Irregular audio detection method and device, electronic equipment and storage medium

Abstract

Embodiments of the present application disclose a method and device for detecting illegal audio, an electronic device and a storage medium. The method comprises: obtaining audio to be detected, extracting human voice in the audio to be detected, and converting the human voice into text information; inputting the text information into an illegal tone detection module to detect whether there is illegal tone in the audio to be detected; wherein the illegal tone detection module detects based on at least one of tone words in the text information and the length of the audio to be detected; and if there is illegal tone in the audio to be detected, inputting the text information into a preset illegal detection model to detect whether the audio to be detected is illegal audio. The technical solution of the embodiments of the present application can quickly and accurately detect illegal audio, while saving labor cost and time cost.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and more specifically, to a method and apparatus for detecting illegal audio, an electronic device, and a computer-readable storage medium. Background Technology

[0002] With the development of multimedia technology, in user-generated content (UGC), users can create their own audio content, or directly adapt song lyrics and then distribute them online, where other users can view the audio content they share. However, some of the audio uploaded to the internet may contain inappropriate content, and works by the same artist often contain similar violations.

[0003] Therefore, before audio content is uploaded, it undergoes violation detection. Most existing violation detection methods rely on directly judging whether the text content violates regulations. This involves converting the audio content into text using speech recognition and then using the text to determine if the audio content violates regulations. Alternatively, audio similarity matching can be used. However, because audio content often contains musical melodies and background noise, the detection rate of existing violation detection methods is low due to the influence of these elements. Secondly, song content may involve many types of violations, complex scenarios, and may involve multiple languages ​​and dialects, making the judgment criteria very difficult.

[0004] Meanwhile, in existing technologies, audio content review is mainly carried out through a dual review process of machine review and manual review. There is an urgent need for a technical solution for violation detection that can reduce the labor and time costs and improve the efficiency of audio content review. This solution would help the operation and review team to prohibit and delete audio content that harms the health and security of the network, thereby building a new pan-entertainment ecosystem that promotes healthy values. Summary of the Invention

[0005] To address the aforementioned technical problems, embodiments of this application provide a method and apparatus for detecting illegal audio, an electronic device, and a computer-readable storage medium, aiming to solve the technical problems of high manpower and time costs required for existing illegal review, as well as low efficiency and accuracy of audio content review.

[0006] Other features and advantages of this application will become apparent from the following detailed description, or may be learned in part from practice of this application.

[0007] According to one aspect of the embodiments of this application, a method for detecting illegal audio is provided, including:

[0008] Acquire the audio to be detected, extract the human voice audio from the audio to be detected, and convert the human voice audio into text information;

[0009] The text information is input into the violation tone detection module to detect whether there is a violation tone in the audio to be detected; the violation tone detection module detects based on at least one of the tone words in the text information and the duration of the audio to be detected;

[0010] If the audio to be detected contains inappropriate tone, the text information is input into the preset violation detection model to detect whether the audio to be detected is inappropriate.

[0011] In another embodiment, the text information is input into the violation tone detection module to detect whether there is a violation tone in the audio to be detected, including:

[0012] Obtain the duration information of the audio to be detected, and extract the number of interjections and words in the text information, as well as the number of characters in the text information;

[0013] Calculate the first ratio of the number of modal particles to the number of words, and calculate the second ratio of the number of text characters to the duration information;

[0014] The presence of inappropriate tone in the audio to be detected is determined based on the first ratio and the second ratio.

[0015] In another embodiment, detecting whether there is an inappropriate tone in the audio to be detected based on a first ratio and a second ratio includes:

[0016] The first ratio is matched with the first detection threshold to obtain the first matching result, and the second ratio is matched with the second detection threshold to obtain the second matching result;

[0017] If the first matching result indicates that the first ratio is greater than the first detection threshold, and the second matching result indicates that the second ratio is less than the second detection threshold, it is determined that there is an illegal tone in the audio to be detected;

[0018] If the first matching result indicates that the first ratio is less than or equal to the first detection threshold, or the second matching result indicates that the second ratio is greater than or equal to the second detection threshold, it is determined that there is no illegal tone in the audio to be detected.

[0019] In another embodiment, before inputting text information into the violation tone detection module to detect whether there is a violation tone in the audio to be detected, the method further includes:

[0020] Extracting voiceprint features from human voice audio yields the target voiceprint features;

[0021] The target voiceprint features are matched with the illegal voiceprint features in the voiceprint database to obtain the feature matching results;

[0022] If the feature matching result indicates that the target voiceprint feature does not match the violation voiceprint feature, the step of inputting the text information into the preset violation tone detection module is executed to detect whether there is a violation tone in the audio to be detected.

[0023] In another embodiment, extracting voiceprint features from human voice audio to obtain target voiceprint features includes:

[0024] The human voice audio is segmented to obtain multiple human voice audio segments;

[0025] Voiceprint features of each human voice audio segment are extracted using a voiceprint model;

[0026] Calculate the average voiceprint feature of multiple voiceprint features, and use the average voiceprint feature as the target voiceprint feature.

[0027] In another embodiment, the method further includes:

[0028] If there is no inappropriate tone in the audio to be detected, the text information is input into the preset semantic model for semantic recognition processing to obtain the semantic recognition result;

[0029] The semantic recognition results determine whether the audio to be detected is illegal.

[0030] In another embodiment, after determining whether the audio to be detected is illegal audio based on the semantic recognition result, the method further includes:

[0031] If the audio to be detected is illegal, the violation category is determined based on the semantic recognition results;

[0032] The corresponding processing procedure is determined based on the type of violation, and the audio to be detected is processed according to the processing procedure.

[0033] According to one aspect of the embodiments of this application, an illegal audio detection device is provided, comprising:

[0034] The acquisition module is configured to acquire the audio to be detected, extract the human voice audio from the audio to be detected, and convert the human voice audio into text information;

[0035] The first detection module is configured to input text information into the violation tone detection module to detect whether there is a violation tone in the audio to be detected; wherein, the violation tone detection module performs detection based on at least one of the tone words in the text information and the duration of the audio to be detected;

[0036] The second detection module is configured to input text information into a preset violation detection model to detect whether the audio to be detected is a violation if there is an illegal tone in the audio to be detected.

[0037] According to one aspect of the embodiments of this application, an electronic device is provided, including: one or more processors; and a storage device for storing one or more programs, which, when executed by one or more processors, cause the electronic device to implement the aforementioned illegal audio detection method.

[0038] According to one aspect of the embodiments of this application, a computer-readable storage medium is provided, on which computer-readable instructions are stored, which, when executed by a computer's processor, cause the computer to perform the above-described illegal audio detection method.

[0039] According to one aspect of the embodiments of this application, a computer program product or computer program is provided, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the illegal audio detection method provided in the various alternative embodiments described above.

[0040] In the technical solution provided in the embodiments of this application, when it is necessary to detect audio to be detected, the human voice audio in the audio to be detected is extracted and converted into text information. The extracted human voice audio will greatly reduce the interference of background music, thereby improving the recognition accuracy of converting human voice audio into text information. This application also inputs the converted text information into a violation tone detection module to detect whether there is a violation tone in the audio to be detected. The violation tone detection module performs detection based on at least one of the tone words in the text information and the duration of the audio to be detected. When there is a violation tone in the audio to be detected, the tone words and duration are different from normal audio. Therefore, the violation tone detection module can accurately detect whether there is a violation tone in the audio to be detected. When a violation tone is detected in the audio to be detected, the text information is further input into a preset violation detection model to detect whether the audio to be detected is a violation audio, which can perform violation detection more quickly and accurately. It is evident that the technical solution adopted in this application eliminates the need for dedicated manual review, thereby reducing manpower and time costs. It also helps the operations review team to prohibit and delete audio content that harms the health and security of the network, thus building a new pan-entertainment ecosystem that promotes healthy values.

[0041] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0042] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort. In the drawings:

[0043] Figure 1 This is a schematic diagram of one implementation environment involved in this application;

[0044] Figure 2 This is a flowchart of a method for detecting illegal audio related to this application;

[0045] Figure 3 This is a flowchart of a method for detecting illegal audio in another embodiment of this application;

[0046] Figure 4 This is a flowchart of step S310 in one embodiment of this application;

[0047] Figure 5 This is a flowchart of step S220 in one embodiment of this application;

[0048] Figure 6 This is a flowchart of step S530 in one embodiment of this application;

[0049] Figure 7 This is a flowchart of a method for detecting illegal audio in another embodiment of this application;

[0050] Figure 8 This is a flowchart of step S720 in one embodiment of this application;

[0051] Figure 9 This is a flowchart of a method for detecting illegal audio in another embodiment of this application;

[0052] Figure 10 This is a block diagram of an illegal audio detection device involved in this application;

[0053] Figure 11 A schematic diagram of the structure of a computer system suitable for implementing the electronic device of the present application is shown. Detailed Implementation

[0054] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0055] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0056] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0057] It should also be noted that "multiple" as mentioned in this application refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0058] Please see Figure 1 , Figure 1 This is a schematic diagram of an implementation environment related to this application. The implementation environment includes a terminal 110 and a server 120, which communicate with each other via a wired or wireless network.

[0059] Terminal 110 runs applications for creating user-generated content and applications for playing audio and video. In some of these playback applications, user-generated content can be created directly. When a user uploads user-generated content to an application for playback, if the application is a public platform—meaning the user-generated content can be played by other users—the uploaded content needs to be reviewed to determine if it contains any prohibited content. If prohibited content is found, the user is not allowed to successfully upload the user-generated content.

[0060] Server 120 is equipped with a method for detecting violations in user-uploaded content, enabling the detection of violations in audio, images, or videos.

[0061] Among them, terminal 110 can be any electronic device capable of running creation and playback applications, such as smartphones, tablets, laptops, and computers. Server 120 can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. This document does not impose any restrictions on these aspects.

[0062] Figure 2 This is a flowchart illustrating an illegal audio detection method according to an exemplary embodiment. The method can be applied to... Figure 1 The implementation environment shown, and by Figure 1 The server 120 in the illustrated embodiment environment is specifically executed.

[0063] like Figure 2 As shown, in an exemplary embodiment, the illegal audio detection method may include steps S210 to S230, which are described in detail below:

[0064] Step S210: Obtain the audio to be detected, extract the human voice audio from the audio to be detected, and convert the human voice audio into text information.

[0065] In this embodiment, when a user uploads audio, the uploaded audio is used as the audio to be detected, and a detection request is generated. Upon receiving the detection request, the audio to be detected corresponding to the detection request is obtained. The audio to be detected includes background audio, melody audio, and human voice audio, etc. By suppressing and separating the background audio and melody audio, the human voice audio in the audio to be detected is extracted, and the human voice audio is converted into text information. This application, by suppressing and separating background audio and melody audio, will greatly reduce the interference of music on voiceprint features and speech features, and significantly improve the accuracy of subsequent conversion of human voice audio into text information and extraction of voiceprint features.

[0066] Specifically, when extracting human voice audio, the audio to be detected is input into a preset human voice separation model for processing. The human voice separation model can be a deep neural network model such as U-Net, Wave-U-Net, WaveGlow, Deep Clustering, TasNet (Time-domain audio separation Network), Conv-TasNet (Convolutional Time-domain audio separation Network), or trained using attention-based TasNet or Conv-TasNet. The human voice separation model separates the audio to be detected into two audio files: the human voice audio and all other audio files. In this embodiment, only the human voice audio is retained for subsequent processing.

[0067] The extracted human voice audio is processed for speech recognition to obtain text information. Specifically, the human voice audio is input into a pre-trained speech recognition model for speech recognition to obtain text information.

[0068] In one exemplary embodiment of this application, please refer to Figure 3 Before inputting the text information into the violation tone detection module in step S220 to detect whether there is a violation tone in the audio to be detected, the violation audio detection method also includes steps S310 to S330, which are described in detail below:

[0069] Step S310: Extract the voiceprint features from the human voice audio to obtain the target voiceprint features.

[0070] In this embodiment, voiceprint features are extracted from human voice audio as target voiceprint features. Specifically, this extraction can be performed using a trained voiceprint model, which can be trained based on a deep learning TDNN (Time Delay Neural Network) network structure.

[0071] Step S320: Match the target voiceprint features with the illegal voiceprint features in the voiceprint database to obtain the feature matching result.

[0072] In this embodiment of the application, the voiceprint features of the segments of illegal audio discovered or reported during operation are extracted in advance, the x-vectors of the extracted voiceprints are combined into a voiceprint library, the publishers corresponding to the illegal audio are marked as sensitive personnel, and the voiceprint features of the extracted illegal audio are used as illegal voiceprint features.

[0073] The target voiceprint features are matched with the illegal voiceprint features to obtain the feature matching results. The feature matching results are divided into two categories: one is that the target voiceprint features and the illegal voiceprint features do not match, and the other is that the target voiceprint features and the illegal voiceprint features match.

[0074] Step S330: If the feature matching result indicates that the target voiceprint feature does not match the violation voiceprint feature, execute the step of inputting the text information into the preset violation tone detection module to detect whether there is a violation tone in the audio to be detected.

[0075] In this embodiment, if the feature matching result indicates that the target voiceprint feature does not match the violation voiceprint feature, then the step of inputting text information into a preset violation tone detection module is executed to detect whether there is a violation tone in the audio to be detected. If the target voiceprint feature matches the violation voiceprint feature, then there is no need to perform the subsequent step of inputting text information into the violation tone detection module for processing, and the publisher is directly prohibited from publishing the audio to be detected on the application.

[0076] In another embodiment of this application, when the target voiceprint features match the illegal voiceprint features, the text information can be input into a preset semantic model for semantic recognition processing to obtain the semantic recognition result, and then the detection audio can be further determined based on the semantic recognition result to determine whether the audio to be detected is illegal audio.

[0077] In this embodiment, by matching voiceprint features, not all audio to be detected needs to be detected by the violation tone detection module and violation detection model, which greatly saves server computing resources and improves the detection efficiency of violation detection.

[0078] In one exemplary embodiment of this application, please refer to Figure 4 In step S310, the voiceprint features in the human voice audio are extracted to obtain the target voiceprint features, including steps S410 to S430, which are described in detail below:

[0079] Step S410: The human voice audio is segmented to obtain multiple human voice audio segments.

[0080] In this embodiment, the human voice audio is segmented to obtain individual audio segments. During segmentation, a pre-set number of segments can be configured, and the human voice audio is cut into multiple segments according to a fixed number of segments. In another embodiment, a pre-set duration for each audio segment can be configured, and the human voice audio is cut into segments of fixed duration. The aforementioned pre-set number of segments and duration can be set as needed, and this application does not impose any restrictions on this.

[0081] Step S420: Extract the voiceprint features of each human voice audio segment using the voiceprint model.

[0082] In this embodiment of the application, each voice audio segment is input into the voiceprint model for voiceprint recognition, and the voiceprint feature x-vector of each voice audio segment is extracted.

[0083] Step S430: Calculate the average voiceprint feature of multiple voiceprint features, and use the average voiceprint feature as the target voiceprint feature.

[0084] In this embodiment, multiple voiceprint features are averaged using x-vectors to obtain an average voiceprint feature. This average voiceprint feature is then used as the target voiceprint feature and matched against illegal voiceprint features in the voiceprint database.

[0085] Step S220: Input the text information into the violation tone detection module to detect whether there is a violation tone in the audio to be detected; wherein, the violation tone detection module performs detection based on at least one of the tone words in the text information and the duration of the audio to be detected.

[0086] In this embodiment, a pre-set irregular tone detection module is included. This module detects irregular tone based on at least one of the interjections in the text information and the duration of the audio to be detected. The converted text information is input into the irregular tone detection module to detect whether there is an irregular tone in the audio to be detected. In this embodiment, irregular tone can be understood as expressing illegal content through interjections. Therefore, if an irregular tone exists in the audio to be detected, the interjections and their duration will differ from normal audio.

[0087] In one exemplary embodiment of this application, please refer to Figure 5 In step S220, the text information is input into the violation tone detection module to detect whether there is a violation tone in the audio to be detected, including steps S510 to S530, which are described in detail below:

[0088] Step S510: Obtain the duration information of the audio to be detected, and extract the number of interjections and words in the text information, as well as the number of characters in the text information.

[0089] In this embodiment, the duration information of the audio to be detected is obtained, and the text information is segmented. During segmentation, tools such as Jieba, SnowNLP, THULAC (THU Lexical Analyzer for Chinese), and NLPIR can be used. During segmentation, each segmented word is tagged with its part of speech, and modal particles are extracted from these tagged words. The number of modal particles, the total number of words, and the number of characters in the text information are then counted.

[0090] Step S520: Calculate the first ratio of the number of modal particles to the number of words, and calculate the second ratio of the number of text characters to the duration information.

[0091] In this embodiment, the number of modal particles is divided by the number of words to obtain a first ratio, and the number of text characters is divided by the duration information to obtain a second ratio.

[0092] Step S530: Detect whether there is an illegal tone in the audio to be detected based on the first ratio and the second ratio.

[0093] In this embodiment, typically, when there is an inappropriate tone in the audio to be detected, the proportion of non-interjections in the human voice audio is relatively small, while the proportion of interjections such as "ah," "oh," "ha," and "okay" increases significantly. Furthermore, the number of recognized characters is also much smaller than in normal audio of the same duration. Therefore, the presence of an inappropriate tone in the audio to be detected can be determined based on the first ratio and the second ratio.

[0094] In one exemplary embodiment of this application, please refer to Figure 6 In step S530, the presence of an inappropriate tone in the audio to be detected is determined based on the first ratio and the second ratio. This includes steps S610 to S630, which are detailed below:

[0095] Step S610: Match the first ratio with the first detection threshold to obtain the first matching result, and match the second ratio with the second detection threshold to obtain the second matching result.

[0096] In this embodiment, a first detection threshold and a second detection threshold are preset. The first ratio is matched with the first detection threshold to obtain a first matching result, and the second ratio is matched with the second detection threshold to obtain a second matching result.

[0097] Step S620: If the first matching result indicates that the first ratio is greater than the first detection threshold, and the second matching result indicates that the second ratio is less than the second detection threshold, it is determined that there is an illegal tone in the audio to be detected.

[0098] In this embodiment of the application, if the first ratio is greater than the first detection threshold and the second ratio is less than the second detection threshold, it indicates that the proportion of non-interjections in the human voice audio is small, while the proportion of interjections is large. At the same time, the number of characters identified is much smaller than that of normal audio for the same audio length, and it can be determined that there is an illegal tone in the audio to be detected.

[0099] Step S630: If the first matching result indicates that the first ratio is less than or equal to the first detection threshold, or the second matching result indicates that the second ratio is greater than or equal to the second detection threshold, it is determined that there is no illegal tone in the audio to be detected.

[0100] In this embodiment of the application, if either the first ratio is less than or equal to the first detection threshold, or the second ratio is greater than or equal to the second detection threshold, it indicates that the proportion of non-interjections in the human voice audio is relatively large, while the proportion of interjections is relatively small, or the number of recognized characters is also normal audio for the same audio length, and it can be determined that there is no irregular tone in the audio to be detected.

[0101] Step S230: If there is an illegal tone in the audio to be detected, input the text information into the preset illegal detection model to detect whether the audio to be detected is illegal audio.

[0102] In this embodiment of the application, the preset violation detection model can be trained based on the Panns-cnn10 model or the Ecapa-tdnn model. During training, two types of samples are set: positive samples and negative samples. The positive samples mainly use natural sounds, which may include animal calls, natural sounds and water sounds, non-conversational human voices, household sounds, urban noise, etc.

[0103] Negative samples are made up of labeled, non-compliant audio.

[0104] After training the Panns-CNN10 model or Ecapa-TDNN to obtain a baseline model using positive and negative samples, transfer training was performed. During transfer training, data such as songs, recorded courses, and audiobooks were incorporated to make the final violation detection model more accurate and reduce the overall training time. In one embodiment of this application, the Ecapa-TDNN model is preferred for training the violation detection model.

[0105] After identifying the presence of inappropriate tone in the audio to be detected, the text information is input into a preset violation detection model to further detect whether the audio to be detected is inappropriate, thus enabling more accurate violation detection.

[0106] In this embodiment of the application, if the audio to be detected is detected as illegal audio according to the violation detection model, the audio to be detected is input to the manual review channel for further review. The manual review result indicates that the audio to be detected is illegal audio. Then, the target voiceprint features extracted from the audio to be detected are stored in the voiceprint library as illegal voiceprint features, and the publisher of the audio to be detected is marked as a sensitive person.

[0107] In this embodiment, when it is necessary to detect the audio to be detected, the human voice audio is extracted from the audio to be detected and converted into text information. The extracted human voice audio will greatly reduce the interference of background music, thereby improving the recognition accuracy of converting human voice audio into text information. This application also inputs the converted text information into a violation tone detection module to detect whether there is a violation tone in the audio to be detected. The violation tone detection module performs detection based on at least one of the tone words in the text information and the duration of the audio to be detected. When there is a violation tone in the audio to be detected, the tone words and duration are different from normal audio. Therefore, the violation tone detection module can accurately detect whether there is a violation tone in the audio to be detected. When a violation tone is detected in the audio to be detected, the text information is further input into a preset violation detection model to detect whether the audio to be detected is a violation audio, which can perform violation detection more quickly and accurately. It is evident that the technical solution adopted in this application eliminates the need for dedicated manual review, thereby reducing manpower and time costs. It also helps the operations review team to prohibit and delete audio content that harms the health and security of the network, thus building a new pan-entertainment ecosystem that promotes healthy values.

[0108] In one exemplary embodiment of this application, please refer to Figure 7 The method for detecting illegal audio also includes steps S710 and S720, which are detailed below:

[0109] Step S710: If there is no inappropriate tone in the audio to be detected, input the text information into the preset semantic model for semantic recognition processing to obtain the semantic recognition result.

[0110] In this embodiment of the application, after the violation tone detection module detects that there is no violation tone in the audio to be detected, the text information is input into the preset semantic model for semantic recognition processing to obtain the semantic recognition result.

[0111] Step S720: Determine whether the audio to be detected is illegal audio based on the semantic recognition result.

[0112] In this embodiment of the application, the semantics expressed by the audio to be detected can be determined based on the semantic recognition results, and thus it can be determined whether the audio to be detected is illegal audio.

[0113] In one exemplary embodiment of this application, please refer to Figure 8 After determining whether the audio to be detected is illegal audio based on the semantic recognition result in step S720, the illegal audio detection method also includes steps S810 and S820, which are described in detail below:

[0114] Step S810: If the audio to be detected is illegal audio, determine the violation category based on the semantic recognition result.

[0115] In this embodiment of the application, when it is determined that the audio to be detected is illegal audio, the violation type is further determined based on the semantic recognition result. There are multiple violation types.

[0116] Step S820: Determine the corresponding processing procedure based on the violation category, and process the audio to be detected according to the processing procedure.

[0117] In this embodiment, each violation category has a corresponding processing flow, such as prohibited and suspected. In the prohibited processing flow, the audio to be detected is prohibited from being published on the network. In the suspected processing flow, it is necessary to further determine whether it is a violation audio. The audio to be detected is processed according to the determined processing flow.

[0118] In one exemplary embodiment of this application, please refer to Figure 9 , Figure 9 This is an exemplary embodiment of a method for detecting illegal audio, including steps S910 to S960, which are described in detail below:

[0119] Step S910: If a detection request is received, obtain the audio to be detected according to the detection request.

[0120] In this embodiment of the application, when a user uploads audio, the uploaded audio is used as the audio to be detected, and a detection request is generated. When the detection request is received, the audio to be detected corresponding to the detection request is obtained.

[0121] Step S920: Extract the human voice audio from the audio to be detected.

[0122] In this embodiment, the audio to be detected is processed according to a preset human voice separation model, and the human voice audio in the audio to be detected is extracted by suppressing and separating the background audio, melody audio, etc.

[0123] Step S930: Extract the voiceprint features from the human voice audio to obtain the target voiceprint features, and match the target voiceprint features with the illegal voiceprint features in the voiceprint database to obtain the feature matching result.

[0124] In this embodiment, step S930 is consistent with the description in steps S310 and S320 above, and will not be repeated here.

[0125] Step S940: If the feature matching result indicates that the target voiceprint feature does not match the illegal voiceprint feature, the text information is input into the preset semantic model for semantic recognition processing to obtain the semantic recognition result, and the human voice audio is converted into text information.

[0126] In this embodiment of the application, if the feature matching result indicates that the target voiceprint feature does not match the illegal voiceprint feature, the text information is directly input into the preset semantic model for semantic recognition processing to obtain the semantic recognition result, and the human voice audio is converted into text information.

[0127] Step S950: If the semantic recognition result indicates that the audio to be detected is not illegal audio, the text information is input into the preset illegal tone detection module to detect whether there is illegal tone in the audio to be detected; wherein, the illegal tone detection module performs detection based on at least one of the tone words in the text information and the duration of the audio to be detected.

[0128] In this embodiment of the application, when the semantic recognition result indicates that the audio to be detected is not illegal audio, the converted text information is input into the preset illegal tone detection module to detect whether there is illegal tone in the audio to be detected. The description of how the tone detection module detects whether there is illegal tone in the audio to be detected is consistent with the description of the aforementioned steps S510 to S530 and steps S610 to S630, and will not be repeated here.

[0129] Step S960: If there is an illegal tone in the audio to be detected, input the text information into the preset illegal detection model to detect whether the audio to be detected is illegal audio.

[0130] In this embodiment of the application, if there is an illegal tone in the audio to be detected, the text information is input into the preset illegal detection model to detect whether the audio to be detected is illegal audio. The relevant description of step S960 is the same as the relevant description of step S230 above, and will not be repeated here.

[0131] In one exemplary embodiment of this application, please refer to Figure 10 , Figure 10 An illegal audio detection device is shown according to an exemplary embodiment, comprising:

[0132] The acquisition module 1010 is configured to acquire the audio to be detected, extract the human voice audio from the audio to be detected, and convert the human voice audio into text information;

[0133] The first detection module 1020 is configured to input text information into the violation tone detection module to detect whether there is a violation tone in the audio to be detected; wherein, the violation tone detection module performs detection based on at least one of the tone words in the text information and the duration of the audio to be detected;

[0134] The second detection module 1030 is configured to input text information into a preset violation detection model to detect whether the audio to be detected is a violation if there is an illegal tone in the audio to be detected.

[0135] In one exemplary embodiment of this application, the first detection module 1020 includes:

[0136] The first extraction submodule is configured to obtain the duration information of the audio to be detected, and extract the number of interjections and words in the text information, as well as the number of characters in the text information;

[0137] The first calculation submodule is configured to calculate a first ratio of the number of modal particles to the number of words, and a second ratio of the number of text characters to the duration information;

[0138] The detection submodule is configured to detect whether there is an inappropriate tone in the audio to be detected based on a first ratio and a second ratio.

[0139] In one exemplary embodiment of this application, the detection submodule includes:

[0140] The matching unit is configured to match a first ratio with a first detection threshold to obtain a first matching result, and to match a second ratio with a second detection threshold to obtain a second matching result;

[0141] The first determining unit is configured to determine that there is an illegal tone in the audio to be detected if the first matching result indicates that the first ratio is greater than the first detection threshold and the second matching result indicates that the second ratio is less than the second detection threshold.

[0142] The second determining unit is configured to determine that there is no illegal tone in the audio to be detected if the first matching result indicates that the first ratio is less than or equal to the first detection threshold, or the second matching result indicates that the second ratio is greater than or equal to the second detection threshold.

[0143] In one exemplary embodiment of this application, the illegal audio detection device further includes:

[0144] The extraction module is configured to extract voiceprint features from human voice audio to obtain target voiceprint features;

[0145] The matching module is configured to match the target voiceprint features with the illegal voiceprint features in the voiceprint database to obtain the feature matching results;

[0146] The execution module is configured to, if the feature matching result indicates that the target voiceprint feature does not match the violation voiceprint feature, execute the step of inputting text information into the preset violation tone detection module to detect whether there is a violation tone in the audio to be detected.

[0147] In one exemplary embodiment of this application, the extraction module includes:

[0148] The segmentation processing submodule is configured to segment human voice audio to obtain multiple human voice audio segments;

[0149] The second extraction submodule is configured to extract the voiceprint features of each human voice audio segment through a voiceprint model;

[0150] The second calculation submodule is configured to calculate the average voiceprint feature of multiple voiceprint features and use the average voiceprint feature as the target voiceprint feature.

[0151] In one exemplary embodiment of this application, the illegal audio detection device further includes:

[0152] The semantic recognition module is configured to input text information into a preset semantic model for semantic recognition processing if there is no inappropriate tone in the audio to be detected, and obtain the semantic recognition result.

[0153] The first determination module is configured to determine whether the audio to be detected is illegal audio based on the semantic recognition results.

[0154] In one exemplary embodiment of this application, the illegal audio detection device further includes:

[0155] The second determination module is configured to determine the violation category based on the semantic recognition result if the audio to be detected is a violation audio.

[0156] The third module is configured to determine the corresponding processing flow based on the violation category, and process the audio to be detected through the processing flow.

[0157] It should be noted that the illegal audio detection device provided in the above embodiments and the illegal audio detection method provided in the above embodiments belong to the same concept. The specific way in which each module, sub-module and unit performs operations has been described in detail in the method embodiments, and will not be repeated here.

[0158] Embodiments of this application also provide an electronic device, including: one or more processors; and a storage device for storing one or more programs, which, when executed by the one or more processors, cause the electronic device to implement the illegal audio detection method provided in the above embodiments.

[0159] Methods for detecting illegal audio include:

[0160] Acquire the audio to be detected, extract the human voice audio from the audio to be detected, and convert the human voice audio into text information;

[0161] The text information is input into the violation tone detection module to detect whether there is a violation tone in the audio to be detected; the violation tone detection module detects based on at least one of the tone words in the text information and the duration of the audio to be detected;

[0162] If the audio to be detected contains inappropriate tone, the text information is input into the preset violation detection model to detect whether the audio to be detected is inappropriate.

[0163] Figure 11 A schematic diagram of the structure of a computer system suitable for implementing the electronic device of the present application is shown.

[0164] It should be noted that, Figure 11 The computer system 1100 of the electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0165] like Figure 11 As shown, the computer system 1100 includes a Central Processing Unit (CPU) 1101, which can perform various appropriate actions and processes based on programs stored in Read-Only Memory (ROM) 1102 or programs loaded from storage portion 1108 into Random Access Memory (RAM) 1103, such as performing the methods described in the above embodiments. Various programs and data required for system operation are also stored in RAM 1103. The CPU 1101, ROM 1102, and RAM 1103 are interconnected via bus 1104. An Input / Output (I / O) interface 1105 is also connected to bus 1104.

[0166] The following components are connected to I / O interface 1105: an input section 1106 including a keyboard, mouse, etc.; an output section 1107 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 1108 including a hard disk, etc.; and a communication section 1109 including a network interface card such as a LAN (Local Area Network) card, modem, etc. The communication section 1109 performs communication processing via a network such as the Internet. A drive 1110 is also connected to I / O interface 1105 as needed. Removable media 1111, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 1110 as needed so that computer programs read from them can be installed into storage section 1108 as needed.

[0167] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 1109, and / or installed from removable medium 1111. When the computer program is executed by central processing unit (CPU) 1101, it performs various functions defined in the system of this application.

[0168] It should be noted that the computer-readable medium shown in the embodiments of this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: 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), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a 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. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying a computer-readable computer program. The transmitted data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The computer program contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wired, etc., or any suitable combination thereof.

[0169] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. Each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0170] The units described in the embodiments of this application can be implemented in software or hardware, and the described units can also be located in a processor. The names of these units do not necessarily limit the specific unit itself.

[0171] Another aspect of this application provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the method described above. This computer-readable storage medium may be included in the electronic device described in the above embodiments, or it may exist independently and not assembled into the electronic device.

[0172] Another aspect of this application provides a computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the illegal audio detection method provided in the various embodiments described above.

[0173] Methods for detecting illegal audio include:

[0174] Acquire the audio to be detected, extract the human voice audio from the audio to be detected, and convert the human voice audio into text information;

[0175] The text information is input into the violation tone detection module to detect whether there is a violation tone in the audio to be detected; the violation tone detection module detects based on at least one of the tone words in the text information and the duration of the audio to be detected;

[0176] If the audio to be detected contains inappropriate tone, the text information is input into the preset violation detection model to detect whether the audio to be detected is inappropriate.

[0177] The above description is merely a preferred exemplary embodiment of this application and is not intended to limit the implementation of this application. Those skilled in the art can easily make corresponding modifications or alterations based on the main concept and spirit of this application. Therefore, the scope of protection of this application should be determined by the scope of protection claimed in the claims.

Claims

1. A method for detecting illegal audio, characterized in that, include: Acquire the audio to be detected, extract the human voice audio from the audio to be detected, and convert the human voice audio into text information; The text information is input into the violation tone detection module to detect whether there is a violation tone in the audio to be detected; If the audio to be detected contains an inappropriate tone, the text information is input into a preset violation detection model to detect whether the audio to be detected is an inappropriate audio. The preset violation detection model is trained based on the Panns-cnn10 model or the Ecapa-tdnn model. The step of inputting the text information into the violation tone detection module to detect whether there is a violation tone in the audio to be detected includes: obtaining the duration information of the audio to be detected, and extracting the number of tone words and the number of words in the text information, as well as the number of characters in the text information; calculating a first ratio of the number of tone words to the number of words, and calculating a second ratio of the number of characters in the text to the duration information; matching the first ratio with a first detection threshold to obtain a first matching result, and matching the second ratio with a second detection threshold to obtain a second matching result; if the first matching result indicates that the first ratio is greater than the first detection threshold, and the second matching result indicates that the second ratio is less than the second detection threshold, it is determined that there is a violation tone in the audio to be detected; if the first matching result indicates that the first ratio is less than or equal to the first detection threshold, or the second matching result indicates that the second ratio is greater than or equal to the second detection threshold, it is determined that there is no violation tone in the audio to be detected.

2. The method as described in claim 1, characterized in that, Before the text information is input into the violation tone detection module to detect whether there is a violation tone in the audio to be detected, the method further includes: The target voiceprint features are obtained by extracting the voiceprint features from the human voice audio. The target voiceprint features are matched with the illegal voiceprint features in the voiceprint database to obtain the feature matching results; If the feature matching result indicates that the target voiceprint feature does not match the violation voiceprint feature, the step of inputting the text information into the preset violation tone detection module to detect whether there is a violation tone in the audio to be detected is executed.

3. The method as described in claim 2, characterized in that, The step of extracting the voiceprint features from the human voice audio to obtain the target voiceprint features includes: The human voice audio is segmented to obtain multiple human voice audio segments; Voiceprint features of each of the human voice audio segments are extracted using a voiceprint model; Calculate the average voiceprint feature of the multiple voiceprint features, and use the average voiceprint feature as the target voiceprint feature.

4. The method according to any one of claims 1 to 3, characterized in that, The method further includes: If there is no inappropriate tone in the audio to be detected, the text information is input into a preset semantic model for semantic recognition processing to obtain the semantic recognition result; Based on the semantic recognition results, it is determined whether the audio to be detected is illegal audio.

5. The method as described in claim 4, characterized in that, After determining whether the audio to be detected is illegal audio based on the semantic recognition result, the method further includes: If the audio to be detected is illegal audio, the violation category is determined based on the semantic recognition result; The corresponding processing procedure is determined based on the violation category, and the audio to be detected is processed through the processing procedure.

6. A device for detecting illegal audio, characterized in that, include: The acquisition module is configured to acquire the audio to be detected, extract the human voice audio from the audio to be detected, and convert the human voice audio into text information; The first detection module is configured to input the text information into the violation tone detection module to detect whether there is a violation tone in the audio to be detected; The second detection module is configured to input the text information into a preset violation detection model to detect whether the audio to be detected is a violation audio if there is an illegal tone in the audio to be detected. The preset violation detection model is trained based on the Panns-cnn10 model or the Ecapa-tdnn model. The first detection module is further configured to: acquire the duration information of the audio to be detected; extract the number of interjections and words in the text information, as well as the number of characters in the text information; calculate a first ratio of the number of interjections to the number of words, and calculate a second ratio of the number of characters to the duration information; match the first ratio with a first detection threshold to obtain a first matching result, and match the second ratio with a second detection threshold to obtain a second matching result; if the first matching result indicates that the first ratio is greater than the first detection threshold, and the second matching result indicates that the second ratio is less than the second detection threshold, it is determined that there is an illegal tone in the audio to be detected; if the first matching result indicates that the first ratio is less than or equal to the first detection threshold, or the second matching result indicates that the second ratio is greater than or equal to the second detection threshold, it is determined that there is no illegal tone in the audio to be detected.

7. An electronic device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the electronic device to implement the illegal audio detection method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, It stores computer-readable instructions that, when executed by the computer's processor, cause the computer to perform the illegal audio detection method according to any one of claims 1 to 5.

Images