, vibration signal data, and interference signal data to obtain voice signal features, vibration signal features, and interference signal features. Based on the eavesdropping intrusion detection algorithm, the vibration signal features and interference signal features are processed for eavesdropping processing calculation to obtain the eavesdropping intrusion effect coefficient. Based on the voice signal features, sound source localization is performed to assist privacy judgment processing to obtain the sound source auxiliary localization coefficient. The eavesdropping intrusion effect coefficient and the sound source auxiliary localization coefficient are sent to the privacy analysis module. The privacy analysis module is used to determine whether speech should be transcribed based on the magnitude of the eavesdropping intrusion effect coefficient and the sound source assisted localization coefficient. The speech transcription module executes or pauses speech transcription based on whether the corresponding speech should be transcribed.
2. The speech-to-text device for a lawyer's consultation room with privacy protection function according to claim 1, characterized in that, The method based on the eavesdropping intrusion detection algorithm performs eavesdropping processing calculations on vibration signal features and interference signal features, and then performs calculations based on the vibration signal risk factor and interference signal risk factor obtained through the vibration signal features and interference signal features.
3. A speech-to-text device for a lawyer's consultation room with privacy protection function according to claim 2, characterized in that, The calculation formula for the eavesdropping processing is as follows: Where K is the eavesdropping and intrusion effectiveness coefficient, Z risk G is a vibration signal risk factor. risk α represents the risk factor for interference signals, β represents the vibration confidence weighting coefficient, and β represents the interference confidence weighting coefficient.
4. A speech-to-text device for a lawyer's interview room with privacy protection function according to claim 3, characterized in that, The vibration signal risk factor is calculated based on the characteristics of the vibration signal, using the following formula: In the formula, min() is the minimum value function, Zi is the vibration signal characteristic, T is the preset fixed duration, Z0 is the preset standard vibration characteristic coefficient, and w1 is the vibration signal environmental influence coefficient. i is the feature number of the vibration signal feature and the interference signal feature, and i = 1, 2, 3, ..., n, where n is the total number of vibration signal features and interference signal features.
5. A speech-to-text device for a lawyer's interview room with privacy protection function according to claim 4, characterized in that, The interference signal risk factor is calculated based on the characteristics of the interference signal, using the following formula: In the formula, Gi represents the characteristics of the interference signal, G0 represents the preset standard interference characteristic coefficient, w2 represents the environmental influence coefficient of the interference signal, and γ represents the preset correlation factor between interference and speech.
6. A speech-to-text device for a lawyer's interview room with privacy protection function according to claim 1, characterized in that, The privacy-aided sound source localization process based on speech signal features yields the following formula for calculating the sound source localization coefficient: In the formula, F is the sound source auxiliary localization coefficient, max() is the maximum value function, Si is the speech signal feature, S0 is the preset standard speech feature coefficient, and w3 is the speech signal environmental influence coefficient.
7. A speech-to-text device for a lawyer's consultation room with privacy protection function according to claim 5, characterized in that, The eavesdropping intrusion effect coefficient K and the sound source auxiliary positioning coefficient F are both in the range of 0-1.
8. A speech-to-text device for a lawyer's consultation room with privacy protection function according to claim 1, characterized in that, The analysis process of the privacy analysis module includes the following steps: When K is less than 0.3 and F is less than 0.7, there is no eavesdropping, and a transcription execution signal is sent to the speech transcription module. When K is less than 0.3 and F is less than 0.7, the equipment is abnormal and should be inspected and repaired. When K is greater than 0.3 and F is greater than 0.7, check the source of the vibration signal and send a pause transcription signal to the speech transcription module. When K is greater than 0.3 and F is less than 0.7, this checks the source of the interference signal and sends a pause transcription signal to the speech transcription module.