Audio processing method, device, electronic device and storage medium

Abstract

The present application provides an audio processing method, device, electronic device and storage medium, the method comprising: obtaining audio data to be processed at a current moment, and extracting current voiceprint features from the audio data to be processed online; obtaining at least one historical cluster at a moment before the current moment, each historical cluster being associated with a user tag of the historical user; performing offline clustering on the current voiceprint features based on at least one historical cluster to obtain a current cluster, the current cluster being associated with a user tag of a user to which the corresponding voiceprint feature belongs, and at least one cluster in the current cluster corresponds to the current voiceprint feature; determining a user tag mapping relationship online based on user tags associated with historical clusters and user tags associated with current clusters; and updating the user tag of the user to which the current voiceprint feature belongs online in the current cluster based on the user tag mapping relationship, and outputting at least the updated user tag of the current voiceprint feature.

Description

Technical Field

[0001] The present application relates to the field of audio processing technology, and in particular to an audio processing method, device, electronic device and storage medium. Background Art

[0002] In some application scenarios, it is necessary to cluster the audio data of the same user into one category from continuous multi-user audio data and output the corresponding user label.

[0003] However, how to accurately and efficiently cluster audio data and output corresponding user tags becomes a problem. Summary of the invention

[0004] This application provides the following technical solutions:

[0005] On one hand, the present application provides an audio processing method, comprising:

[0006] Acquire the audio data to be processed at the current moment, and extract the current voiceprint feature online from the audio data to be processed;

[0007] Obtaining at least one historical cluster at a time before the current time, wherein the at least one historical cluster is obtained by clustering at least one historical voiceprint feature corresponding to at least one historical user, each of the historical clusters corresponds to at least one historical voiceprint feature of a historical user, and is associated with a user tag of the historical user;

[0008] Offline clustering of the current voiceprint feature based on the at least one historical cluster is performed to obtain a current cluster, wherein the current cluster is associated with a user tag of a user to which the corresponding voiceprint feature belongs, and at least one cluster in the current cluster corresponds to the current voiceprint feature;

[0009] Determining a user tag mapping relationship online based on the user tags associated with the historical clusters and the user tags associated with the current clusters;

[0010] Based on the user label mapping relationship, the user label of the user to which the current voiceprint feature belongs is updated online in the current cluster, and at least the updated user label of the current voiceprint feature is output.

[0011] The online determination of the user tag mapping relationship based on the user tags associated with the historical clusters and the user tags associated with the current clusters includes:

[0012] Obtaining a user tag associated with the current cluster and at least one user tag associated with the at least one historical cluster;

[0013] Establishing a tag pair mapping relationship for each user tag associated with the current cluster, each tag pair mapping relationship includes a matching factor between the user tag associated with the current cluster and the user tag associated with each historical cluster, wherein the matching factor represents a matching degree between the user tag associated with the current cluster and the user tag associated with each historical cluster;

[0014] Determine a matching factor in each of the tag pair mapping relationships whose matching degree satisfies a condition;

[0015] Based on the matching factors of the respective matching degrees satisfying the conditions, a mapping relationship between the user tags associated with the current cluster and the user tags associated with the historical cluster is determined.

[0016] Each tag pair mapping relationship is obtained by comparing a user tag associated with the current cluster with each user tag associated with the historical cluster;

[0017] The determining of the matching factors of the tag pair mapping relationships in which the matching degree satisfies the condition includes:

[0018] Based on each of the tag pair mapping relationships, N groups of matching factors are obtained, each group of matching factors includes one matching factor in each of the tag pair mapping relationships, wherein N is less than the number of user tags associated with the historical clustering;

[0019] A group of matching factors having the largest sum of matching factors among the multiple groups of matching factors is determined as the matching factors whose matching degrees meet the conditions in the mapping relationships of the respective tag pairs.

[0020] The method further comprises:

[0021] Based on the user tag updated by the current voiceprint feature, updating the user tag associated with the current cluster;

[0022] Based on the updated user tags associated with the current cluster, at least a portion of the user tags output at a time before the current time is updated.

[0023] The at least one historical cluster is determined by:

[0024] Obtain multiple historical voiceprint features before the current moment;

[0025] Based on each of the historical voiceprint features as an original historical cluster, determine the similarity between every two original historical clusters, merge the two original historical clusters with the highest similarity into one historical cluster, and obtain at least one historical cluster;

[0026] In the case that the number of the at least one historical cluster does not reach the set number, returning to the step of determining the similarity between every two of the original historical clusters until the number of the at least one historical cluster reaches the set number, and saving the at least one historical cluster;

[0027] The set number is greater than or equal to the number of users corresponding to the multiple historical voiceprint features before the current moment.

[0028] The performing offline clustering on the current voiceprint feature based on the at least one historical cluster to obtain a current cluster includes:

[0029] Initially determining a current cluster using the current voiceprint feature as a first target cluster, wherein the initially determined current cluster includes the first target cluster and the at least one historical cluster;

[0030] Determining the similarity between every two clusters in the initially determined current clusters;

[0031] Determine whether the highest similarity among the similarities between each two clusters is lower than a similarity threshold;

[0032] If it is lower than the similarity threshold, the clustering is terminated, and the current clustering is finally obtained;

[0033] If it is not lower than the similarity threshold, merging the two clusters corresponding to the highest similarity to obtain a second target cluster;

[0034] The initially determined current clusters are updated based on the second target clusters, and the step of determining the similarity between every two clusters in the initially determined current clusters is returned to be executed.

[0035] Another aspect of the present application provides an audio processing device, comprising:

[0036] An extraction module, used to obtain the audio data to be processed at the current moment, and extract the current voiceprint features online from the audio data to be processed;

[0037] an acquisition module, configured to acquire at least one historical cluster at a time before the current time, wherein the at least one historical cluster is obtained by clustering at least one historical voiceprint feature corresponding to at least one historical user, each of the historical clusters corresponds to at least one historical voiceprint feature of a historical user, and is associated with a user tag of the historical user;

[0038] A clustering module, configured to perform offline clustering on the current voiceprint feature based on the at least one historical cluster to obtain a current cluster, wherein the current cluster is associated with a user tag of a user to which the corresponding voiceprint feature belongs, and at least one cluster in the current cluster corresponds to the current voiceprint feature;

[0039] A determination module, configured to determine a user tag mapping relationship online based on the user tags associated with the historical clusters and the user tags associated with the current clusters;

[0040] An updating module, configured to update the user label of the user to which the current voiceprint feature belongs online in the current cluster based on the user label mapping relationship;

[0041] At least output the user label after the current voiceprint feature is updated;

[0042] The output module is used to output at least the updated user label of the current voiceprint feature.

[0043] The determination module is specifically used for:

[0044] Obtaining a user tag associated with the current cluster and at least one user tag associated with the at least one historical cluster;

[0045] Establishing a tag pair mapping relationship for each user tag associated with the current cluster, each tag pair mapping relationship includes a matching factor between the user tag associated with the current cluster and the user tag associated with each historical cluster, wherein the matching factor represents a matching degree between the user tag associated with the current cluster and the user tag associated with each historical cluster;

[0046] Determine a matching factor in each of the tag pair mapping relationships whose matching degree satisfies a condition;

[0047] Based on the matching factors whose respective matching degrees satisfy the conditions, a mapping relationship between the user tags associated with the current cluster and the user tags associated with the historical cluster is determined.

[0048] A third aspect of the present application provides an electronic device, including:

[0049] Memory and processor;

[0050] The memory is used to store at least one set of instruction sets;

[0051] The processor is used to call and execute the instruction set in the memory, and execute any one of the audio processing methods described above by executing the instruction set.

[0052] A fourth aspect of the present application provides a storage medium storing a computer program for implementing any one of the audio processing methods described above, wherein the computer program is executed by a processor to implement any one of the audio processing methods described above.

[0053] In this application, the audio data to be processed at the current moment is obtained, the current voiceprint features are extracted online from the audio data to be processed, at least one historical clustering at a moment before the current moment is obtained, and the current voiceprint features are clustered offline based on at least one historical clustering, so that the current voiceprint features are clustered offline based on the existing clustering, which can avoid the cumulative errors caused by the initial clustering errors and improve the accuracy of clustering. At the same time, the current voiceprint features are extracted online and clustered offline, which can reduce the clustering delay and ensure the efficiency of clustering.

[0054] In addition, based on the user tags associated with historical clusters and the user tags associated with the current cluster, the user tag mapping relationship is determined online, and based on the user tag mapping relationship, the user tag of the user to whom the current voiceprint feature belongs is updated online in the current cluster, so that the updated user tag of the current voiceprint feature can more accurately represent the cluster corresponding to the current voiceprint feature, and the updated user tag of the current voiceprint feature is output, so that the user can accurately determine the user corresponding to the current voiceprint feature, thereby improving the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0056] Figure 1 It is a schematic diagram of an implementation scenario of a clustering method in the prior art;

[0057] Figure 2 It is a flowchart of an audio processing method provided in the first embodiment of the present application;

[0058] Figure 3 This is a schematic diagram of an implementation scenario of an audio processing method provided in the first embodiment of the present application;

[0059] Figure 4 This is another implementation scenario schematic diagram of an audio processing method provided in the first implementation mode of the present application;

[0060] Figure 5 It is a flowchart of an audio processing method provided in the second embodiment of the present application;

[0061] Figure 6 is a flowchart of an audio processing method provided in the third embodiment of the present application;

[0062] Figure 7 This is a schematic diagram of an implementation scenario of an audio processing method provided in the third embodiment of the present application;

[0063] Figure 8 is a schematic diagram of another implementation scenario of an audio processing method provided in the third embodiment of the present application;

[0064] Fig. 9 It is a structural schematic diagram of an audio processing device provided by the present application. DETAILED DESCRIPTION

[0065] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0066] In the context of needing to cluster the audio data of the same user into one category from continuous multi-user audio data, currently, the method can be used Figure 1 Clustering is performed using the offline clustering method or the online clustering method shown.

[0067] Among them, the inventors found that:

[0068] The offline clustering method requires obtaining all audio data (such as Figure 1 Medium i …x i+1 and x j ) After that, clustering is performed based on all the audio data. However, this method cannot meet the real-time requirements.

[0069] However, the online clustering method can obtain one audio data online each time (such as Figure 1 Medium j ), then the audio data is compared with the previous audio data (such as Figure 1 The accuracy of this clustering method depends on the quality of the initial clustering. If the audio of the same speaker is initially divided into two categories, irreversible cumulative errors will be caused later.

[0070] In the process of analyzing the above existing clustering methods, the inventors found that the above offline clustering method and online clustering method cannot guarantee the accuracy and efficiency of clustering at the same time. Therefore, the inventors propose a new audio data method to solve the above problem.

[0071] Reference Figure 2 , is a flowchart of an audio processing method provided in the first embodiment of the present application. The data processing method provided in the present application can be applied to electronic devices. The present application does not limit the product type of the electronic device, such as Figure 2 As shown, the method may include but is not limited to the following steps:

[0072] Step S101: Acquire the audio data to be processed at the current moment, and extract the current voiceprint features online from the audio data to be processed.

[0073] The audio data to be processed at the current moment includes at least one piece of audio data of at least one user.

[0074] Specifically, the voiceprint recognition model can be used to extract voiceprint features online from the audio data to be processed, and the extracted voiceprint features are used as current voiceprint features.

[0075] Extracting the user's voiceprint features online from the audio data to be processed may include: deleting noise data online from the audio data to be processed to obtain target data, and extracting the user's voiceprint features online from the target data.

[0076] It can be understood that the online extraction of the current voiceprint features from the audio data to be processed can ensure the real-time extraction of the current voiceprint features.

[0077] Step S102: obtaining at least one historical cluster at a time before the current time, wherein the at least one historical cluster is obtained by clustering at least one historical voiceprint feature corresponding to at least one historical user, each historical cluster corresponds to at least one historical voiceprint feature of a historical user, and is associated with a user tag of the historical user.

[0078] Obtaining at least one historical clustering at a time before the current time may include but is not limited to:

[0079] S1021. Obtain multiple historical voiceprint features before the current moment.

[0080] This step may include:

[0081] Acquire historical audio data before the current moment, where the historical audio data includes at least one piece of audio data of at least one historical user;

[0082] One or more historical voiceprint features are extracted from each audio data of the historical user to obtain multiple historical voiceprint features before the current moment.

[0083] Specifically, when the audio data of a historical user is in a stable state, a historical voiceprint feature can be extracted from at least one audio data of the historical user to obtain a historical voiceprint feature corresponding to the historical user; or, when the audio data of a historical user is in an unstable state (such as intermittent coughing), multiple historical voiceprint features may be extracted from at least one audio data of the historical user to obtain multiple historical voiceprint features corresponding to the historical user.

[0084] For example, if the historical audio data before the current moment includes audio data m1, m2, m3, m4, m5, m6, m7, m8, m9, m10 and m11, m1, m2, m3, m4, m5, m6, m7, m8, m9, m10 and m11 are arranged in the order of the generation time of the audio data from early to late, m1, m2, m3, m4, m5 belong to user A, m6, m7, m8 belong to historical user B, m9, m10 and m11 belong to historical user A, historical voiceprint features m11, m22, m33, m44, m55 are extracted from m1, m2, m3, m4, m5 respectively, historical voiceprint features m66, m77, m88 are extracted from m6, m7, m8, and historical voiceprint features m99, m100 and m110 are extracted from m9, m10 and m11 respectively.

[0085] S1022: Cluster multiple historical voiceprint features to obtain at least one historical cluster, each historical cluster corresponds to at least one historical voiceprint feature of a historical user, and is associated with a user tag of the historical user.

[0086] For example, m11, m22, m33, m44, m55, m66, m77, m88, m99, m100 and m110 are clustered to obtain historical cluster a, historical cluster b and historical cluster c. Historical cluster a corresponds to historical voiceprint features m11, m22, m33, m44, m99, m100 and m110, historical cluster b corresponds to historical voiceprint feature m55, and historical cluster c corresponds to historical voiceprint features m66, m77 and m88.

[0087] This step may include but is not limited to:

[0088] S10221. Based on each historical voiceprint feature as an original historical cluster, determine the similarity between every two original historical clusters, merge the two original historical clusters with the highest similarity into one historical cluster, and obtain at least one historical cluster.

[0089] S10222. When the number of at least one historical cluster does not reach the set number, return to the step of determining the similarity between every two original historical clusters until the number of at least one historical cluster reaches the set number, and save at least one historical cluster.

[0090] The set number is greater than or equal to the number of users corresponding to multiple historical voiceprint features before the current moment.

[0091] When the number of at least one historical cluster reaches a set number, at least one historical cluster is saved to shorten the time to obtain at least one historical cluster and improve efficiency. The efficiency of offline clustering of current voiceprint features based on at least one historical cluster can be improved by using the saved at least one historical cluster.

[0092] It should be noted that different historical clusters are associated with different user labels, which are used to distinguish different users. The user labels associated with each historical voiceprint feature corresponding to the same historical cluster are the same. For example, the above historical cluster a can be associated with user label 1, historical cluster b can be associated with user label 2, and historical cluster c can be associated with user label 3. It should be noted that the output of the user labels associated with the historical voiceprint features corresponding to historical clusters a, b, and c is also output in the order of the generation time of the audio data from early to late, that is, the output is 11112333111.

[0093] It is understandable that although m55 and m11, m22, m33, and m44 are all extracted from the audio data of historical user A, the different states of historical user A may cause the voiceprint features in audio data m5 to be different from the voiceprint features in other audio data, which in turn causes the historical voiceprint features of historical user A to be clustered into two classes. For example, audio data m5 is the audio data of user A in a coughing state, and audio data m11, m22, m33, and m44 are the audio data of user A in a normal state, which causes m55 to be different from m11, m22, m33, and m44.

[0094] In this embodiment, the user tag is different from the user ID. At different times, the user tag associated with the same historical cluster may change, while the user ID does not change over time. For example, the user ID of historical user A is aa at any time; at the current moment, the user tag associated with historical cluster a corresponding to historical user A is 2, and at a moment before the current moment, the user tag associated with historical cluster a was 1.

[0095] Step S103: Offline clustering is performed on the current voiceprint feature based on at least one historical cluster to obtain a current cluster. The current cluster is associated with a user tag of a user to which the corresponding voiceprint feature belongs. At least one cluster in the current cluster corresponds to the current voiceprint feature.

[0096] In this embodiment, it is possible but not limited to: using a distributed clustering (kmeans) algorithm to perform offline clustering on the current voiceprint features based on at least one historical clustering to obtain the current clustering; or, it is possible but not limited to: using a hierarchical clustering algorithm to perform offline clustering on the current voiceprint features based on at least one historical clustering to obtain the current clustering.

[0097] Wherein, using a hierarchical clustering algorithm, the current voiceprint feature is clustered offline based on at least one historical cluster to obtain the current cluster, which may include but is not limited to:

[0098] S1031. Initially determine a current cluster using a current voiceprint feature as a first target cluster, where the initially determined current cluster includes the first target cluster and at least one historical cluster.

[0099] S1032: Determine the similarity between every two clusters in the initially determined current clusters.

[0100] Determining the similarity between each two clusters in the initially determined current clusters may include:

[0101] S10321, obtaining the voiceprint features corresponding to the first cluster and the voiceprint features corresponding to the second cluster in every two clusters in the current clusters initially determined;

[0102] S10322. Calculate the similarity between the voiceprint features corresponding to the first cluster and the voiceprint features corresponding to the second cluster.

[0103] If there is one voiceprint feature corresponding to the first cluster and one voiceprint feature corresponding to the second cluster, the similarity between the voiceprint feature corresponding to the first cluster and the voiceprint feature corresponding to the second cluster can be directly calculated.

[0104] If there are multiple voiceprint features corresponding to the first cluster or the second cluster, a first processing may be performed on the multiple voiceprint features corresponding to the first cluster or the multiple voiceprint features corresponding to the second cluster to obtain a first processing result, and the similarity between the multiple voiceprint features corresponding to the first cluster and the voiceprint features corresponding to the second cluster is calculated based on the first processing result; or, the similarity between the voiceprint features corresponding to the first cluster or the multiple voiceprint features corresponding to the second cluster is calculated based on the first processing result. Specifically, it is possible but not limited to performing an average operation processing on the voiceprint features corresponding to the first cluster or the second cluster to obtain an average voiceprint feature, and calculating the similarity between the average voiceprint feature corresponding to the first cluster and the voiceprint feature corresponding to the second cluster; or, calculating the similarity between the voiceprint features corresponding to the first cluster and the average voiceprint features corresponding to the second cluster.

[0105] S1033: Determine whether the highest similarity among the similarities between every two clusters is lower than a similarity threshold.

[0106] The similarity threshold can be set as needed and is not limited in this application.

[0107] S1034: If the similarity is lower than the threshold, the clustering is terminated, and the current clustering is finally obtained.

[0108] It can be understood that if the highest similarity among the similarities between every two clusters is lower than the similarity threshold, then the similarity between every two clusters is lower than the similarity threshold.

[0109] S1035: If it is not lower than the similarity threshold, the two clusters corresponding to the highest similarity are merged to obtain a second target cluster.

[0110] S1037. Update the initially determined current cluster based on the second target cluster, and return to execute step S1032.

[0111] Step S104: determining a user tag mapping relationship online based on the user tags associated with the historical clusters and the user tags associated with the current clusters.

[0112] In this embodiment, the current voiceprint feature is clustered offline based on at least one historical cluster. Part of the voiceprint features corresponding to at least one historical cluster may be clustered into another cluster. When the part of the voiceprint features is clustered into another cluster, it affects the change of the user tag associated with the historical cluster, which in turn affects the user tag of the user to whom the current voiceprint feature belongs in the current cluster. For example, if the current voiceprint feature is m120, at least one historical cluster is historical cluster a, historical cluster b and historical cluster c as described in step S102. Based on at least one historical cluster, m120 is clustered offline. The current clusters are cluster a and cluster b. Cluster a corresponds to voiceprint features m11, m22, m33, m44, m55, m99, m100, and m110. Cluster b corresponds to voiceprint features m66, m77, m88, and m120. The user label corresponding to cluster a in the current cluster is 1, and the user label corresponding to cluster b is 2. Among them, m55 changes from corresponding to historical cluster b to corresponding to cluster a, affecting the user label associated with at least one historical cluster from 11112333111 to 11111222111, and affecting the user label of the user to whom the current voiceprint feature belongs in the current cluster to be 2.

[0113] In order to determine the above influence, the user tag mapping relationship may be determined online based on the user tags associated with the historical clusters and the user tags associated with the current clusters.

[0114] Step S105: based on the user label mapping relationship, the user label of the user to which the current voiceprint feature belongs is updated online in the current cluster, and at least the updated user label of the current voiceprint feature is output.

[0115] Based on the user label mapping relationship, the user label of the user to which the current voiceprint feature belongs is updated online in the current cluster, so that the updated user label of the current voiceprint feature can more accurately represent the cluster corresponding to the current voiceprint feature.

[0116] In this embodiment, the audio data to be processed at the current moment is obtained, the current voiceprint features are extracted online from the audio data to be processed, at least one historical clustering at a moment before the current moment is obtained, and the current voiceprint features are clustered offline based on the at least one historical clustering, so that the current voiceprint features are clustered offline based on the existing clustering, which can avoid the cumulative errors caused by the initial clustering errors and improve the accuracy of clustering. At the same time, the current voiceprint features are extracted online and clustered offline, which can reduce the clustering delay and ensure the efficiency of clustering.

[0117] Furthermore, based on the user tags associated with the historical clusters and the user tags associated with the current clusters, the user tag mapping relationship is determined online, and based on the user tag mapping relationship, the user tag of the user to whom the current voiceprint feature belongs is updated online in the current cluster, so that the updated user tag of the current voiceprint feature corresponds to the user tag associated with the historical clusters, which can more accurately characterize the cluster corresponding to the current voiceprint feature, and output the updated user tag of the current voiceprint feature, so that the user can accurately determine the user corresponding to the current voiceprint feature, thereby improving the user experience. For example, based on the above example, Figure 3As shown, the current moment is represented as moment j, and the moment before the current moment is represented as moment j-1. Based on the user tags 11112333111 associated with the historical clusters and the user tags 111112221112 associated with the current clusters, the user tag mapping relationship is determined online: the user tag 1 associated with the current cluster corresponds to the user tag 1 associated with the historical clusters, the user tag 2 associated with the current cluster corresponds to the user tag 3 associated with the historical clusters, and the user tag 3 associated with the current cluster corresponds to the user tag 2 associated with the historical clusters. If the user tag mapping relationship is not based on the user tag mapping relationship, the user tag of the user to whom the current voiceprint feature belongs is updated online, and the current voiceprint feature is updated online. The user label of the user to which the voiceprint feature belongs is determined to be 2. If it corresponds to the user label 11112333111 associated with the historical cluster, it will be mistakenly judged that the current voiceprint features m120 and m55 correspond to the same cluster. However, based on the user label mapping relationship, the user label 2 of the user to which m120 belongs is updated online in the user label 111112221112 associated with the current cluster, and the updated user label of m120 is 3. If it corresponds to the user label 11112333111 associated with the historical cluster, it can be determined that m120 and m66, m77, and m88 correspond to the same cluster, which can more accurately characterize the cluster corresponding to the current voiceprint feature.

[0118] For example, Figure 4 As shown, at least the updated user label 3 of the current voiceprint feature m120 can be output.

[0119] As another optional embodiment of the present application, refer to Figure 5 , is a flowchart of an audio processing method provided in the second embodiment of the present application. This embodiment is mainly a refinement of the audio processing method described in the first embodiment above. Figure 5 As shown, the method may include but is not limited to the following steps:

[0120] Step S201: Obtain the audio data to be processed at the current moment, and extract the current voiceprint features online from the audio data to be processed.

[0121] Step S202: obtaining at least one historical cluster at a time before the current time, wherein the at least one historical cluster is obtained by clustering at least one historical voiceprint feature corresponding to at least one historical user, each historical cluster corresponds to at least one historical voiceprint feature of a historical user, and is associated with a user tag of the historical user.

[0122] Step S203: Offline clustering is performed on the current voiceprint feature based on at least one historical cluster to obtain a current cluster. The current cluster is associated with a user tag of a user to which the corresponding voiceprint feature belongs. At least one cluster in the current cluster corresponds to the current voiceprint feature.

[0123] The detailed process of steps S201 - S203 can refer to the related introduction of steps S101 - S103 in the first embodiment, which will not be repeated here.

[0124] Step S204: Obtain a user tag associated with the current cluster and at least one user tag associated with at least one historical cluster.

[0125] Step S205: Establish a tag pair mapping relationship for each user tag associated with the current cluster, each tag pair mapping relationship includes a matching factor between the user tag associated with the current cluster and the user tag associated with each historical cluster, and the matching factor represents the degree of matching between the user tag associated with the current cluster and the user tag associated with each historical cluster.

[0126] In this implementation, it is possible but not limited to compare the user tags associated with the current cluster with the user tags associated with each historical cluster in pairs to obtain a tag pair mapping relationship of the user tags associated with the current cluster.

[0127] When comparing the user tags associated with the current cluster with the user tags associated with each historical cluster, if the number of current clusters is less than the number of historical clusters, n virtual current clusters can be established based on the difference between the number of current clusters and the number of historical clusters, where n is equal to the difference between the number of current clusters and the number of historical clusters, and the virtual current cluster is different from the current cluster. The virtual current cluster can correspond to one or more virtual voiceprint features.

[0128] Comparing the user labels associated with the current cluster with the user labels associated with each historical cluster in pairs can be understood as: comparing the user labels associated with the voiceprint features corresponding to the current cluster with the user labels associated with the historical voiceprint features corresponding to each historical cluster in pairs.

[0129] Among them, corresponding to the method of comparing the user tags associated with the current cluster and the user tags associated with each historical cluster in pairs, the matching factor of the user tags associated with the current cluster and the user tags associated with each historical cluster can be the number of user tags associated with the historical voiceprint features corresponding to the historical clusters aligned with the user tags associated with the voiceprint features of the current cluster.

[0130] For example, referring to the current clusters in the example of the first embodiment, i.e., cluster a and cluster b, the user tags associated with cluster a and cluster b are 111112221112, and user tag 1 associated with cluster a and user tag 1 associated with historical cluster a, user tag 2 associated with historical cluster b and user tag 3 associated with historical cluster c are compared in pairs to obtain the tag pair mapping relationship shown in Table 1.

[0131] As shown in Table 1, 7, 1, and 0 belong to label pair mapping relationship 1 (i.e., the label pair mapping relationship of user label 1 associated with cluster a). In label pair mapping relationship 1, 7 represents the number of user labels 11111111 associated with the voiceprint features corresponding to cluster a, which are aligned with the historical mapping of user labels 11111111 associated with the voiceprint features corresponding to cluster a; 1 represents the number of user labels 2 associated with the voiceprint features corresponding to cluster a, which are aligned with the historical mapping of user labels 11111111 associated with the voiceprint features corresponding to cluster a; 0 represents the number of user labels 3 associated with the voiceprint features corresponding to cluster a, which are aligned with the historical mapping of user labels 11111111 associated with the voiceprint features corresponding to cluster a;

[0132] 0, 0, 3 belong to label pair mapping relationship 2 (i.e., the label pair mapping relationship of user label 2 associated with cluster b). The first 0 in label pair mapping relationship 2 represents the number of user labels 1 associated with the voiceprint features corresponding to cluster a that are aligned with the user label 2222 associated with the voiceprint features corresponding to cluster b. The second 0 represents the number of user labels 2 associated with the voiceprint features corresponding to cluster b that are aligned with the voiceprint features corresponding to cluster a that are aligned with the user label 2222 associated with the voiceprint features corresponding to cluster b. 3 represents the number of user labels 3 associated with the voiceprint features corresponding to cluster a that are aligned with the user label 2222 associated with the voiceprint features corresponding to cluster b.

[0133] 0, 0, 0 belongs to label pair mapping relationship 3 (i.e., the label pair mapping relationship of user label 3 associated with the virtual current cluster). The first 0 in label pair mapping relationship 3 represents the number of user label 1s associated with the voiceprint features corresponding to the historical cluster a mapped by the user label 3 associated with the voiceprint features corresponding to the virtual current cluster. The second 0 represents the number of user label 2s associated with the voiceprint features corresponding to the historical cluster a mapped by the user label 3 associated with the voiceprint features corresponding to the virtual current cluster. 3 represents the number of user label 3s associated with the voiceprint features corresponding to the virtual current cluster.

[0134] Label pair mapping relationship 1 Label pair mapping relationship 2 Label pair mapping relationship 3 7 0 0 1 0 0 0 3 0

[0135] Step S206: Determine the matching factor of each tag pair mapping relationship whose matching degree meets the condition.

[0136] This step may include:

[0137] S2061. Based on each tag pair mapping relationship, N groups of matching factors are obtained, each group of matching factors includes one matching factor in each tag pair mapping relationship, wherein N is less than the number of user tags associated with the historical clustering.

[0138] Optionally, each group of matching factors includes one of the matching factors in each label pair mapping relationship, and the user label associated with the current cluster and the user label associated with the historical cluster to which each matching factor in the group of matching factors belongs are different.

[0139] For example, as shown in Table 1, based on label pair mapping relationships 1, 2, and 3, two groups of matching factors can be obtained. The first group of matching factors may include: 7 in label pair mapping relationship 1, 3 in label pair mapping relationship 2, and the second 0 in label pair mapping relationship 3; the second group of matching factors may include: 1 in label pair mapping relationship 1, 3 in label pair mapping relationship 2, and the first 0 in label pair mapping relationship 3.

[0140] S2062: Determine a group of matching factors with the largest sum of matching factors among the multiple groups of matching factors as the matching factors whose matching degrees meet the conditions in the mapping relationships of each label pair.

[0141] For example, in step S2061, the sum of the matching factors in the first group of matching factors is 10, and the sum of the matching factors in the second group of matching factors is 4, then the first group of matching factors is determined to be the matching factors whose matching degrees in each tag pair mapping relationship meet the conditions.

[0142] Step S207: Determine the mapping relationship between the user tags associated with the current cluster and the user tags associated with the historical cluster based on the matching factors whose matching degrees meet the conditions.

[0143] Specifically, the user tags associated with the current cluster and the user tags associated with the historical cluster to which each matching factor with a matching degree that satisfies the condition belongs can be determined, and the user tags associated with the current cluster and the user tags associated with the historical cluster to which each matching factor with a matching degree that satisfies the condition belongs can be determined to be in a one-to-one relationship.

[0144] For example, based on the matching factors in the first group of matching factors, it is determined that the user label 1 associated with cluster a in the current cluster to which matching factor 7 in the first group of matching factors belongs and the user label 1 associated with the historical cluster are in a one-to-one mapping relationship, the user label 2 associated with cluster b in the current cluster to which matching factor 3 in the first group of matching factors belongs and the user label 3 associated with the historical cluster are in a one-to-one mapping relationship, and the user label 3 associated with the virtual current cluster in the current cluster to which matching factor 0 in the first group of matching factors belongs and the user label 2 associated with the historical cluster are in a one-to-one mapping relationship.

[0145] Steps S204-S207 are a specific implementation of step S104 in the first implementation.

[0146] Step S208: based on the user label mapping relationship, the user label of the user to which the current voiceprint feature belongs is updated online in the current cluster, and at least the updated user label of the current voiceprint feature is output.

[0147] The detailed process of step S208 may refer to a specific implementation of step S105 in the first implementation.

[0148] In this embodiment, the audio data to be processed at the current moment is obtained, the current voiceprint features are extracted online from the audio data to be processed, at least one historical clustering at a moment before the current moment is obtained, and the current voiceprint features are clustered offline based on the at least one historical clustering, so that the current voiceprint features are clustered offline based on the existing clustering, which can avoid the cumulative errors caused by the initial clustering errors and improve the accuracy of clustering. At the same time, the current voiceprint features are extracted online and clustered offline, which can reduce the clustering delay and ensure the efficiency of clustering.

[0149] Furthermore, by obtaining a user tag associated with the current cluster and at least one user tag associated with at least one historical cluster, a tag pair mapping relationship of each user tag associated with the current cluster is established respectively, each tag pair mapping relationship includes a matching factor between the user tag associated with the current cluster and the user tag associated with each historical cluster, the matching factor characterizing the degree of matching between the user tag associated with the current cluster and each user tag associated with the historical cluster, the matching factor whose matching degree satisfies a condition in each tag pair mapping relationship is determined, based on the matching factors whose matching degree satisfies the condition, the mapping relationship between the user tag associated with the current cluster and the user tag associated with the historical cluster is determined, based on the user tag mapping relationship, the user tag of the user to which the current voiceprint feature belongs is updated online in the current cluster, so that the updated user tag of the current voiceprint feature corresponds to the user tag associated with the historical cluster, which can more accurately characterize the cluster corresponding to the current voiceprint feature, and output the updated user tag of the current voiceprint feature, so that the user can accurately determine the user corresponding to the current voiceprint feature, thereby improving the user experience.

[0150] As another optional embodiment of the present application, refer to Figure 6 , is a flowchart of an audio processing method provided in the third embodiment of the present application. This embodiment is mainly an extension of the audio processing method described in the first embodiment above. Figure 6 As shown, the method may include but is not limited to the following steps:

[0151] Step S301, obtaining the audio data to be processed at the current moment, and extracting the current voiceprint features online from the audio data to be processed;

[0152] Step S302: obtaining at least one historical cluster at a time before the current time, wherein the at least one historical cluster is obtained by clustering at least one historical voiceprint feature corresponding to at least one historical user, each of the historical clusters corresponds to at least one historical voiceprint feature of a historical user, and is associated with a user tag of the historical user;

[0153] Step S303: Offline clustering is performed on the current voiceprint feature based on the at least one historical cluster to obtain a current cluster, wherein the current cluster is associated with a user tag of a user to which the corresponding voiceprint feature belongs, and at least one cluster in the current cluster corresponds to the current voiceprint feature;

[0154] Step S304: determining a user tag mapping relationship online based on the user tags associated with the historical clustering and the user tags associated with the current clustering;

[0155] Step S305: based on the user label mapping relationship, the user label of the user to which the current voiceprint feature belongs is updated online in the current cluster, and at least the updated user label of the current voiceprint feature is output.

[0156] The detailed process of steps S301 - S305 can refer to the relevant introduction of steps S101 - S105 in the first embodiment, which will not be repeated here.

[0157] Step S306: based on the updated user tag of the current voiceprint feature, update the user tag associated with the current cluster.

[0158] In this embodiment, based on the updated user tag of the current voiceprint feature and the user tag mapping relationship, the user tag associated with the current cluster is updated. Figure 7 As shown, the updated user label of m120 is 3 and the user label mapping relationship, and the user label 111112221112 associated with the current cluster is updated to 111113331113.

[0159] Step S307: Based on the updated user tags associated with the current cluster, update at least a portion of the user tags output at a time before the current time.

[0160] In this implementation, based on the updated user tags associated with the current cluster, a portion of the user tags output at a time before the current time may be updated.

[0161] It is also possible to update all user tags output at a time before the current time based on the updated user tags associated with the current cluster. Figure 8As shown, based on the updated user label 111113331113 associated with the current cluster, all user labels output at one moment before the current moment are updated, and 1111113331113 is output.

[0162] In this embodiment, the audio data to be processed at the current moment is obtained, the current voiceprint features are extracted online from the audio data to be processed, at least one historical clustering at a moment before the current moment is obtained, and the current voiceprint features are clustered offline based on the at least one historical clustering, so that the current voiceprint features are clustered offline based on the existing clustering, which can avoid the cumulative errors caused by the initial clustering errors and improve the accuracy of clustering. At the same time, the current voiceprint features are extracted online and clustered offline, which can reduce the clustering delay and ensure the efficiency of clustering.

[0163] In addition, based on the user tags associated with historical clusters and the user tags associated with the current cluster, the user tag mapping relationship is determined online. Based on the user tag mapping relationship, the user tag of the user to whom the current voiceprint feature belongs is updated online in the current cluster, so that the updated user tag of the current voiceprint feature corresponds to the user tag associated with the historical cluster. This can more accurately characterize the cluster corresponding to the current voiceprint feature, and output the updated user tag of the current voiceprint feature, so that the user can accurately determine the user corresponding to the current voiceprint feature, thereby improving the user experience.

[0164] Furthermore, based on the updated user tag of the current voiceprint feature, the user tag associated with the current cluster is updated, and based on the updated user tag associated with the current cluster, at least a part of the user tag output at a moment before the current moment is updated to achieve the update of the user tag outputted historically.

[0165] Corresponding to the above-mentioned audio processing method implementation method provided by the present application, the present application also provides an implementation method of an audio processing device.

[0166] In this embodiment, if Fig. 9 As shown, the audio processing apparatus may include: an extraction module 100 , an acquisition module 200 , a clustering module 300 , a determination module 400 , an update module 500 and an output module 600 .

[0167] The extraction module 100 is used to obtain the audio data to be processed at the current moment, and extract the current voiceprint features online from the audio data to be processed;

[0168] The acquisition module 200 is used to acquire at least one historical cluster at a time before the current time, where the at least one historical cluster is obtained by clustering at least one historical voiceprint feature corresponding to at least one historical user, each historical cluster corresponds to at least one historical voiceprint feature of a historical user, and is associated with a user tag of the historical user;

[0169] The clustering module 300 is used to perform offline clustering on the current voiceprint feature based on at least one historical cluster to obtain a current cluster, wherein the current cluster is associated with a user label of a user to which the corresponding voiceprint feature belongs, and at least one cluster in the current cluster corresponds to the current voiceprint feature;

[0170] A determination module 400, configured to determine a user tag mapping relationship online based on user tags associated with historical clusters and user tags associated with current clusters;

[0171] An updating module 500 is used to update the user label of the user to which the current voiceprint feature belongs online in the current cluster based on the user label mapping relationship;

[0172] At least output the user label after the current voiceprint feature is updated.

[0173] The output module 600 is used to at least output the user label after the current voiceprint feature is updated.

[0174] In this implementation, the determination module 400 is specifically used to:

[0175] Obtaining a user tag associated with a current cluster and at least one user tag associated with at least one historical cluster;

[0176] Establishing a tag pair mapping relationship for each user tag associated with the current cluster, each tag pair mapping relationship includes a matching factor between the user tag associated with the current cluster and the user tag associated with each historical cluster, and the matching factor represents the matching degree between the user tag associated with the current cluster and the user tag associated with each historical cluster;

[0177] Determine the matching factor of each tag pair mapping relationship whose matching degree meets the conditions;

[0178] Based on the matching factors of each matching degree that meet the conditions, a mapping relationship between the user tags associated with the current cluster and the user tags associated with the historical cluster is determined.

[0179] In this implementation, each tag pair mapping relationship is obtained by comparing a user tag associated with a current cluster with a user tag associated with each historical cluster;

[0180] The determination module 400 may be specifically used for:

[0181] Based on the mapping relationship of each tag pair, N groups of matching factors are obtained, each group of matching factors includes one matching factor in each tag pair mapping relationship, where N is less than the number of user tags associated with the historical clustering;

[0182] A group of matching factors with the largest sum of matching factors among the multiple groups of matching factors is determined as the matching factor whose matching degree in each label pair mapping relationship meets the condition.

[0183] In this implementation, the audio processing device may further include:

[0184] Output update module for:

[0185] Based on the updated user tag of the current voiceprint feature, update the user tag associated with the current cluster;

[0186] Based on the updated user tags associated with the current cluster, at least a portion of the user tags output at a time before the current time is updated.

[0187] In this implementation, the audio processing device may further include:

[0188] A historical cluster determination module is used to:

[0189] Obtain multiple historical voiceprint features before the current moment;

[0190] Based on each historical voiceprint feature as an original historical cluster, the similarity between each two original historical clusters is determined, and the two original historical clusters with the highest similarity are merged into one historical cluster to obtain at least one historical cluster;

[0191] In the case where the number of at least one historical cluster does not reach the set number, returning to the step of determining the similarity between every two original historical clusters until the number of at least one historical cluster reaches the set number, and saving at least one historical cluster;

[0192] The set number is greater than or equal to the number of users corresponding to multiple historical voiceprint features before the current moment.

[0193] In this implementation, the clustering module 300 may be specifically used for:

[0194] Initially determining a current cluster using the current voiceprint feature as a first target cluster, wherein the initially determined current cluster includes the first target cluster and at least one historical cluster;

[0195] Determine the similarity between each two clusters in the initially determined current clusters;

[0196] Determine whether the highest similarity among the similarities between every two clusters is lower than a similarity threshold;

[0197] If it is lower than the similarity threshold, the clustering is terminated and the current clustering is finally obtained;

[0198] If it is not lower than the similarity threshold, the two clusters corresponding to the highest similarity are merged to obtain the second target cluster;

[0199] The initially determined current clusters are updated based on the second target clusters, and the step of determining the similarity between every two clusters in the initially determined current clusters is returned to be executed.

[0200] Corresponding to the above-mentioned embodiment of an audio processing method provided by the present application, the present application also provides an embodiment of an electronic device applying the audio processing method.

[0201] The electronic device may include the following structure:

[0202] The memory 10 and the processor 20 .

[0203] A memory 10, used to store at least one set of instruction sets;

[0204] The processor 20 is used to call and execute the instruction set in the memory 10, and execute the audio processing method introduced in any one of the first to third embodiments by executing the instruction set.

[0205] Corresponding to the above-mentioned audio processing method embodiment provided by the present application, the present application also provides an embodiment of a storage medium.

[0206] In this embodiment, the storage medium stores a computer program for implementing the audio processing method as described in any one of the aforementioned embodiments, and the computer program is executed by a processor to implement the audio processing method as described in any one of the aforementioned embodiments.

[0207] It should be noted that each embodiment focuses on the differences from other embodiments, and the same or similar parts between the embodiments can be referred to each other. For the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0208] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0209] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0210] It can be known from the description of the above implementation methods that those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be essentially or partly contributed to the prior art in the form of a software product, which can be stored in a storage medium such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present application or certain parts of the embodiments.

[0211] The above is a detailed introduction to an audio processing method, device, electronic device and storage medium provided by the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for general technical personnel in this field, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. An audio processing method, comprising: Acquire the audio data to be processed at the current moment, and extract the current voiceprint feature online from the audio data to be processed; Obtaining at least one historical cluster at a time before the current time, wherein the at least one historical cluster is obtained by clustering at least one historical voiceprint feature corresponding to at least one historical user, each of the historical clusters corresponds to at least one historical voiceprint feature of a historical user, and is associated with a user tag of the historical user; Offline clustering of the current voiceprint feature based on the at least one historical cluster is performed to obtain a current cluster, wherein the current cluster is associated with a user tag of a user to which the corresponding voiceprint feature belongs, and at least one cluster in the current cluster corresponds to the current voiceprint feature; Determining a user tag mapping relationship online based on the user tags associated with the historical clusters and the user tags associated with the current clusters; Based on the user label mapping relationship, the user label of the user to which the current voiceprint feature belongs is updated online in the current cluster, and at least the updated user label of the current voiceprint feature is output.

2. According to the method of claim 1, the online determination of the user tag mapping relationship based on the user tags associated with the historical clusters and the user tags associated with the current clusters comprises: Obtaining a user tag associated with the current cluster and at least one user tag associated with the at least one historical cluster; Establishing a tag pair mapping relationship for each user tag associated with the current cluster, each tag pair mapping relationship includes a matching factor between the user tag associated with the current cluster and the user tag associated with each historical cluster, wherein the matching factor represents a matching degree between the user tag associated with the current cluster and the user tag associated with each historical cluster; Determine a matching factor in each of the tag pair mapping relationships whose matching degree satisfies a condition; Based on the matching factors of the respective matching degrees satisfying the conditions, a mapping relationship between the user tags associated with the current cluster and the user tags associated with the historical cluster is determined.

3. According to the method of claim 2, each of the tag pair mapping relationships is obtained by comparing a user tag associated with the current cluster with each user tag associated with the historical cluster; The determining of the matching factors of the tag pair mapping relationships in which the matching degree satisfies the condition includes: Based on each of the tag pair mapping relationships, N groups of matching factors are obtained, each group of matching factors includes one matching factor in each of the tag pair mapping relationships, wherein N is less than the number of user tags associated with the historical clustering; A group of matching factors having the largest sum of matching factors among the multiple groups of matching factors is determined as the matching factors whose matching degrees meet the conditions in the mapping relationships of the respective tag pairs.

4. The method according to claim 1, further comprising: Based on the user tag updated by the current voiceprint feature, updating the user tag associated with the current cluster; Based on the updated user tags associated with the current cluster, at least a portion of the user tags output at a time before the current time is updated.

5. The method according to claim 1, wherein the at least one historical cluster is determined by: Obtain multiple historical voiceprint features before the current moment; Based on each of the historical voiceprint features as an original historical cluster, determine the similarity between every two original historical clusters, merge the two original historical clusters with the highest similarity into one historical cluster, and obtain at least one historical cluster; In the case that the number of the at least one historical cluster does not reach the set number, returning to the step of determining the similarity between every two of the original historical clusters until the number of the at least one historical cluster reaches the set number, and saving the at least one historical cluster; in, The set number is greater than or equal to the number of users corresponding to the multiple historical voiceprint features before the current moment.

6. The method according to claim 1, wherein the performing offline clustering on the current voiceprint feature based on the at least one historical cluster to obtain a current cluster comprises: Initially determining a current cluster using the current voiceprint feature as a first target cluster, wherein the initially determined current cluster includes the first target cluster and the at least one historical cluster; Determining the similarity between every two clusters in the initially determined current clusters; Determine whether the highest similarity among the similarities between each two clusters is lower than a similarity threshold; If it is lower than the similarity threshold, the clustering is terminated, and the current clustering is finally obtained; If it is not lower than the similarity threshold, merging the two clusters corresponding to the highest similarity to obtain a second target cluster; The initially determined current clusters are updated based on the second target clusters, and the step of determining the similarity between every two clusters in the initially determined current clusters is returned to be executed.

7. An audio processing device, comprising: An extraction module, used to obtain the audio data to be processed at the current moment, and extract the current voiceprint features online from the audio data to be processed; an acquisition module, configured to acquire at least one historical cluster at a time before the current time, wherein the at least one historical cluster is obtained by clustering at least one historical voiceprint feature corresponding to at least one historical user, each of the historical clusters corresponds to at least one historical voiceprint feature of a historical user, and is associated with a user tag of the historical user; A clustering module, configured to perform offline clustering on the current voiceprint feature based on the at least one historical cluster to obtain a current cluster, wherein the current cluster is associated with a user tag of a user to which the corresponding voiceprint feature belongs, and at least one cluster in the current cluster corresponds to the current voiceprint feature; A determination module, configured to determine a user tag mapping relationship online based on the user tags associated with the historical clusters and the user tags associated with the current clusters; An updating module, configured to update the user label of the user to which the current voiceprint feature belongs online in the current cluster based on the user label mapping relationship; At least output the updated user label of the current voiceprint feature; An output module is used to output at least the updated user label of the current voiceprint feature.

8. The device according to claim 7, wherein the determining module is specifically configured to: Obtaining a user tag associated with the current cluster and at least one user tag associated with the at least one historical cluster; Establishing a tag pair mapping relationship for each user tag associated with the current cluster, each tag pair mapping relationship includes a matching factor between the user tag associated with the current cluster and the user tag associated with each historical cluster, wherein the matching factor represents a matching degree between the user tag associated with the current cluster and the user tag associated with each historical cluster; Determine a matching factor in each of the tag pair mapping relationships whose matching degree satisfies a condition; Based on the matching factors of the respective matching degrees satisfying the conditions, a mapping relationship between the user tags associated with the current cluster and the user tags associated with the historical cluster is determined.

9. An electronic device, comprising: Memory and processor; The memory is used to store at least one set of instruction sets; The processor is used to call and execute the instruction set in the memory, and execute the audio processing method according to any one of claims 1 to 6 by executing the instruction set.

10. A storage medium storing a computer program for implementing the audio processing method according to any one of claims 1 to 6, wherein the computer program is executed by a processor to implement the audio processing method according to any one of claims 1 to 6.

Images