Voice information processing method and device

Abstract

This disclosure provides a voice information processing method and apparatus. The method includes: receiving target voice input from a user; vectorizing the target voice to obtain a first vector; acquiring the user's location information; generating prompt information including the location information; and vectorizing the prompt information to obtain a second vector; inputting the first vector and the second vector into a natural language processing model; the natural language processing model acquiring a set of points of interest based on the second vector; and identifying whether the first vector includes points of interest based on the set of points of interest. This can improve the accuracy of identifying points of interest from voice and also improve the recall rate of points of interest.

Description

Technical Field

[0001] This disclosure relates to the fields of computer and speech recognition technology, and in particular to a speech information processing method and device. Background Technology

[0002] With the development of terminal devices and artificial intelligence technology, human-computer interaction has gradually evolved into multimodal human-computer interaction, including text and voice. Voice interaction mainly refers to users sending voice information to terminal devices, which then perform Automatic Speech Recognition (ASR) on the voice information, converting the user's input into text, and executing corresponding instructions or providing responses based on the text.

[0003] In some application scenarios, the voice information input by users may include points of interest (POIs) from geographic information systems. Currently, the accuracy of identifying POIs from voice information is low. Summary of the Invention

[0004] This disclosure provides a voice information processing method and apparatus to overcome the current problem of low accuracy in identifying points of interest from voice information.

[0005] In a first aspect, embodiments of this disclosure provide a voice information processing method, including:

[0006] Receive the target speech input by the user, and vectorize the target speech to obtain a first vector;

[0007] Obtain the user's location information, generate a prompt message including the location information, and vectorize the prompt message to obtain a second vector;

[0008] The first vector and the second vector are input into a natural language processing model. The natural language processing model obtains a set of interest points based on the second vector and identifies whether the first vector contains interest points based on the set of interest points.

[0009] Secondly, embodiments of this disclosure provide a voice information processing device, including:

[0010] A receiving unit is configured to receive target speech input by a user and vectorize the target speech to obtain a first vector.

[0011] The acquisition unit is configured to acquire the user's location information, generate prompt information including the location information, and vectorize the prompt information to obtain a second vector.

[0012] The recognition unit is used to input the first vector and the second vector into the natural language processing model, so that the natural language processing model can obtain a set of interest points based on the second vector, and recognize whether the first vector includes interest points based on the set of interest points.

[0013] Thirdly, embodiments of this disclosure provide an electronic device, including: a processor and a memory;

[0014] The memory stores computer-executed instructions;

[0015] The processor executes computer execution instructions stored in the memory, causing the at least one processor to perform the voice information processing method as described in the first aspect and various possible designs of the first aspect.

[0016] Fourthly, embodiments of this disclosure provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the voice information processing method described in the first aspect and various possible designs of the first aspect.

[0017] Fifthly, embodiments of this disclosure provide a computer program product, including a computer program that, when executed by a processor, implements the voice information processing method as described in the first aspect and various possible designs of the first aspect.

[0018] The voice information processing method and device provided in this embodiment input voice information and prompt information including the user's location information into a natural language processing model. The natural language processing model identifies the voice information based on a set of points of interest associated with the location indicated by the location information, which can improve the accuracy of identifying points of interest from the voice information and also improve the recall rate of points of interest. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 Flowchart of the voice information processing method provided in the embodiments of this disclosure Figure 1 ;

[0021] Figure 2 This is a schematic diagram of an application scenario;

[0022] Figure 3Flowchart of the voice information processing method provided in the embodiments of this disclosure Figure 2 ;

[0023] Figure 4 This is a schematic diagram of an application scenario;

[0024] Figure 5 This is a structural block diagram of a voice information processing device provided in an embodiment of the present disclosure;

[0025] Figure 6 This is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of this disclosure. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0027] To identify points of interest (POIs in this disclosure refer to POIs in a geographic information system), related technologies can pre-set a list of hotspot POIs and pre-load this list as a hot word stimulus word list into the speech recognition model. If a word in the user's input speech information exists in the list of hotspot POIs, the speech recognition model will stimulate the recognition path of this POI entry to improve the model's recognition performance for POIs.

[0028] The aforementioned scheme requires storing hotspots of interest in a list beforehand. The accuracy and recall of interest point recognition decrease as the number of entries in the list increases; the more entries in the list, the worse the recognition performance of individual interest point entries. Furthermore, interest points with the same or similar pronunciations in the hotspot list are easily confused, making it difficult to accurately identify interest points based on the user's speech. Therefore, related technologies suffer from low accuracy in recognizing interest points in speech.

[0029] The solution provided in this disclosure inputs voice and prompts including the user's geographic location information into a natural language processing model. The natural language processing model then uses the prompts to obtain a set of points of interest associated with the geographic location information. By referring to the set of points of interest, the model can infer and identify points of interest in the voice information, thereby improving the accuracy of identifying points of interest from the voice.

[0030] It should be understood that in the solutions provided in this disclosure, the user's geographical location information is either actively provided by the user or obtained with the user's authorization.

[0031] Please refer to Figure 1 , Figure 1 Flowchart of the voice information processing method provided in this disclosure Figure 1 ,like Figure 1 As shown, the method includes the following steps:

[0032] S101: Receive the target speech input by the user and vectorize the target speech to obtain the first vector.

[0033] The entity executing the voice information processing method can be a terminal device, specifically a pre-defined application client running on the terminal device; or a server, specifically a pre-defined application server running on the server to provide services to the application client on the terminal device. In some application scenarios, the aforementioned pre-defined application can convert user-inputted voice into text; in other application scenarios, the aforementioned pre-defined application can perform pre-defined operations based on user-inputted voice.

[0034] The aforementioned execution entity can receive target speech input by the user. The target speech can be any speech input by the user. The aforementioned execution entity can use various algorithms for vectorizing speech information to vectorize the target speech, obtaining a first vector.

[0035] As an implementation, an audio encoder can be used to convert the target speech into a lower-dimensional numerical representation, known as audio embedding. Specifically, this can include the following processes: First, preprocessing. In this step, the target speech is resampled to ensure it has a preset sampling rate, noise is removed, and the noise-removed target speech is segmented. Second, feature extraction. In this step, features are extracted from each segment of the target speech. Feature extraction can be performed using various methods, such as Mel-frequency cepstral coefficients, Mel-spectral graphs, and wavelet transforms. The extracted features aim to capture key information from the target speech while reducing the amount of data. Third, encoding. In this step, an encoder model is used to encode the extracted audio features. The encoder model can be a machine learning model, such as a support vector machine; it can also be a deep learning-based model, such as a convolutional neural network, a recurrent neural network, or an encoder-decoder architecture. The encoder model learns how to map the input audio features into a fixed-size vector space, obtaining a first vector that retains the main semantic information of the target speech.

[0036] S102: Obtain the user's location information, generate a prompt message including the location information, and vectorize the prompt message to obtain a second vector.

[0037] The aforementioned executing entity can obtain the user's location information based on user authorization. This location information represents the user's position in space. In one example, the location information includes the approximate address of the user's current location.

[0038] In some implementations, step S102 includes: obtaining location information input by the user; or, based on authorization, obtaining the user's location information from a location service application on the user's terminal.

[0039] As an example, users can manually enter location information. Understandably, users can enter location information through text or voice, or by selecting a location on a map.

[0040] The location information entered by the user can be, for example, the user's current location. It can also be a location other than the user's current location. In this example, even if the user's input location is not their current location, it can still be considered that the target audio input by the user is highly likely related to the point of interest of that location.

[0041] As an example, the user's terminal device (i.e., the user's terminal) runs a location service application, which can obtain the user's location information with the user's authorization. In this example, the obtained location information can be the user's current location.

[0042] It should be noted that there is no sequential relationship between steps S101 and S102. Step S102 can be performed before step S101 or simultaneously with step S101.

[0043] In some implementations, the prompt message is represented as text, and the second vector is a vector obtained by vectorizing the text. In these implementations, the prompt message may also include other prompt content, such as the output format, the logic for processing the first vector, and examples, which are typically represented in text form to ensure that the prompt message is not misinterpreted.

[0044] In one example, the aforementioned prompt information may include a field for writing location information, where the text representing the location information can be written. The prompt information can be encoded using text encoding to obtain a second vector corresponding to the prompt information. Furthermore, the prompt information may also include indication information that instructs the natural language processing model to identify whether the first vector contains interest points based on a set of interest points associated with the location information.

[0045] The process of vectorizing the above prompt information is as follows: First, preprocessing. In this step, the prompt information is segmented into words or sub-word units. Second, model or algorithm selection. In this step, it is determined which model or algorithm to use to generate the vector. The aforementioned models can include, for example, statistical method-based models and deep learning models. Statistical method-based models include the Bag of Words (BoW) model, TF-IDF (Term Frequency-Inverse Document Frequency), etc. Deep learning models include pre-trained language models, etc. Third, feature extraction or encoding. In this step, for statistical method-based models, each word or n-gram (a combination of n consecutive words) is mapped into a vector space, forming a sparse matrix. For deep learning models, the text passes through neural network layers, which can capture contextual information and output dense vector representations. For example, after the input text passes through multiple bidirectional encoders, each token (here, a token refers to an entity in the text, which can be a word, character, sub-word, or punctuation mark) will obtain a corresponding vector representation, thus obtaining the second vector of the prompt information. The above second vector contains all the information of the prompt information.

[0046] S103: Input the first vector and the second vector into the natural language processing model. The natural language processing model obtains the set of interest points based on the second vector and identifies whether the first vector contains interest points based on the set of interest points.

[0047] In some implementations, step S103 includes concatenating the first vector and the second vector and then inputting the concatenated vector into a natural language processing model. The prompt information includes textual content containing prompt information, including location information. The second vector is a vector obtained by vectorizing the text; while the first vector is a vector obtained by vectorizing the target speech. Therefore, the natural language processing model can identify the first vector and the second vector from the concatenated vector.

[0048] The first and second vectors are concatenated and then input into the natural language processing model so that the model can obtain the user's real-time location information and then obtain the points of interest corresponding to the real-time location information, which helps to improve the accuracy of identifying points of interest from speech.

[0049] The natural language processing model can use the first vector as the data to be processed and identify the position information and the processing logic for processing the first vector from the second vector. Since the second vector includes the user's position information, this position information can trigger the natural language model to determine a set of points of interest associated with this position information according to its own corpus. In addition, the logic for processing the first vector in the above prompt information can, for example, include the following steps for processing the first vector: first, obtain a set of points of interest associated with the position indicated by the position information, and then preferentially refer to the set of points of interest to identify the text corresponding to the first vector.

[0050] The above points of interest can refer to locations or entities in a geographic information system that can arouse users' interest or attention, such as hotels, shopping malls, scenic spots, banks, and hospitals, etc.

[0051] In one example, multiple points of interest and the addresses of each point of interest can be obtained, and a set of points of interest associated with the user's position information can be determined according to the user's position information and the addresses of each point of interest. For example, a set of the above points of interest is composed of multiple points of interest whose distance from the user's location is less than or equal to a preset distance threshold.

[0052] After determining the set of points of interest associated with the position information, each point of interest in the set of points of interest can be referred to to detect whether the target voice information includes a point of interest. For example, multiple points of interest in the set of points of interest can be used as a reference, and according to a preset speech recognition method, it can be determined whether the target voice information includes a point of interest.

[0053] Specifically, several frames of speech in the target voice information correspond to one phoneme, and several phonemes can be mapped to one character.

[0054] Since the pronunciations of different characters may be the same or similar, it is possible to map to incorrect characters when mapping several phonemes to characters. For example, the pronunciations of both "yi" and "yi" are "yi". When recognizing the target voice information, an incorrect recognition result may be obtained, and thus it may be impossible to recognize the accurate point of interest. For example, when there are two points of interest, "yi shopping mall" and "yi shopping mall", it is very likely to confuse the two, and the accuracy rate of the point of interest recognized from the voice is low. In addition, it is also possible that no point of interest can be recognized from the voice.

[0055] The set of points of interest associated with the user's position information can be used as a reference set for converting audio to text. If the pronunciations of several phonemes and the pronunciations of several phonemes are the same as the characters of each point of interest in the set of points of interest, the above several factors can be recognized as points of interest. In this way, even if there are other characters with the same or similar pronunciation as one or more characters in the point of interest, since the point of interest is used as a reference set, the above several factors can be preferentially converted into points of interest.

[0056] The natural language processing model can extract one or more interest points associated with the location indicated by the location information from its own corpus based on the second vector, forming an interest point set. Then, the target speech is recognized by referring to the interest point set.

[0057] Natural language processing (NLP) models can use a set of points of interest (POIs) as auxiliary information for identifying the first vector. That is, during the process of mapping the first vector to text, the NLP model can prioritize mapping each POI from the POI set. Therefore, for words with the same pronunciation as POIs, the NLP model will identify the POIs in the first vector with a higher probability when referring to the POI set associated with the user's location information.

[0058] Please refer to Figure 2 , Figure 2 This is a diagram illustrating an application scenario. For example... Figure 2 As shown, users can input voice into the execution entity of the voice information processing method via voice input. Any voice input by the user can be considered as the target voice. Assuming the user inputs the target voice "Where is e-shopping mall?", after receiving the user's target voice 211, the execution entity can perform audio vectorization to obtain a first vector. This first vector is then input into the concatenation module 23. Furthermore, the execution entity can generate prompt information 221 including the user's geographical location and perform text vectorization on the prompt information to obtain a second vector. This second vector is then input into the concatenation module 23. The concatenation module 23 can concatenate the first and second vectors to obtain a concatenated vector. The concatenation module 23 can then input the concatenated vector into the natural language processing model 24. The natural language processing model can identify the geographical location from the second vector and search its learned corpus for at least one point of interest associated with the geographical location to form a set of points of interest. For example, the set of points of interest could be {e-shopping mall, XX subway entrance}. The natural language processing model can refer to the set of points of interest to identify the first vector, obtaining the corresponding target voice recognition result converted into the text "Where is e-shopping mall?". Instead of identifying the target speech as "Where is Yishang Mall?", where "Yishang Mall" could be a point of interest that is far from the user's location.

[0059] In this embodiment, by generating prompt information including location information, the second vector corresponding to the prompt information and the first vector of the target speech are input into a natural language processing model. The natural language processing model then identifies the target speech based on a set of interest points associated with the location information. This improves the problem of misidentification of interest points in speech and increases the accuracy of interest points identified from user speech. Furthermore, since the set of interest points is referenced when identifying the target language, multiple phonemes with the same pronunciation as the interest points are likely to be identified as interest points when identifying the target speech, thus improving the recall rate of interest points.

[0060] Please refer to Figure 3 , Figure 3 Flowchart of the voice information processing method provided in this disclosure Figure 2 ,like Figure 3 The method shown includes, in addition to, with Figure 1 In addition to step S301, which is the same as step S101 in the illustrated embodiment, the following steps are also included:

[0061] S302: Obtain the user's location information and retrieve at least one point of interest near the location indicated by the location information.

[0062] In this embodiment, the executing entity of the voice information processing method can be a terminal device, specifically an application client running on the terminal device; the aforementioned executing entity can be a server, specifically an application server running on the server.

[0063] After receiving the user's location information, the system can search for points of interest near the location indicated by the location information in an external knowledge base, and obtain at least one point of interest.

[0064] The aforementioned external knowledge base could be, for example, a geographic database that the aforementioned executing entity has the authority to access, a social networking platform focused on geographic location, a map service application, etc.

[0065] In one example, one or more geographic databases can be accessed first. A spatial query can then be performed within these databases, centered on the location indicated by the location information, with a preset search radius, to obtain at least one point of interest associated with that location. In another example, a search for points of interest near the location can be performed on a location-focused social networking platform to obtain at least one point of interest. In yet another example, the aforementioned entity can access the aforementioned map application, input location information and a search radius within the map application, and receive at least one point of interest returned by the map service application.

[0066] S303: Determine at least one target interest point from at least one interest point, and construct an interest point set from the at least one target interest point.

[0067] In one example, all of the above-mentioned points of interest can be used as target points of interest.

[0068] In one example, at least one point of interest can be filtered to select at least one target point of interest from the at least one point of interest.

[0069] In some embodiments, step S303 above includes the following steps:

[0070] First, sort at least one point of interest according to preset rules.

[0071] Secondly, select at least one target point of interest from the ranking results.

[0072] In one example, at least one point of interest can be sorted according to its distance from the location indicated by the location information above, for example, sorting at least one point of interest in ascending order of distance to obtain a sorting result.

[0073] In one example, at least one point of interest can be sorted according to its access frequency, for example, by sorting it in descending order of access frequency, to obtain the sorting result.

[0074] In one example, at least one point of interest can be sorted according to the category to which each point of interest belongs. For example, the sorting order of different categories can be preset, and for each category, at least one point of interest belonging to that category can be sorted according to the sorting order of that category to obtain the sorting result of the point of interest in that category.

[0075] One or more target points of interest can be selected from the above sorting results. In one example, points of interest with serial numbers less than a preset serial number threshold can be selected as target points of interest. The preset serial number threshold can be an integer greater than or equal to 1. The preset serial number threshold can be set according to specific application scenarios, and there are no restrictions here.

[0076] In these implementations, at least one point of interest is sorted according to preset rules, and a target point of interest is selected from the sorting results. These preset rules can be set according to different needs. The resulting target point of interest can, to some extent, meet the specific needs of the user, help improve the accuracy of interest points recognized from speech, and contribute to improving the user experience.

[0077] S304: Obtain the user's location information, generate a prompt message including the location information and a set of points of interest, and vectorize the prompt message to obtain a second vector.

[0078] In one example, the above prompt message may include a location information field and a point of interest (POI) field. Users can enter their location information in the location information field and a set of POIs, including at least one target POI, in the POI field.

[0079] The aforementioned execution entity can vectorize the prompt information containing location information and at least one target point of interest to obtain a second vector. For example, the prompt information represented in text form can be vectorized to obtain the second vector.

[0080] S305: Input the first vector and the second vector into the natural language processing model. The natural language processing model extracts information about the set of interest points from the second vector and identifies whether the first vector contains interest points based on the information about the set of interest points.

[0081] In addition to location information, the above prompt information may also include target interest points determined from the search results obtained by searching for interest points in an external knowledge base through steps S302 to S304. The location information and target interest points can enhance the prompt information. Therefore, the natural language processing model does not need to search for interest points associated with the location information. It can directly use the target interest points in the prompt information as auxiliary information for recognizing the target speech.

[0082] Please refer to Figure 4 , Figure 4 For example, a schematic diagram of an application scenario. Figure 4As shown, users can input voice information into the execution entity of the voice information processing method via voice input. Any voice input by the user can be considered as the target voice. Assuming the user inputs the target voice "Where is e-shopping mall?", after receiving the user's target voice 411, the target voice 411 can be audio vectorized 412 to obtain a first vector, which is then input into the splicing module 43. Furthermore, the execution entity can generate prompt information 421 including the user's geographical location, which can then be sent to the retrieval module 422. The retrieval module 422 can search for points of interest (POIs) in the geographic database 423 based on the location information, within a space centered on the geographical location and at a preset radius from the center, obtaining multiple candidate POIs. The retrieval module 422 can also sort the multiple candidate POIs and select at least one target POI from the sorting results to form a POI set. The POI set is written into the prompt information to obtain enhanced prompt information 424. The enhanced prompt information 424 can be text vectorized 425 to obtain a second vector. The second vector is then input into the splicing module 43. The concatenation module 43 can concatenate the first vector and the second vector and input them into the natural language processing model 44. The natural language processing model 44 can identify the first vector and the second vector based on the received concatenated vector, and then parse the set of points of interest from the second vector. Schematic, the set of points of interest may include {e shopping mall, XX subway entrance, YY hospital}. The result of the natural language processing model's identification of the first vector is "Where is e shopping mall?", and the natural language processing model can output the above "Where is e shopping mall?" to the aforementioned execution entity for display.

[0083] In this embodiment, points of interest (POIs) are retrieved from an external database based on location information, and the prompt information is enhanced based on the retrieval results. The enhanced prompt information may include a set of POIs associated with the location information. The natural language processing (NLP) model can identify the set of POIs from the second vector, and convert the first vector into text using this set of POIs as a reference. When converting the first vector into text, the target POIs in the set of POIs are preferentially referenced. In this process, the NLP model does not need to retrieve POIs associated with the location information from its own corpus; it can directly use the set of POIs in the prompt information as auxiliary information for speech recognition, which can improve the efficiency of speech recognition.

[0084] exist Figure 1 or Figure 3 In some embodiments of the voice information processing method shown, the method further includes:

[0085] In response to the recognition that the target speech includes points of interest, the target speech is converted into text including the points of interest.

[0086] Specifically, a natural language processing model can convert the target speech information into text that includes identified points of interest.

[0087] It should be noted that this method can also perform other tasks based on the identified points of interest, such as generating navigation paths to the points of interest and generating descriptive information about the points of interest.

[0088] In these implementations, points of interest in the speech can be accurately converted into corresponding text during the speech-to-text process, thus improving the accuracy of speech-to-text conversion.

[0089] Corresponding to the voice information processing method in the above embodiments, Figure 5 This is a structural block diagram of a voice information processing device provided in an embodiment of this disclosure. For ease of explanation, only the parts relevant to the embodiments of this disclosure are shown. (Refer to...) Figure 5 The device includes: a receiving unit 501, an acquisition unit 502, and an identification unit 503.

[0090] The receiving unit 501 is used to receive the target speech input by the user and vectorize the target speech to obtain a first vector.

[0091] The acquisition unit 502 is used to acquire the user's location information, generate prompt information including the location information, and vectorize the prompt information to obtain a second vector.

[0092] The recognition unit 503 is used to input the first vector and the second vector into the natural language processing model, so that the natural language processing model can obtain a set of interest points based on the second vector, and recognize whether the first vector includes interest points based on the set of interest points.

[0093] In one embodiment of this disclosure, device 50 further includes a retrieval unit (not shown in the figures), the retrieval unit being used for:

[0094] Retrieve at least one point of interest near the location indicated by the location information;

[0095] Determine at least one target interest point from at least one interest point, and generate an interest point set including at least one target interest point;

[0096] The identification unit 503 is further used for:

[0097] Generate prompts that include location information and a set of points of interest.

[0098] In one embodiment of this disclosure, the retrieval unit is further configured to:

[0099] Sort at least one point of interest according to preset rules;

[0100] Select at least one target point of interest from the sorting results.

[0101] In one embodiment of this disclosure, the identification unit 503 is further configured to:

[0102] The first and second vectors are input into the natural language processing model, which extracts information about the set of interest points from the second vector and identifies whether the first vector contains interest points based on the information about the set of interest points.

[0103] In one embodiment of this disclosure, the prompt information is represented by text, and the second vector is a vector obtained by vectorizing the text.

[0104] In one embodiment of this disclosure, the identification unit 503 is further configured to:

[0105] The first and second vectors are concatenated and then input into the natural language processing model.

[0106] In one embodiment of this disclosure, the acquisition unit 502 is further configured to:

[0107] Get the location information input by the user; or,

[0108] Based on authorization, the user's location information is obtained from the location service application on the user's terminal.

[0109] In one embodiment of this disclosure, device 50 further includes a conversion unit (not shown), the conversion unit being used for:

[0110] In response to the recognition that the target speech includes points of interest, the target speech is converted into text including the points of interest.

[0111] The device provided in this embodiment can be used to execute the technical solutions of the above method embodiments. Its implementation principle and technical effect are similar, and will not be described again here.

[0112] To implement the above embodiments, this disclosure also provides an electronic device.

[0113] refer to Figure 6The diagram illustrates a structural schematic of an electronic device 600 suitable for implementing embodiments of the present disclosure. The electronic device 600 can be a terminal device or a server. The terminal device can include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, personal digital assistants (PDAs), tablet computers, portable media players (PMPs), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 6 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0114] like Figure 6 As shown, electronic device 600 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 602 or a program loaded from storage device 608 into random access memory (RAM) 603. RAM 603 also stores various programs and data required for the operation of electronic device 600. The processing unit 601, ROM 602, and RAM 603 are interconnected via bus 604. Input / output (I / O) interface 605 is also connected to bus 604.

[0115] Typically, the following devices can be connected to I / O interface 605: input devices 606 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 607 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 608 including, for example, magnetic tapes, hard disks, etc.; and communication devices 609. Communication device 609 allows electronic device 600 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 6 An electronic device 600 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively.

[0116] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 609, or installed from a storage device 608, or installed from a ROM 602. When the computer program is executed by the processing device 601, it performs the functions defined in the methods of embodiments of this disclosure.

[0117] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of 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 or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can 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 program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0118] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0119] The aforementioned computer-readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the methods shown in the above embodiments.

[0120] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0121] 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 disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing 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 the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, 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.

[0122] The units described in the embodiments of this disclosure can be implemented in software or in hardware. The name of a unit does not necessarily limit the unit itself; for example, the first acquisition unit can also be described as "a unit that acquires at least two Internet Protocol addresses".

[0123] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.

[0124] In a first aspect, according to one or more embodiments of the present disclosure, a voice information processing method is provided, comprising:

[0125] Receive the target speech input by the user, and vectorize the target speech to obtain the first vector;

[0126] Obtain the user's location information, generate a prompt message including the location information, and vectorize the prompt message to obtain a second vector;

[0127] The first and second vectors are input into the natural language processing model. The natural language processing model obtains a set of interest points based on the second vector and identifies whether the first vector contains interest points based on the set of interest points.

[0128] According to one or more embodiments of this disclosure, the method further includes:

[0129] Retrieve at least one point of interest near the location indicated by the location information;

[0130] Determine at least one target interest point from at least one interest point, and generate an interest point set including at least one target interest point;

[0131] And generate prompts including location information, including:

[0132] Generate prompts that include location information and a set of points of interest.

[0133] According to one or more embodiments of this disclosure, determining at least one target point of interest from at least one point of interest includes:

[0134] Sort at least one point of interest according to preset rules;

[0135] Select at least one target point of interest from the sorting results.

[0136] According to one or more embodiments of this disclosure, a first vector and a second vector are input into a natural language processing model, which obtains a set of interest points based on the second vector, and identifies whether the first vector includes interest points based on the set of interest points, including:

[0137] The first and second vectors are input into the natural language processing model, which extracts information about the set of interest points from the second vector and identifies whether the first vector contains interest points based on the information about the set of interest points.

[0138] According to one or more embodiments of this disclosure, the prompt information is represented by text, and the second vector is a vector obtained by vectorizing the text.

[0139] According to one or more embodiments of this disclosure, inputting a first vector and a second vector into a natural language processing model includes: concatenating the first vector and the second vector and then inputting them into the natural language processing model.

[0140] According to one or more embodiments of this disclosure, obtaining a user's location information includes:

[0141] Get the location information input by the user; or,

[0142] Based on authorization, the user's location information is obtained from the location service application on the user's terminal.

[0143] According to one or more embodiments of this disclosure, the method further includes:

[0144] In response to the recognition that the target speech includes points of interest, the target speech is converted into text including the points of interest.

[0145] Secondly, according to one or more embodiments of this disclosure, a voice information processing device is provided, comprising:

[0146] The receiving unit is used to receive the target speech input by the user and vectorize the target speech to obtain a first vector.

[0147] The acquisition unit is used to acquire the user's location information, generate prompt information including the location information, and vectorize the prompt information to obtain a second vector;

[0148] The recognition unit is used to input the first vector and the second vector into the natural language processing model, and the natural language processing model obtains the set of interest points based on the second vector, and recognizes whether the first vector contains interest points based on the set of interest points.

[0149] According to one or more embodiments of this disclosure, the device further includes a retrieval unit, the retrieval unit being configured to:

[0150] Retrieve at least one point of interest near the location indicated by the location information;

[0151] Determine at least one target interest point from at least one interest point, and generate an interest point set including at least one target interest point;

[0152] The identification unit is further used for:

[0153] Generate prompts that include location information and a set of points of interest.

[0154] In one embodiment of this disclosure, the retrieval unit is further configured to:

[0155] Sort at least one point of interest according to preset rules;

[0156] Select at least one target point of interest from the sorting results.

[0157] According to one or more embodiments of this disclosure, the identification unit is further configured to:

[0158] The first and second vectors are input into the natural language processing model, which extracts information about the set of interest points from the second vector and identifies whether the first vector contains interest points based on the information about the set of interest points.

[0159] According to one or more embodiments of this disclosure, the prompt information is represented by text, and the second vector is a vector obtained by vectorizing the text.

[0160] According to one or more embodiments of this disclosure, the identification unit is further configured to:

[0161] The first and second vectors are concatenated and then input into the natural language processing model.

[0162] According to one or more embodiments of this disclosure, the acquisition unit is further configured to:

[0163] Get the location information input by the user; or,

[0164] Based on authorization, the user's location information is obtained from the location service application on the user's terminal.

[0165] According to one or more embodiments of this disclosure, the device further includes a conversion unit, the conversion unit being configured to:

[0166] In response to the recognition that the target speech includes points of interest, the target speech is converted into text including the points of interest.

[0167] Thirdly, according to one or more embodiments of the present disclosure, an electronic device is provided, comprising: at least one processor and a memory;

[0168] The memory stores the instructions that the computer executes;

[0169] At least one processor executes computer execution instructions stored in memory, causing at least one processor to perform the voice information processing method as described in the first aspect above and various possible designs of the first aspect.

[0170] Fourthly, according to one or more embodiments of the present disclosure, a computer-readable storage medium is provided, which stores computer-executable instructions that, when executed by a processor, implement the voice information processing method described in the first aspect and various possible designs of the first aspect.

[0171] Fifthly, according to one or more embodiments of the present disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the voice information processing method as described in the first aspect and various possible designs of the first aspect.

[0172] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

[0173] Furthermore, while the operations are described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in a sequential order. In certain environments, multitasking and parallel processing may be advantageous. Similarly, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

[0174] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. A method for processing voice information, comprising: Receive the target speech input by the user, and vectorize the target speech to obtain a first vector; Obtain the user's location information, generate a prompt message including the location information, and vectorize the prompt message to obtain a second vector; The first vector and the second vector are input into a natural language processing model. The natural language processing model obtains a set of interest points based on the second vector and identifies whether the first vector contains interest points based on the set of interest points.

2. The method of claim 1, wherein, The method further includes: Retrieve at least one point of interest near the location indicated by the location information; Determine at least one target interest point from the at least one interest point, and generate an interest point set including the at least one target interest point; And the generation of prompt information including the location information includes: Generate a prompt message that includes the location information and the set of points of interest.

3. The method of claim 2, wherein, Determining at least one target point of interest from the at least one point of interest includes: The at least one point of interest is sorted according to a preset rule; Select at least one target point of interest from the sorting results.

4. The method of claim 2, wherein, The step of inputting the first vector and the second vector into a natural language processing model, whereby the natural language processing model obtains a set of interest points based on the second vector, and identifies whether the first vector includes interest points based on the set of interest points, includes: The first vector and the second vector are input into a natural language processing model, which extracts information about the set of interest points from the second vector and identifies whether the first vector contains interest points based on the information about the set of interest points.

5. The method of claim 1, wherein, The prompt information is represented by text, and the second vector is a vector obtained by vectorizing the text.

6. The method according to any one of claims 1 to 5, characterized in that, The step of inputting the first vector and the second vector into the natural language processing model includes: The first vector and the second vector are concatenated and then input into the natural language processing model.

7. The method according to any one of claims 1 to 5, characterized in that, The acquisition of the user's location information includes: Obtain the location information input by the user; or, Based on authorization, the user's location information is obtained from the location service application on the user's terminal.

8. The method according to any one of claims 1-5, characterized in that, The method further includes: In response to the recognition that the target speech includes points of interest, the target speech is converted into text including the points of interest.

9. A voice information processing device, comprising: A receiving unit is configured to receive target speech input by a user and vectorize the target speech to obtain a first vector. The acquisition unit is configured to acquire the user's location information, generate prompt information including the location information, and vectorize the prompt information to obtain a second vector. The recognition unit is used to input the first vector and the second vector into the natural language processing model, so that the natural language processing model can obtain a set of interest points based on the second vector, and recognize whether the first vector includes interest points based on the set of interest points.

10. An electronic device, characterized in that, include: Processor and memory; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the voice information processing method as described in any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by the processor, implement the voice information processing method as described in any one of claims 1 to 8.

12. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the voice information processing method as described in any one of claims 1 to 8.

Images