Audio data desensitization method and device, equipment and storage medium
By using an automatic speech recognition engine and sliding window splicing technology, combined with the timestamps and character positions of audio data, efficient audio data desensitization is achieved, solving the problem of insufficient accuracy in audio data desensitization in existing technologies and improving the reliability of data processing.
Patent Information
- Application Number
- CN202511786328.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-01
- Publication Date
- 2026-02-24
AI Technical Summary
Existing data anonymization technologies are ineffective at handling unstructured data, especially audio data, resulting in insufficient accuracy and reliability of anonymization, and failing to meet the development and testing needs of various scenarios.
An automatic speech recognition engine is used for text conversion and timestamp segmentation. Text data is concatenated using a preset sliding window, and audio data is desensitized through sensitive entity extraction and character position data.
It improves the accuracy and reliability of audio data anonymization, ensuring data processing quality in various scenarios.
Smart Images
Figure CN121565151A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of computer technology and can be applied to the field of financial technology, especially to the field of privacy computing technology, specifically to a method, apparatus, device and storage medium for de-identifying audio data. Background Technology
[0002] Audio data anonymization technology is an important technology for protecting sensitive information in audio data in the context of ensuring privacy and security.
[0003] Current data anonymization technologies have relatively simple built-in algorithms, mostly supporting only the anonymization of structured data. They struggle to address the anonymization of unstructured data and cannot meet the development and testing needs of various types of industrial data in diverse scenarios. On the other hand, a common approach to unstructured data anonymization is to preprocess different types of unstructured data, converting them into text, and then capturing sensitive information from the text for anonymization. However, unstructured data such as audio is mostly preprocessed using artificial intelligence models before being converted into text. Due to the technical limitations of the algorithms used in the preprocessing, it is difficult to guarantee the accuracy of the converted text information. Summary of the Invention
[0004] This application provides a method, apparatus, device, and storage medium for desensitizing audio data, so as to improve the accuracy and reliability of audio data desensitization.
[0005] According to one aspect of this application, a method for de-identifying audio data is provided, the method comprising:
[0006] Based on an automatic speech recognition engine, the received audio data to be processed is converted into text and divided into timestamps to obtain at least two text data to be processed and the timestamp data of the text data to be processed.
[0007] Based on a preset sliding window, the at least two text data to be processed are concatenated using the timestamp data to obtain concatenated text data; wherein, the preset sliding window is determined based on the business scenario and historically sensitive data type of the audio data to be processed.
[0008] Sensitive entity extraction is performed on the concatenated text data to obtain the sensitive entities to be processed in the concatenated text data and the character position data of the sensitive entities to be processed in the concatenated text data;
[0009] Based on the timestamp data, the concatenated text data, the sensitive entities to be processed, and the character position data, the audio data to be processed is desensitized to obtain the target audio data.
[0010] According to another aspect of this application, an audio data desensitization device is provided, the device comprising:
[0011] The data conversion module is used to perform text conversion and timestamp division on the received audio data to be processed based on the automatic speech recognition engine, so as to obtain at least two text data to be processed and the timestamp data of the text data to be processed.
[0012] The text splicing module is used to splice at least two text data to be processed based on a preset sliding window and the timestamp data to obtain spliced text data; wherein, the preset sliding window is determined according to the business scenario and historical sensitive data type of the audio data to be processed;
[0013] The entity extraction module is used to extract sensitive entities from the concatenated text data to obtain the sensitive entities to be processed in the concatenated text data and the character position data of the sensitive entities to be processed in the concatenated text data.
[0014] The data desensitization module is used to desensitize the audio data to be processed based on the timestamp data, the concatenated text data, the sensitive entities to be processed, and the character position data to obtain the target audio data.
[0015] According to another aspect of this application, an electronic device is provided, the electronic device comprising:
[0016] One or more processors;
[0017] Memory, used to store one or more programs;
[0018] When the one or more programs are executed by the one or more processors, the one or more processors implement any of the audio data desensitization methods provided in the embodiments of this application.
[0019] According to another aspect of this application, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements any of the audio data desensitization methods provided in the embodiments of this application.
[0020] According to another aspect of this application, a computer program product is provided, including a computer program that, when executed by a processor, implements any of the audio data desensitization methods provided in the embodiments of this application.
[0021] This application utilizes an automatic speech recognition engine to perform text conversion and timestamp segmentation on received audio data to obtain at least two text data sets to be processed and their timestamp data. Based on a preset sliding window, the at least two text data sets are concatenated according to the timestamp data to obtain concatenated text data. The preset sliding window is determined based on the business scenario and historical sensitive data types of the audio data to be processed. Sensitive entities are extracted from the concatenated text data to obtain the sensitive entities to be processed and their character position data within the concatenated text data. Based on the timestamp data, the concatenated text data, the sensitive entities to be processed, and the character position data, the audio data to be processed is de-identified to obtain the target audio data. This technical solution, by combining text character positions and audio timestamps for audio data de-identification, helps improve the accuracy and reliability of audio data de-identification. Attached Figure Description
[0022] Figure 1 This is a flowchart of an audio data desensitization method according to Embodiment 1 of this application;
[0023] Figure 2 This is a flowchart of an audio data desensitization method according to Embodiment 2 of this application;
[0024] Figure 3 This is a schematic diagram of an audio data desensitization device according to Embodiment 3 of this application;
[0025] Figure 4 This is a schematic diagram of the structure of an electronic device that implements the audio data desensitization method of Embodiment 4 of this application. Detailed Implementation
[0026] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.
[0027] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0028] Furthermore, it should be noted that the collection, storage, use, processing, transmission, provision, and disclosure of the text data and timestamp data involved in the technical solution of this application all comply with the provisions of relevant laws and regulations and do not violate public order and good morals.
[0029] Example 1
[0030] Figure 1 This is a flowchart of an audio data desensitization method according to Embodiment 1 of this application. This embodiment is applicable to situations where audio data needs to be desensitized, and can be performed by an audio data desensitization device. This audio data desensitization device can be implemented in hardware and / or software, and can be configured in a computer device, such as a server. Figure 1 As shown, the method includes:
[0031] S110. Based on the automatic speech recognition engine, the received audio data to be processed is converted into text and divided into timestamps to obtain at least two text data to be processed and the timestamp data of the text data to be processed.
[0032] The automatic speech recognition engine refers to a speech recognition system that not only converts speech in audio into text but also includes a timestamp for each word or sentence in the transcription result. This timestamp indicates the start and end times of the word or sentence in the original audio. The audio data to be processed refers to the audio that needs to be anonymized. The text data to be processed refers to the text extracted sentence by sentence from the audio. The timestamp data refers to the time information corresponding to the text data to be processed. Each timestamp marks the start and end times of a word or sentence in the original audio.
[0033] For example, an automatic language recognition engine is used to convert the received audio data to be processed into text data sentence by sentence, resulting in at least two text data to be processed. A timestamp is assigned to the beginning and end of each text data to be processed, resulting in timestamp data of the text data to be processed.
[0034] S120. Based on a preset sliding window, according to the timestamp data, the at least two text data to be processed are concatenated by a sliding window to obtain concatenated text data; wherein, the preset sliding window is determined according to the business scenario and historically sensitive data type of the audio data to be processed.
[0035] In text processing, a pre-defined sliding window refers to a window size set in advance for segmenting or concatenating data. The size and step of the sliding window can be adjusted according to business needs, typically within a certain time range or text length. For example, the window can move every few words to extract a subset of the text. Concatenating text data refers to combining multiple text data sets based on timestamps and window size using a sliding window. The business scenario refers to the requirements and context of audio data processing. Historically sensitive data types refer to certain data categories that may require special attention or processing during past data processing.
[0036] Optionally, the text concatenation order of the at least two text data to be processed is determined based on the timestamp data; based on a preset sliding window, the at least two text data to be processed are concatenated using a sliding window according to the timestamp data and the text concatenation order to obtain concatenated text data.
[0037] The text concatenation order refers to the order in which the text data to be processed is concatenated using a sliding window. For example, if there are two text data to be processed, text A corresponds to timestamps from 0 to 5 seconds, while text B corresponds to timestamps from 5 to 10 seconds, then the concatenation order should be text A first, followed by text B.
[0038] For example, because sensitive information may exist in multiple clauses, multiple clauses need to be concatenated consecutively to contain the complete sensitive information. For example, a card number, 6214, 6800, 0696, 7699, may require four clauses to contain the complete card number. In order to fully identify sensitive information, the identification results of the clauses are concatenated using a sliding window. According to the preset size of the sliding window, the identification results of consecutively adjacent clauses are concatenated. For example, if the sliding window is set to 4, it means concatenating clauses 1 to 4, clause 2 to 5, clause 3 to 6, and so on. The size of the sliding window can be flexibly adjusted according to the type of sensitive information and the business scenario. In this step, the start timestamp of the first identification result in the consecutive clause identification results is used as the start of the sliding window, and the end timestamp of the last identification result in the consecutive clause identification results is used as the end of the sliding window. The concatenated text is used as the identification text of the sliding window.
[0039] S130. Perform sensitive entity extraction on the concatenated text data to obtain the sensitive entities to be processed in the concatenated text data and the character position data of the sensitive entities to be processed in the concatenated text data.
[0040] Among them, sensitive entities to be processed refer to entity information identified as sensitive in the concatenated text data, which require further processing, protection, or modification. Character position data refers to the specific position of each sensitive entity within the concatenated text data; this position data is usually given through character indexes, such as the start and end positions of a sensitive entity in the text.
[0041] Optionally, based on an entity recognition model, sensitive entities are extracted from the concatenated text data to obtain candidate sensitive entities in the concatenated text data; wherein, the entity recognition model is trained on a deep learning model according to the business scenario and historical sensitive data types of the audio data to be processed; based on post-processing rules, entity verification is performed on the candidate sensitive entities according to the concatenated text data to obtain the sensitive entities to be processed in the concatenated text data.
[0042] Candidate sensitive entities are potential sensitive entities identified from the concatenated text data during the initial sensitive entity extraction process. Post-processing rules refer to the specific specifications and methods for further processing the data after the initial processing. For sensitive entities in the concatenated text data, post-processing rules help determine how to verify, correct, filter, or replace these entities; it can include removing redundant information, verifying the accuracy of sensitive entities, and performing further protection processing on sensitive information (such as de-identification or encryption) according to actual needs.
[0043] For example, entity verification in post-processing rules can be as follows: using regular expressions to further verify whether the identified sensitive information is in a valid format; introducing a sensitive word dictionary for supplementary identification (such as common organization names, phone number prefixes, common place names, etc.); and verifying the context, such as trigger words like "ID number is" or "phone number is" to confirm sensitive word entities.
[0044] Understandably, further entity verification of the entity recognition model results through post-processing rules can effectively improve the accuracy and reliability of sensitive entity recognition.
[0045] S140. Based on the timestamp data, the concatenated text data, the sensitive entities to be processed, and the character position data, the audio data to be processed is desensitized to obtain the target audio data.
[0046] The target audio data refers to the anonymized audio data.
[0047] For example, based on the sensitive entity to be processed, the concatenated text data, and the character position data, the position of the sensitive entity to be processed in the concatenated text data is determined, and based on the timestamp data, the position is mapped onto the audio data to be processed to obtain the time period in the audio data to be processed that needs to be desensitized; the time period is then muted to obtain the target audio data.
[0048] This application embodiment utilizes an automatic speech recognition engine to perform text conversion and timestamp segmentation on received audio data to be processed, obtaining at least two text data sets to be processed and their timestamp data. Based on a preset sliding window, the at least two text data sets to be processed are concatenated according to the timestamp data to obtain concatenated text data. The preset sliding window is determined based on the business scenario and historical sensitive data types of the audio data to be processed. Sensitive entities are extracted from the concatenated text data to obtain the sensitive entities to be processed in the concatenated text data and their character position data within the concatenated text data. Based on the timestamp data, the concatenated text data, the sensitive entities to be processed, and the character position data, the audio data to be processed is desensitized to obtain the target audio data. This technical solution, by combining text character positions and audio timestamps for audio data desensitization, helps improve the accuracy and reliability of audio data desensitization.
[0049] Example 2
[0050] Figure 2This is a flowchart of an audio data desensitization method according to Embodiment 2 of this application. Based on the technical solutions of the above embodiments, this embodiment refines the process of "desensitizing the audio data to be processed according to the timestamp data, the concatenated text data, the sensitive entity to be processed, and the character position data to obtain target audio data" into "replacing the sensitive entity to be processed in the concatenated text data with placeholders to obtain desensitized text data; performing text consistency verification on the concatenated text data and the desensitized text data to obtain the text difference range between the desensitized text data and the concatenated text data; desensitizing the audio data to be processed according to the text difference range, the timestamp data, and the character position data to obtain target audio data." It should be noted that for parts not detailed in this embodiment, please refer to the relevant descriptions in other embodiments. Figure 2 As shown, the method includes:
[0051] S210. Based on the automatic speech recognition engine, the received audio data to be processed is converted into text and timestamped to obtain at least two text data to be processed and the timestamp data of the text data to be processed.
[0052] S220. Based on a preset sliding window, the at least two text data to be processed are spliced together using the timestamp data to obtain spliced text data.
[0053] S230. Perform sensitive entity extraction on the concatenated text data to obtain the sensitive entities to be processed in the concatenated text data and the character position data of the sensitive entities to be processed in the concatenated text data.
[0054] S240. Replace the sensitive entities to be processed in the concatenated text data with placeholders to obtain desensitized text data.
[0055] Placeholders are symbols or strings used to represent sensitive entities; they are representative but do not expose sensitive information. Placeholders are typically predefined, generic characters or strings used to indicate the location of sensitive entities. Anonymous text data refers to text data that has been anonymized using placeholders or specific symbols.
[0056] Specifically, sensitive entities to be processed in the concatenated text data are replaced with placeholders or specific symbols to obtain desensitized text data.
[0057] For example, the concatenated text data is: My name is Zhang San, and my ID number is 440981199001017654; the desensitized text data can be: My name is **, and my ID number is ******************.
[0058] It should be noted that when replacing sensitive entities with placeholders or specific symbols, the number of placeholders or specific symbols should be the same as the number of characters in the sensitive entity.
[0059] S250. Perform text consistency verification on the spliced text data and the desensitized text data to obtain the text difference range between the desensitized text data and the spliced text data.
[0060] Text consistency verification refers to comparing concatenated text data and anonymized text data to ensure that they maintain consistency in content structure. The text difference range refers to the range of character positions corresponding to the differences between the concatenated and anonymized text data during the text consistency verification process. In other words, during verification, the system determines the locations where the anonymized text differs from the original concatenated text; these differences are usually caused by substitution operations during the anonymization process (e.g., sensitive information is replaced by placeholders or symbols).
[0061] In one alternative implementation, if the content length of the de-identified text data is zero, or the content lengths of the de-identified text data and the concatenated text data are inconsistent, the result is returned directly.
[0062] Content length refers to the number of bytes or characters in the text.
[0063] Understandably, if the text content length is 0, it means that there is no valid data or content in the text. This may be because the input data is empty, or because some operations during processing (such as filtering, deletion, or replacement) have caused the content to be completely cleared. If the length of the processed content is inconsistent with the original content, it may mean that some data was mistakenly deleted, lost, or modified during the data anonymization process. By verifying text consistency, the quality and integrity of the data at each stage can be guaranteed. Returning the data directly when the content length is 0 or inconsistent can effectively prevent the spread of errors and ensure the correctness of the data anonymization or processing process.
[0064] S260. Based on the text difference range, the timestamp data, and the character position data, the audio data to be processed is desensitized to obtain the target audio data.
[0065] Optionally, based on the character position data, the start and end positions of the text difference interval in the concatenated text data are determined; based on the start and end positions of the difference interval and the timestamp data, the audio data to be processed is desensitized to obtain the target audio data.
[0066] The start position of the difference interval refers to the initial position of the textual differences in the original text. The end position of the difference interval refers to the final position of the textual differences in the original text.
[0067] Furthermore, the start and end positions of the difference interval are compared with the timestamp data to obtain the target time interval of the text difference interval in the audio data to be processed; the audio data in the audio data to be processed that corresponds to the target time interval is desensitized to obtain the target audio data.
[0068] The target time interval refers to the time range in the audio data that corresponds to the difference interval in the text. By comparing the difference interval in the text with the timestamp data in the audio, a precise time period (usually in seconds or milliseconds) is determined, and this time period is the target time interval.
[0069] This application embodiment utilizes an automatic speech recognition engine to perform text conversion and timestamp segmentation on received audio data to be processed, obtaining at least two text data sets to be processed and their timestamp data. Based on a preset sliding window, the at least two text data sets to be processed are concatenated according to the timestamp data to obtain concatenated text data. The preset sliding window is determined based on the business scenario and historical sensitive data types of the audio data to be processed. Sensitive entities are extracted from the concatenated text data to obtain the sensitive entities to be processed and their character positions within the concatenated text data. The sensitive entities to be processed in the concatenated text data are replaced with placeholders to obtain de-identified text data. Text consistency is verified between the concatenated text data and the de-identified text data to obtain the text difference range between the de-identified text data and the concatenated text data. Based on the text difference range, timestamp data, and character position data, the audio data to be processed is de-identified to obtain the target audio data. This technical solution, by combining text character positions and audio timestamps for audio data de-identification, helps improve the accuracy and reliability of audio data de-identification.
[0070] Example 3
[0071] Figure 3 This is a schematic diagram of an audio data desensitization device according to Embodiment 3 of this application. It is applicable to situations requiring audio data desensitization. The audio data desensitization device can be implemented in hardware and / or software, and can be configured in a computer device, such as a server. Figure 3 As shown, the device includes:
[0072] The data conversion module 310 is used to perform text conversion and timestamp division on the received audio data to be processed based on an automatic speech recognition engine, so as to obtain at least two text data to be processed and timestamp data of the text data to be processed.
[0073] The text splicing module 320 is used to splice at least two text data to be processed based on a preset sliding window and the timestamp data to obtain spliced text data; wherein, the preset sliding window is determined according to the business scenario and historical sensitive data type of the audio data to be processed;
[0074] The entity extraction module 330 is used to extract sensitive entities from the concatenated text data to obtain the sensitive entities to be processed in the concatenated text data and the character position data of the sensitive entities to be processed in the concatenated text data.
[0075] The data desensitization module 340 is used to desensitize the audio data to be processed based on the timestamp data, the concatenated text data, the sensitive entities to be processed, and the character position data to obtain the target audio data.
[0076] This application embodiment utilizes an automatic speech recognition engine to perform text conversion and timestamp segmentation on received audio data to be processed, obtaining at least two text data sets to be processed and their timestamp data. Based on a preset sliding window, the at least two text data sets to be processed are concatenated according to the timestamp data to obtain concatenated text data. The preset sliding window is determined based on the business scenario and historical sensitive data types of the audio data to be processed. Sensitive entities are extracted from the concatenated text data to obtain the sensitive entities to be processed in the concatenated text data and their character position data within the concatenated text data. Based on the timestamp data, the concatenated text data, the sensitive entities to be processed, and the character position data, the audio data to be processed is desensitized to obtain the target audio data. This technical solution, by combining text character positions and audio timestamps for audio data desensitization, helps improve the accuracy and reliability of audio data desensitization.
[0077] Optionally, the data anonymization module 340 includes:
[0078] An entity replacement unit is used to replace the sensitive entities to be processed in the concatenated text data with placeholders to obtain desensitized text data.
[0079] A text verification unit is used to perform text consistency verification on the concatenated text data and the desensitized text data to obtain the text difference range between the desensitized text data and the concatenated text data;
[0080] The data desensitization unit is used to desensitize the audio data to be processed based on the text difference range, the timestamp data, and the character position data to obtain the target audio data.
[0081] Optionally, the data anonymization unit includes:
[0082] The position determination subunit is used to determine the start position and end position of the text difference interval in the concatenated text data based on the character position data.
[0083] The data desensitization subunit is used to desensitize the audio data to be processed based on the start position of the difference interval, the end position of the difference interval, and the timestamp data to obtain the target audio data.
[0084] Optional, data anonymization subunit, specifically used for:
[0085] The start and end positions of the difference interval are compared with the timestamp data to obtain the target time interval of the text difference interval in the audio data to be processed.
[0086] The audio data corresponding to the target time interval in the audio data to be processed is de-identified to obtain the target audio data.
[0087] Optional, text concatenation module 320, specifically used for:
[0088] Based on the timestamp data, determine the text concatenation order of the at least two text data to be processed;
[0089] Based on a preset sliding window, the at least two text data to be processed are spliced together using a sliding window according to the timestamp data and the text splicing order to obtain spliced text data.
[0090] Optional, entity extraction module 330, specifically used for:
[0091] Based on the entity recognition model, sensitive entities are extracted from the concatenated text data to obtain candidate sensitive entities in the concatenated text data; wherein, the entity recognition model is trained on a deep learning model according to the business scenario and historical sensitive data types of the audio data to be processed;
[0092] Based on post-processing rules, entity verification is performed on the candidate sensitive entities according to the concatenated text data to obtain the sensitive entities to be processed in the concatenated text data.
[0093] The audio data desensitization device provided in this application can execute the audio data desensitization method provided in any embodiment of this application, and has the corresponding functional modules and beneficial effects for executing each audio data desensitization method.
[0094] According to embodiments of this application, this application also provides an electronic device, a readable storage medium, and a computer program product.
[0095] Example 4
[0096] Figure 4 This is a schematic diagram of the structure of an electronic device 410 implementing the audio data desensitization method of the embodiments of this application. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or claimed herein.
[0097] like Figure 4 As shown, the electronic device 410 includes at least one processor 411 and a memory, such as a read-only memory 412 or a random access memory 413, communicatively connected to the at least one processor 411. The memory stores computer programs executable by the at least one processor. The processor 411 can perform various appropriate actions and processes based on the computer program stored in the read-only memory 412 or loaded from storage unit 418 into the random access memory 413. The random access memory 413 can also store various programs and data required for the operation of the electronic device 410. The processor 411, read-only memory 412, and random access memory 413 are interconnected via a bus 414. An input / output interface 415 is also connected to the bus 414.
[0098] Multiple components in electronic device 410 are connected to input / output interface 415, including: input unit 416, such as keyboard, mouse, etc.; output unit 417, such as various types of monitors, speakers, etc.; storage unit 418, such as disk, optical disk, etc.; and communication unit 419, such as network card, modem, wireless transceiver, etc. Communication unit 419 allows electronic device 410 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.
[0099] Processor 411 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 411 include, but are not limited to, central processing units, graphics processing units, various special-purpose artificial intelligence computing chips, various processors running machine learning model algorithms, digital signal processors, and any suitable processor, controller, microcontroller, etc. Processor 411 performs the various methods and processes described above, such as methods for de-identifying audio data.
[0100] In some embodiments, the audio data desensitization method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 418. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 410 via read-only memory 412 and / or communication unit 419. When the computer program is loaded into random access memory 413 and executed by processor 411, one or more steps of the audio data desensitization method described above may be performed. Alternatively, in other embodiments, processor 411 may be configured as the audio data desensitization method by any other suitable means (e.g., by means of firmware).
[0101] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays, application-specific integrated circuits (ASICs), application-specific standard products (ASICs), systems-on-a-chip (SoCs), payload programmable logic devices, computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.
[0102] Computer programs used to implement the methods of this application may be written in any combination of one or more programming languages. These computer programs may be provided to the processor of a general-purpose computer, a special-purpose computer, or other programmable audio data desensitization device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.
[0103] In the context of this application, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory, optical fibers, portable compact disk read-only memory, optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.
[0104] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a cathode ray tube or liquid crystal display monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).
[0105] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.
[0106] A computing system can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product within the cloud computing service system to address the shortcomings of traditional physical hosts and virtual private servers, such as high management difficulty and weak business scalability.
[0107] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this application can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this application can be achieved, and this is not limited herein.
[0108] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.
Claims
1. A method for desensitizing audio data, characterized in that, include: Based on an automatic speech recognition engine, the received audio data to be processed is converted into text and divided into timestamps to obtain at least two text data to be processed and the timestamp data of the text data to be processed. Based on a preset sliding window, the at least two text data to be processed are concatenated using the timestamp data to obtain concatenated text data; wherein, the preset sliding window is determined based on the business scenario and historically sensitive data type of the audio data to be processed. Sensitive entity extraction is performed on the concatenated text data to obtain the sensitive entities to be processed in the concatenated text data and the character position data of the sensitive entities to be processed in the concatenated text data; Based on the timestamp data, the concatenated text data, the sensitive entities to be processed, and the character position data, the audio data to be processed is desensitized to obtain the target audio data.
2. The method according to claim 1, characterized in that, Based on the timestamp data, the concatenated text data, the sensitive entities to be processed, and the character position data, the audio data to be processed is desensitized to obtain the target audio data, including: Replace the sensitive entities to be processed in the concatenated text data with placeholders to obtain desensitized text data; Perform text consistency verification on the concatenated text data and the desensitized text data to obtain the text difference range between the desensitized text data and the concatenated text data; Based on the text difference range, the timestamp data, and the character position data, the audio data to be processed is de-identified to obtain the target audio data.
3. The method according to claim 2, wherein the step of desensitizing the audio data to be processed based on the text difference interval, the timestamp data, and the character position data to obtain the target audio data includes: Based on the character position data, determine the start and end positions of the text difference interval in the concatenated text data; Based on the start position of the difference interval, the end position of the difference interval, and the timestamp data, the audio data to be processed is de-identified to obtain the target audio data.
4. The method according to claim 3, characterized in that, The step of desensitizing the audio data to be processed based on the start position of the difference interval, the end position of the difference interval, and the timestamp data to obtain the target audio data includes: The start and end positions of the difference interval are compared with the timestamp data to obtain the target time interval of the text difference interval in the audio data to be processed. The audio data corresponding to the target time interval in the audio data to be processed is de-identified to obtain the target audio data.
5. The method according to claim 1, characterized in that, Based on a preset sliding window, the at least two text data to be processed are concatenated using the timestamp data to obtain concatenated text data, including: Based on the timestamp data, determine the text concatenation order of the at least two text data to be processed; Based on a preset sliding window, the at least two text data to be processed are spliced together using a sliding window according to the timestamp data and the text splicing order to obtain spliced text data.
6. The method according to claim 1, characterized in that, Sensitive entity extraction is performed on the concatenated text data to obtain the sensitive entities to be processed in the concatenated text data, including: Based on the entity recognition model, sensitive entities are extracted from the concatenated text data to obtain candidate sensitive entities in the concatenated text data; wherein, the entity recognition model is trained on a deep learning model according to the business scenario and historical sensitive data types of the audio data to be processed; Based on post-processing rules, entity verification is performed on the candidate sensitive entities according to the concatenated text data to obtain the sensitive entities to be processed in the concatenated text data.
7. An audio data desensitization device, characterized in that, include: The data conversion module is used to perform text conversion and timestamp division on the received audio data to be processed based on the automatic speech recognition engine, so as to obtain at least two text data to be processed and the timestamp data of the text data to be processed. The text splicing module is used to splice at least two text data to be processed based on a preset sliding window and the timestamp data to obtain spliced text data; wherein, the preset sliding window is determined according to the business scenario and historical sensitive data type of the audio data to be processed; The entity extraction module is used to extract sensitive entities from the concatenated text data to obtain the sensitive entities to be processed in the concatenated text data and the character position data of the sensitive entities to be processed in the concatenated text data. The data desensitization module is used to desensitize the audio data to be processed based on the timestamp data, the concatenated text data, the sensitive entities to be processed, and the character position data to obtain the target audio data.
8. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the audio data desensitization method as described in any one of claims 1-6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the audio data desensitization method as described in any one of claims 1-6.
10. A computer program product comprising a computer program that, when executed by a processor, implements the method for desensitizing audio data according to any one of claims 1-6.