A human-computer interaction method and device based on dialogue prediction

Abstract

This application discloses a human-computer interaction method and apparatus based on dialogue prediction. The dialogue prediction-based human-computer interaction method includes: acquiring user voice information at a first moment; acquiring phoneme information to be recognized; acquiring basic vehicle information and / or user interaction information within a preset time period before the first moment; acquiring a dynamic language model; acquiring text information corresponding to the phoneme information to be recognized based on the phoneme information and the dynamic language model; and generating human-computer interaction command information based on the text information. This application's dialogue prediction-based human-computer interaction method is more closely aligned with the user's current situation based on phoneme information. The text information recognized through this application's dynamic language model is more accurate, and the dynamic language model is constantly updated based on the current vehicle situation and user interaction information, which reduces translation time each time and is more intelligent and convenient.

Description

Technical Field

[0001] This application relates to the field of vehicle human-computer interaction technology, specifically to a human-computer interaction method and device based on dialogue prediction. Background Technology

[0002] Currently, speech recognition uses a fixed model, and in order to optimize the overall recognition rate, the recognition rate is not particularly high for some low-frequency scenarios.

[0003] To improve recognition rates in certain scenarios, one current strategy is to dynamically register hot words. User-inputted speech is prioritized for matching hot words, thus improving the recognition rate of some text. However, even with this approach, it's impossible to automatically update hot words based on the rhythm of the conversation, resulting in a still low recognition rate in some scenarios.

[0004] Therefore, there is a need for a technical solution to address or at least mitigate the aforementioned shortcomings of existing technologies. Summary of the Invention

[0005] The purpose of this invention is to provide a human-computer interaction method based on dialogue prediction to at least solve one of the above-mentioned technical problems.

[0006] One aspect of the present invention provides a human-computer interaction method based on dialogue prediction, the human-computer interaction method based on dialogue prediction comprising:

[0007] Obtain user voice information in the first moment;

[0008] Obtain the phoneme information to be recognized based on the user's voice information;

[0009] Acquire basic vehicle information and / or user interaction information within a preset time period prior to the first moment;

[0010] A dynamic language model is obtained based on basic vehicle information and / or user interaction information.

[0011] The text information corresponding to the phoneme information to be identified is obtained based on the phoneme information to be identified and the dynamic language model;

[0012] Generate human-computer interaction command information based on the text information.

[0013] Optionally, obtaining the dynamic language model based on user interaction information includes:

[0014] Obtain user intent information based on user interaction information;

[0015] Generate dynamically related text information based on the user intent information;

[0016] Register the dynamically associated text information to the dynamic language model;

[0017] The process of obtaining the dynamic language model based on basic vehicle information includes:

[0018] Obtain information about the interactive features in the current vehicle;

[0019] Obtain the first dialogue database associated with each function that can interact with the user, and each first dialogue database includes multiple phrases;

[0020] The phrases in each of the first speech databases are compared with each other and with the phrases in the second speech database. If any phrase has the same phoneme information as at least one other phrase and the semantic information of the two phrases with the same phoneme information is different, then the phrase is called the same pronunciation phrase.

[0021] Collect words with the same pronunciation and group them to form multiple sets of words. Each set of words contains words with the same phoneme information.

[0022] Register each set of word groups to the dynamic language model.

[0023] Optionally, obtaining the dynamic language model based on basic vehicle information and user interaction information includes:

[0024] Obtain user intent information based on user interaction information;

[0025] Generate dynamically related text information based on the user intent information;

[0026] Obtain information about the interactive features in the current vehicle;

[0027] Obtain the first dialogue database associated with each function that can interact with the user, and each first dialogue database includes multiple phrases;

[0028] The phrases in each of the first speech databases are compared with each other and with the phrases in the second speech database. If any phrase has the same phoneme information as at least one other phrase, then the phrase is called a phrase with the same pronunciation.

[0029] Collect words with the same pronunciation and group them to form multiple sets of words. Each set of words contains words with the same phoneme information.

[0030] Register the dynamically associated text information and each set of word groups to the dynamic language model.

[0031] Optionally, generating dynamically related text information based on the user intent information includes:

[0032] Based on the user intent information, obtain the functions that the vehicle and the user need to interact with;

[0033] Obtain a second dialogue database associated with the vehicle and user interaction functions, each of which includes multiple phrases;

[0034] Register each phrase from the second speech database to the dynamic language model.

[0035] Optionally, obtaining the dynamic language model based on basic vehicle information and user interaction information further includes:

[0036] The phrases in each of the first speech databases are compared with each other and with the phrases in the second speech database. If any phrase does not have the same phoneme information as at least one other phrase, then the phrase is called a phrase with different pronunciations.

[0037] Register each word with a different pronunciation to a candidate dynamic language model.

[0038] Optionally, when the dynamic language model is obtained based on vehicle basic information and user interaction information, the step of obtaining the text information corresponding to the phoneme information based on the phoneme information and the dynamic language model includes:

[0039] The phoneme information of each word group in each second speech database is obtained as the first phoneme information;

[0040] The similarity of the phoneme information to be identified with each first phoneme information is calculated to determine whether there is a similarity exceeding a preset threshold. If so, the word group corresponding to the first phoneme information with a similarity exceeding the preset threshold is obtained as the first candidate word group.

[0041] The phoneme information of each group of words is obtained as the third phoneme information;

[0042] The similarity between the phoneme information to be identified and each third phoneme information is calculated to determine whether any similarity exceeds a preset threshold. If not, then...

[0043] The first candidate word group is used as the text information corresponding to the phoneme information to be identified.

[0044] Optionally, when the dynamic language model is obtained based on vehicle basic information and user interaction information, the step of obtaining the text information corresponding to the phoneme information based on the phoneme information and the dynamic language model includes:

[0045] The similarity of the phoneme information to be identified with each third phoneme information is calculated to determine whether there is a similarity exceeding a preset threshold. If so, the set of word groups corresponding to the third phoneme information with similarity exceeding the preset threshold is obtained as the candidate word group set.

[0046] Each phrase in the set of candidate phrases is selected as a second candidate phrase;

[0047] The time when user interaction information is generated, and the time when information on interactive functions in the current vehicle corresponding to the second candidate phrase is generated;

[0048] A first weight is generated based on the time generated from the first moment and the user interaction information;

[0049] Based on the time generated by the first moment and the time generated by the interactive function information of the current vehicle corresponding to each second candidate word group, a second weight is generated for the time generated by the interactive function information of the current vehicle corresponding to each second candidate word group.

[0050] Obtain a historical usage database, which includes at least one pre-device selected word group and the usage weight corresponding to each pre-device selected word;

[0051] Obtain the usage weight corresponding to the pre-selected word group that is the same as the first candidate word group;

[0052] Obtain the usage weight corresponding to the same pre-selected word groups for each second candidate word group;

[0053] Based on the first weight, the second weight, and each used weight, obtain one of the first candidate word groups or each of the second candidate word groups as the text information corresponding to the phoneme information to be identified.

[0054] Optionally, the step of obtaining the text information corresponding to the phoneme information to be identified by one of the first candidate word groups or each of the second candidate word groups according to the first weight, the second weight, and each of the usage weights includes:

[0055] The final weight of the first candidate word group is obtained based on the usage weight of the first candidate word group and the first weight;

[0056] Perform the following operations for each second alternative phrase:

[0057] The final weight of the second candidate word group is obtained based on the usage weight of the second candidate word group and the second weight corresponding to the interactive function information of the current vehicle.

[0058] The first or second candidate word group with the highest final weight among the first candidate word group and the second candidate word group is used as the text information corresponding to the phoneme information to be identified.

[0059] Optionally, when the dynamic language model is obtained based on vehicle basic information and user interaction information, the step of obtaining the text information corresponding to the phoneme information based on the phoneme information and the dynamic language model includes:

[0060] The phoneme information of each word group in each second speech database is obtained as the first phoneme information;

[0061] The similarity of the phoneme information to be identified with each of the first phoneme information is calculated to determine whether any similarity exceeds a preset threshold. If not, then...

[0062] The phoneme information of each group of words is obtained as the third phoneme information;

[0063] The similarity between the phoneme information to be identified and each third phoneme information is calculated to determine whether any similarity exceeds a preset threshold. If not, then...

[0064] The phoneme information of each different pronunciation phrase in each different pronunciation phrase in the candidate dynamic language model is obtained as the fourth phoneme information;

[0065] The similarity between the phoneme information to be identified and each of the fourth phonemes is calculated to determine whether any similarity exceeds a preset threshold. If so, then...

[0066] The pronunciation phrases corresponding to the fourth phoneme information with a similarity exceeding a preset threshold are obtained as the text information corresponding to the phoneme information to be identified.

[0067] This application also provides a human-computer interaction device based on dialogue prediction, the human-computer interaction device based on dialogue prediction comprising:

[0068] User voice information acquisition module, the user voice information acquisition module is used to acquire user voice information at the first moment;

[0069] A phoneme information acquisition module is used to acquire phoneme information to be identified based on user voice information.

[0070] A vehicle basic information acquisition module, which is used to acquire vehicle basic information prior to the first moment.

[0071] User interaction information acquisition module, the user interaction information acquisition module is used to acquire user interaction information before the first moment;

[0072] A dynamic language model acquisition module, which is used to acquire a dynamic language model based on basic vehicle information and / or user interaction information;

[0073] A text information acquisition module is used to acquire text information corresponding to the phoneme information to be identified based on the phoneme information to be identified and the dynamic language model.

[0074] A human-computer interaction command information acquisition module is used to generate human-computer interaction command information based on the text information. Beneficial effects

[0075] The human-computer interaction method based on dialogue prediction in this application obtains a dynamic language model based on vehicle basic information and / or user interaction information when performing corresponding text recognition based on phoneme information. Since this dynamic language model is obtained based on vehicle basic information and / or user interaction information, the generated dynamic language model is not based on a fixed template or hot words, but rather on the user's interaction information and / or vehicle basic information within a certain time period. This allows the dynamic language model to be more closely aligned with the user's current situation, resulting in more accurate text recognition. Furthermore, the dynamic language model is constantly updated based on the current vehicle conditions and user interaction information, reducing translation time each time and making it more intelligent and convenient. Attached Figure Description

[0076] Figure 1 This is a flowchart illustrating a human-computer interaction method based on dialogue prediction according to an embodiment of this application.

[0077] Figure 2 This is a schematic diagram of an electronic device capable of implementing a dialogue prediction-based human-computer interaction method according to an embodiment of this application. Detailed Implementation

[0078] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this application will be described in more detail below with reference to the accompanying drawings. In the drawings, the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The described embodiments are some, but not all, embodiments of this application. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. The embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0079] Figure 1 This is a flowchart illustrating a human-computer interaction method based on dialogue prediction according to an embodiment of this application.

[0080] like Figure 1 The human-computer interaction methods based on dialogue prediction shown include:

[0081] Step 1: Obtain the user's voice information at the first moment;

[0082] Step 2: Obtain the phoneme information to be recognized based on the user's voice information;

[0083] Step 3: Obtain basic vehicle information and / or user interaction information within a preset time period prior to the first moment;

[0084] Step 4: Obtain a dynamic language model based on basic vehicle information and / or user interaction information;

[0085] Step 5: Obtain the text information corresponding to the phoneme information to be identified based on the phoneme information to be identified and the dynamic language model;

[0086] Step 6: Generate human-computer interaction command information based on the text information.

[0087] The human-computer interaction method based on dialogue prediction in this application obtains a dynamic language model based on vehicle basic information and / or user interaction information when performing corresponding text recognition based on phoneme information. Since this dynamic language model is obtained based on vehicle basic information and / or user interaction information, the generated dynamic language model is not based on a fixed template or hot words, but rather on the user's interaction information and / or vehicle basic information within a certain time period. This allows the dynamic language model to be more closely aligned with the user's current situation, resulting in more accurate text recognition. Furthermore, the dynamic language model is constantly updated based on the current vehicle conditions and user interaction information, reducing translation time each time and making it more intelligent and convenient.

[0088] In this embodiment, the phoneme information to be identified is obtained based on the user's voice information. For example, if the user's voice is "replay", the phoneme recognition engine can identify "replay" as "chongbo".

[0089] In this embodiment, obtaining the dynamic language model based on user interaction information includes:

[0090] Obtain user intent information based on user interaction information;

[0091] Generate dynamically related text information based on the user intent information;

[0092] Register dynamically associated text information to the dynamic language model.

[0093] In this embodiment, obtaining the dynamic language model based on basic vehicle information includes:

[0094] Obtain information about the interactive features in the current vehicle;

[0095] Obtain the first dialogue database associated with each function that can interact with the user, and each first dialogue database includes multiple phrases;

[0096] The phrases in each of the first speech databases are compared with each other and with the phrases in the second speech database. If any phrase has the same phoneme information as at least one other phrase and the semantic information of the two phrases with the same phoneme information is different, then the phrase is called the same pronunciation phrase.

[0097] Collect words with the same pronunciation and group them to form multiple sets of words. Each set of words contains words with the same phoneme information.

[0098] Register each set of word groups to the dynamic language model.

[0099] In this embodiment, obtaining the dynamic language model based on basic vehicle information and user interaction information includes:

[0100] Obtain user intent information based on user interaction information;

[0101] Generate dynamically related text information based on the user intent information;

[0102] Obtain information about the interactive features in the current vehicle;

[0103] Obtain the first dialogue database associated with each function that can interact with the user, and each first dialogue database includes multiple phrases;

[0104] The phrases in each of the first speech databases are compared with each other and with the phrases in the second speech database. If any phrase has the same phoneme information as at least one other phrase, then the phrase is called a phrase with the same pronunciation.

[0105] Collect words with the same pronunciation and group them to form multiple sets of words. Each set of words contains words with the same phoneme information.

[0106] Register the dynamically associated text information and each set of word groups to the dynamic language model.

[0107] In this embodiment, generating dynamically related text information based on the user intent information includes:

[0108] Based on the user intent information, obtain the functions that the vehicle and the user need to interact with;

[0109] Obtain a second dialogue database associated with the vehicle and user interaction functions, each of which includes multiple phrases;

[0110] Register each phrase from the second speech database to the dynamic language model.

[0111] In this embodiment, obtaining the dynamic language model based on basic vehicle information and user interaction information further includes:

[0112] The phrases in each of the first speech databases are compared with each other and with the phrases in the second speech database. If any phrase does not have the same phoneme information as at least one other phrase, then the phrase is called a phrase with different pronunciations.

[0113] Register each word with a different pronunciation to a candidate dynamic language model.

[0114] In this embodiment, when the dynamic language model obtains information based on vehicle basic information and user interaction information, obtaining the text information corresponding to the phoneme information based on the phoneme information and the dynamic language model includes:

[0115] The phoneme information of each word group in each second speech database is obtained as the first phoneme information;

[0116] The similarity of the phoneme information to be identified with each first phoneme information is calculated to determine whether there is a similarity exceeding a preset threshold. If so, the word group corresponding to the first phoneme information with a similarity exceeding the preset threshold is obtained as the first candidate word group.

[0117] The phoneme information of each group of words is obtained as the third phoneme information;

[0118] The similarity between the phoneme information to be identified and each third phoneme information is calculated to determine whether any similarity exceeds a preset threshold. If not, then...

[0119] The first candidate word group is used as the text information corresponding to the phoneme information to be identified.

[0120] In this embodiment, when the dynamic language model obtains information based on vehicle basic information and user interaction information, obtaining the text information corresponding to the phoneme information based on the phoneme information and the dynamic language model includes:

[0121] The similarity of the phoneme information to be identified with each third phoneme information is calculated to determine whether there is a similarity exceeding a preset threshold. If so, the set of word groups corresponding to the third phoneme information with similarity exceeding the preset threshold is obtained as the candidate word group set.

[0122] Each phrase in the set of candidate phrases is selected as a second candidate phrase;

[0123] The time when user interaction information is generated, and the time when information on interactive functions in the current vehicle corresponding to the second candidate phrase is generated;

[0124] A first weight is generated based on the time generated from the first moment and the user interaction information;

[0125] Based on the time generated by the first moment and the time generated by the interactive function information of the current vehicle corresponding to each second candidate word group, a second weight is generated for the time generated by the interactive function information of the current vehicle corresponding to each second candidate word group.

[0126] Obtain a historical usage database, which includes at least one pre-device selected word group and the usage weight corresponding to each pre-device selected word;

[0127] Obtain the usage weight corresponding to the pre-selected word group that is the same as the first candidate word group;

[0128] Obtain the usage weight corresponding to the same pre-selected word groups for each second candidate word group;

[0129] Based on the first weight, the second weight, and each used weight, obtain one of the first candidate word groups or each of the second candidate word groups as the text information corresponding to the phoneme information to be identified.

[0130] Using this method, when there are multiple alternative phrases, you can choose the one with the highest probability based on the actual situation.

[0131] In this embodiment, obtaining the text information corresponding to the phoneme information to be identified by one of the first candidate word groups or each of the second candidate word groups according to the first weight, the second weight, and each used weight includes:

[0132] The final weight of the first candidate word group is obtained based on the usage weight of the first candidate word group and the first weight;

[0133] Perform the following operations for each second alternative phrase:

[0134] The final weight of the second candidate word group is obtained based on the usage weight of the second candidate word group and the second weight corresponding to the interactive function information of the current vehicle.

[0135] The first or second candidate word group with the highest final weight among the first candidate word group and the second candidate word group is used as the text information corresponding to the phoneme information to be identified.

[0136] In this embodiment, when the dynamic language model obtains information based on vehicle basic information and user interaction information, obtaining the text information corresponding to the phoneme information based on the phoneme information and the dynamic language model includes:

[0137] The phoneme information of each word group in each second speech database is obtained as the first phoneme information;

[0138] The similarity of the phoneme information to be identified with each of the first phoneme information is calculated to determine whether any similarity exceeds a preset threshold. If not, then...

[0139] The phoneme information of each group of words is obtained as the third phoneme information;

[0140] The similarity between the phoneme information to be identified and each third phoneme information is calculated to determine whether any similarity exceeds a preset threshold. If not, then...

[0141] The phoneme information of each different pronunciation phrase in each different pronunciation phrase in the candidate dynamic language model is obtained as the fourth phoneme information;

[0142] The similarity between the phoneme information to be identified and each of the fourth phonemes is calculated to determine whether any similarity exceeds a preset threshold. If so, then...

[0143] The pronunciation phrases corresponding to the fourth phoneme information with a similarity exceeding a preset threshold are obtained as the text information corresponding to the phoneme information to be identified.

[0144] The following examples further illustrate this application in detail. It is understood that these examples do not constitute any limitation on this application. Example

[0145] Obtain user voice information in the first moment;

[0146] Obtain the phoneme information to be recognized based on the user's voice information;

[0147] Obtain basic vehicle information and / or user interaction information within a preset time period before the first moment (for example, the preset time period is 1 minute, and within 1 minute before the first moment, the basic vehicle information includes whether the user turned on the air conditioner or opened a music software, and the user interaction information is whether the user asked the vehicle to open the in-vehicle address book).

[0148] A dynamic language model is obtained based on basic vehicle information and / or user interaction information.

[0149] In this embodiment, a dynamic language model is obtained based on basic vehicle information and user interaction information.

[0150] Specifically, the following method is used to obtain a dynamic language model based on basic vehicle information and user interaction information:

[0151] The user intent information is obtained based on the user interaction information (as can be seen from the above description, the user intent information is to open the in-vehicle address book).

[0152] Generating dynamically related text information based on the user intent information, specifically, generating dynamically related text information based on the user intent information includes:

[0153] Based on the user intent information, obtain the vehicle and user interaction functions to be performed (in this embodiment, the vehicle and user interaction functions to be performed are the address book functions).

[0154] Obtain the second dialogue database associated with the vehicle and user interaction functions (in the vehicle, each function will pre-store some dialogue, for example, the address book function will pre-store the names of each contact in the address book for voice broadcast, etc.), each second dialogue database includes multiple phrases;

[0155] Register each phrase in the second speech database to the dynamic language model (it is understood that the language model is existing technology and will not be described in detail here).

[0156] Obtain information about the functions in the current vehicle that can interact with the user (in this embodiment, the air conditioning function and the QQ Music software function can both interact with the user).

[0157] Obtain the first dialogue database associated with each function that can interact with the user (for example, the first dialogue database in the air conditioning function may include temperature rise and temperature fall), and each first dialogue database includes multiple phrases;

[0158] The phrases in each of the first speech databases are compared with each other and with the phrases in the second speech database. If any phrase has the same phoneme information as at least one other phrase, then the phrase is called a phrase with the same pronunciation (in this embodiment, there are no phrases with the same phoneme information, so they do not need to be obtained).

[0159] Collect words with the same pronunciation and group them to form multiple sets of words. Each set of words contains words with the same phoneme information.

[0160] Register the dynamically associated text information and each set of word groups to the dynamic language model.

[0161] In this embodiment, obtaining the dynamic language model based on basic vehicle information and user interaction information further includes:

[0162] The phrases in each of the first speech databases are compared with each other and with the phrases in the second speech database. If any phrase does not have the same phoneme information as at least one other phrase, then the phrase is called a phrase with different pronunciations.

[0163] Each different pronunciation phrase is registered to a candidate dynamic language model (in this embodiment, the above-mentioned temperature rise and temperature fall are both different pronunciation phrases, therefore, they are all registered to the candidate dynamic language model).

[0164] In this embodiment, the text information corresponding to the phoneme to be identified is obtained based on the phoneme information to be identified and the dynamic language model. Specifically, this embodiment adopts the following method:

[0165] The phoneme information of each word group in each second speech database is obtained as the first phoneme information;

[0166] The similarity of the phoneme information to be identified with each first phoneme information is calculated to determine whether there is a similarity exceeding a preset threshold. If so, the word group corresponding to the first phoneme information with a similarity exceeding the preset threshold is obtained as the first candidate word group.

[0167] The phoneme information of each group of words is obtained as the third phoneme information;

[0168] The similarity score between the phoneme information to be identified and each third phoneme information is calculated to determine whether any similarity score exceeds a preset threshold. If not, then...

[0169] The first candidate word group is taken as the text information corresponding to the phoneme information to be identified;

[0170] Generate human-computer interaction command information based on the text information. Example

[0171] Obtain the user's voice information at the first moment (in this embodiment, the voice information obtained is for replay).

[0172] The phoneme information to be identified is obtained based on the user's voice information (the replay is used to perform phoneme recognition, thereby obtaining the phoneme information to be identified, chongbo).

[0173] Obtain basic vehicle information and / or user interaction information within a preset time period before the first moment (for example, the preset time period is 1 minute, and within 1 minute before the first moment, the basic vehicle information includes whether the user turned on the air conditioner or opened a music software, and the user interaction information is whether the user asked the vehicle to open the in-vehicle address book).

[0174] A dynamic language model is obtained based on basic vehicle information and / or user interaction information.

[0175] In this embodiment, a dynamic language model is obtained based on basic vehicle information and user interaction information.

[0176] Specifically, the following method is used to obtain a dynamic language model based on basic vehicle information and user interaction information:

[0177] The user intent information is obtained based on the user interaction information (as can be seen from the above description, the user intent information is to open the in-vehicle address book).

[0178] Generating dynamically related text information based on the user intent information, specifically, generating dynamically related text information based on the user intent information includes:

[0179] Based on the user intent information, obtain the vehicle and user interaction functions to be performed (in this embodiment, the vehicle and user interaction functions to be performed are the address book functions).

[0180] Obtain the second dialogue database associated with the vehicle and the user's interactive functions (in the vehicle, each function will pre-store some dialogue, for example, the address book function will pre-store the names of each contact in the address book for voice broadcast, etc.), each second dialogue database includes multiple phrases (for example, in this embodiment, the second dialogue database of the address book includes the name Chongbo).

[0181] Register each phrase in the second speech database to the dynamic language model (that is, register Chongbo to the dynamic language model; it is understood that the language model is existing technology and will not be described in detail here).

[0182] Obtain information about the functions in the current vehicle that can interact with the user (in this embodiment, the air conditioning function and the QQ Music software function can both interact with the user).

[0183] Obtain the first phrase database associated with each function information capable of interacting with the user (for example, in the first phrase database of the air conditioner function, it may include temperature increase and temperature decrease, and in the first phrase database of the QQ Music software, one function is replay, and this function also appears as a phrase in the first phrase database). Each first phrase database includes multiple phrases;

[0184] Compare the phrases in each first phrase database with each other and separately with the phrases in the second phrase database. If any one phrase has the same phoneme information as at least one other phrase, then this phrase is called a same pronunciation phrase (chongbo and replay have the same phoneme information, so chongbo and replay are same pronunciation phrases);

[0185] Obtain each same pronunciation phrase and group them to form multiple sets of phrase collections. Each phrase included in each set of phrase collections has the same phoneme information;

[0186] Register the dynamic associated text information and each set of phrase collections to the dynamic language model.

[0187] In this embodiment, obtaining the dynamic language model according to the vehicle basic information and user interaction information further includes:

[0188] Compare the phrases in each first phrase database with each other and separately with the phrases in the second phrase database. If any one phrase does not have the same phoneme information as at least one other phrase, then this phrase is called a different pronunciation phrase;

[0189] Register each different pronunciation phrase to the alternative dynamic language model (in this embodiment, the above temperature increase and temperature decrease both belong to different pronunciation phrases, so they are both registered to the alternative dynamic language model).

[0190] In this embodiment, obtaining the text information corresponding to the to-be-recognized phoneme information according to the to-be-recognized phoneme information and the dynamic language model. Specifically, this embodiment adopts the following method:

[0191] Respectively obtain the phoneme information (chongbo) of each phrase in each second phrase database as the first phoneme information;

[0192] Calculate the similarity between the to-be-recognized phoneme information (chongbo) and each first phoneme information (chongbo) respectively, so as to determine whether there is a similarity exceeding a preset threshold. If so, obtain the phrase corresponding to the first phoneme information with the similarity exceeding the preset threshold as the first alternative phrase (chongbo);

[0193] The phoneme information of each group of words is obtained as the third phoneme information;

[0194] The similarity of the phoneme information to be identified with each third phoneme information is calculated to determine whether any similarity exceeds a preset threshold. If so (in this embodiment, the replayed phoneme information chongbo exceeds the preset threshold), then...

[0195] Obtain the set of word groups corresponding to third phoneme information with similarity exceeding a preset threshold as the candidate word group set;

[0196] Each phrase in the set of candidate phrases is selected as a second candidate phrase (replay, repeat).

[0197] The time of obtaining user interaction information (e.g., in this embodiment, the user interaction information is that the user requests the vehicle to open the in-vehicle address book, and the time is 10:10:10) and the time of obtaining the function information of the current vehicle that can interact with the user corresponding to the second candidate phrase (e.g., the air conditioning function is 10:10:10, and the time of opening a music software is 10:09:50) are all obtained. Since the hours are the same, only the minutes and seconds are described below for the sake of convenience.

[0198] A first weight is generated based on the first moment (in this embodiment, the first moment is 10 minutes and 20 seconds) and the time generated by the user interaction information;

[0199] The first weight is generated using the following method:

[0200] Set up a weighted database. For example, the weighted database stipulates that if the difference from the first time is within 10 seconds, the weight is 10, and if the difference from the first time is within 20 seconds but more than 10 seconds, the weight is 5. For example, in the above case, the first time is 10 minutes and 20 seconds, and the time to open the car's address book is 10 minutes and 10 seconds, which is a difference of 10 seconds, so the first weight is 10.

[0201] Based on the time generated by the first moment and the interactive function information of the current vehicle corresponding to each second candidate word group, a second weight is generated for the time generated by the interactive function information of the current vehicle corresponding to each second candidate word group. For example, in the above, the time when a music software is opened is 10:09:50, which is 20 seconds different from the first moment, so the second weight is 5.

[0202] Obtain a historical usage database, which includes at least one pre-device selected word group and the usage weight corresponding to each pre-device selected word;

[0203] Obtain the usage weight corresponding to the pre-set alternative phrase that is the same as the first alternative phrase (for example, the word "chongbo" has only been used 3 times in history, so the usage weight is 3);

[0204] Respectively obtain the usage weight corresponding to the pre-set alternative phrase that is the same as each second alternative phrase (for example, the word "chongbo" has been used 10 times in history, so the usage weight is 10. Since "chongbo" also belongs to the first alternative phrase, the usage weight is still 3 as above);

[0205] Obtain one of the first alternative phrase or each second alternative phrase as the text information corresponding to the to-be-recognized phoneme information according to the first weight, the second weight, and each usage weight.

[0206] Specifically, obtain the final weight of the first alternative phrase according to the usage weight of the first alternative phrase and the first weight (the final weight of the word "chongbo" is 10 + 3 = 13, that is, the final weight is 13);

[0207] Perform the following operations for each second alternative phrase:

[0208] Obtain the final weight of the second alternative phrase according to the usage weight of the second alternative phrase and the second weight corresponding to the function information that can interact with the user in the current vehicle corresponding to the second alternative phrase (the final weight of the word "chongbo" is 5 + 10 = 15, that is, the final weight is 15, and the final weight of the word "chongbo" is 10 + 3 = 13, that is, the final weight is 13);

[0209] Obtain the first alternative phrase or the second alternative phrase corresponding to the highest value among the final weights of the first alternative phrase and the final weights of each second alternative phrase as the text information corresponding to the to-be-recognized phoneme information (that is, obtain "chongbo").

[0210] Perform corresponding actions according to "chongbo". For example, if a certain song is being played at that time, replay the song.

[0211] Example 3:

[0212] Obtain the user voice information at the first moment (in this embodiment, the obtained user voice information is "turn on hotspot");

[0213] Obtain the to-be-recognized phoneme information according to the user voice information (perform phoneme recognition on "chongbo" to obtain the to-be-recognized phoneme information "kairedian");

[0214] Obtain basic vehicle information and / or user interaction information within a preset time period before the first moment (for example, the preset time period is 1 minute, and within 1 minute before the first moment, the basic vehicle information includes whether the user turned on the air conditioner or opened a music software, and the user interaction information is whether the user asked the vehicle to open the in-vehicle address book).

[0215] A dynamic language model is obtained based on basic vehicle information and / or user interaction information.

[0216] In this embodiment, a dynamic language model is obtained based on basic vehicle information and user interaction information.

[0217] Specifically, the following method is used to obtain a dynamic language model based on basic vehicle information and user interaction information:

[0218] The user intent information is obtained based on the user interaction information (as can be seen from the above description, the user intent information is to open the in-vehicle address book).

[0219] Generating dynamically related text information based on the user intent information, specifically, generating dynamically related text information based on the user intent information includes:

[0220] Based on the user intent information, obtain the vehicle and user interaction functions to be performed (in this embodiment, the vehicle and user interaction functions to be performed are the address book functions).

[0221] Obtain the second dialogue database associated with the vehicle and the user's interactive functions (in the vehicle, each function will pre-store some dialogue, for example, the address book function will pre-store the names of each contact in the address book for voice broadcast, etc.), each second dialogue database includes multiple phrases (for example, in this embodiment, the second dialogue database of the address book includes the name Chongbo).

[0222] Register each phrase in the second speech database to the dynamic language model (that is, register Chongbo to the dynamic language model; it is understood that the language model is existing technology and will not be described in detail here).

[0223] Obtain information about the functions in the current vehicle that can interact with the user (in this embodiment, the air conditioning function and the QQ Music software function can both interact with the user).

[0224] Obtain the first dialogue database associated with each function that can interact with the user (for example, the first dialogue database in the air conditioning function may include turning on the hotspot and turning on the coldspot, while one function in the first dialogue database of QQ Music software is replay, which also appears as a phrase in the first dialogue database), and each first dialogue database includes multiple phrases.

[0225] The phrases in each of the first speech databases are compared with each other and with the phrases in the second speech database. If any phrase has the same phoneme information as at least one other phrase, then the phrase is called a phrase with the same pronunciation.

[0226] Collect words with the same pronunciation and group them to form multiple sets of words. Each set of words contains words with the same phoneme information.

[0227] Register the dynamically associated text information and each set of word groups to the dynamic language model.

[0228] In this embodiment, obtaining the dynamic language model based on basic vehicle information and user interaction information further includes:

[0229] The phrases in each of the first speech databases are compared with each other and with the phrases in the second speech database. If any phrase does not have the same phoneme information as at least one other phrase, then the phrase is called a phrase with different pronunciations.

[0230] Each different pronunciation phrase is registered to the candidate dynamic language model (in this embodiment, the above-mentioned hot spots and cold spots are different pronunciation phrases, so they are all registered to the candidate dynamic language model).

[0231] In this embodiment, the text information corresponding to the phoneme to be identified is obtained based on the phoneme information to be identified and the dynamic language model. Specifically, this embodiment adopts the following method:

[0232] The phoneme information (chongbo) of each word group in each second speech database is obtained as the first phoneme information;

[0233] The similarity score of the phoneme information to be identified (chongbo) is calculated with each of the first phoneme information (chongbo) to determine whether any of them exceeds a preset threshold. If not, then...

[0234] The phoneme information of each group of words is obtained as the third phoneme information;

[0235] The similarity between the phoneme information to be identified and each third phoneme information is calculated to determine whether any similarity exceeds a preset threshold. If not, then...

[0236] The phoneme information of each different pronunciation phrase in each candidate dynamic language model is obtained as the fourth phoneme information (kairedian, kailengdian).

[0237] The similarity score between the phoneme information to be identified and each of the fourth phonemes is calculated to determine whether any similarity score exceeds a preset threshold. If so (kairedian), then...

[0238] The pronunciation phrases corresponding to the fourth phoneme information with a similarity exceeding a preset threshold are obtained as the text information corresponding to the phoneme information to be identified (hotspots).

[0239] In this embodiment, the user interaction information can also be the vehicle's question information. For example, the vehicle's question information is "Who do you want to call? Please say the contact's name." In this case, the user intent information pointed to by the vehicle's question information is still the address book.

[0240] This application also provides a dialogue prediction-based human-computer interaction device, which includes a user voice information acquisition module, a phoneme information acquisition module, a vehicle basic information acquisition module, a user interaction information acquisition module, a dynamic language model acquisition module, a text information acquisition module, and a human-computer interaction command information acquisition module.

[0241] The user voice information acquisition module is used to acquire user voice information in the first moment.

[0242] The phoneme information acquisition module is used to acquire phoneme information to be recognized based on user voice information;

[0243] The vehicle basic information acquisition module is used to acquire the vehicle's basic information up to the first moment.

[0244] The user interaction information acquisition module is used to acquire user interaction information prior to the first moment.

[0245] The dynamic language model acquisition module is used to acquire a dynamic language model based on basic vehicle information and / or user interaction information;

[0246] The text information acquisition module is used to acquire the text information corresponding to the phoneme information to be identified based on the phoneme information to be identified and the dynamic language model;

[0247] The human-computer interaction command information acquisition module is used to generate human-computer interaction command information based on the text information.

[0248] It is understandable that the above description of the method also applies to the description of the apparatus.

[0249] This application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the computer program to implement the above-described dialogue prediction-based human-computer interaction method.

[0250] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, enables the above-described human-computer interaction method based on dialogue prediction.

[0251] Figure 2 This is an exemplary structural diagram of an electronic device capable of implementing a dialogue prediction-based human-computer interaction method according to an embodiment of this application.

[0252] like Figure 2 As shown, the electronic device includes an input device 501, an input interface 502, a central processing unit 503, a memory 504, an output interface 505, and an output device 506. The input interface 502, central processing unit 503, memory 504, and output interface 505 are interconnected via a bus 507. The input device 501 and output device 506 are connected to the bus 507 via the input interface 502 and output interface 505, respectively, and thus connected to other components of the electronic device. Specifically, the input device 504 receives input information from the outside and transmits it to the central processing unit 503 via the input interface 502. The central processing unit 503 processes the input information based on computer-executable instructions stored in the memory 504 to generate output information, temporarily or permanently storing the output information in the memory 504, and then transmitting the output information to the output device 506 via the output interface 505. The output device 506 outputs the output information to the outside of the electronic device for user use.

[0253] In other words, Figure 2 The illustrated electronic device may also be implemented as including: a memory storing computer-executable instructions; and one or more processors, which can be coupled when executing the computer-executable instructions. Figure 1 The method for human-computer interaction based on dialogue prediction is described.

[0254] In one embodiment, Figure 2 The electronic device shown can be implemented as including: a memory 504 configured to store executable program code; and one or more processors 503 configured to run the executable program code stored in the memory 504 to perform the dialogue prediction-based human-computer interaction method in the above embodiments.

[0255] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0256] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0257] Computer-readable media include both permanent and non-permanent, removable and non-removable media, and information storage can be achieved by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, DVD or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0258] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0259] Furthermore, it is clear that the word "comprising" does not exclude other units or steps. Multiple units, modules, or devices recited in the apparatus claims may also be implemented by a single unit or overall apparatus via software or hardware.

[0260] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code, which includes one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutively marked blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or the overall flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0261] In this embodiment, the processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0262] Memory can be used to store computer programs and / or modules. The processor implements various functions of the device / terminal equipment by running or executing the computer programs and / or modules stored in the memory, and by accessing data stored in the memory. Memory can mainly include a program storage area and a data storage area. The program storage area can store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area can store data created based on the use of the mobile phone (such as audio data, phonebook, etc.). In addition, memory can include high-speed random access memory, and can also include non-volatile memory, such as hard disks, RAM, plug-in hard disks, smart media cards (SMC), secure digital (SD) cards, flash cards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.

[0263] In this embodiment, if the modules / units integrated into the device / terminal equipment are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. Although this application discloses preferred embodiments as described above, it is not intended to limit this application. Any person skilled in the art can make possible changes and modifications without departing from the spirit and scope of this application. Therefore, the scope of protection of this application should be determined by the scope defined in the claims of this application.

[0264] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0265] Furthermore, it is clear that the word "comprising" does not exclude other units or steps. Multiple units, modules, or devices recited in the apparatus claims may also be implemented by a single unit or overall apparatus via software or hardware.

[0266] Although the present invention has been described in detail above with general descriptions and specific embodiments, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, all such modifications or improvements made without departing from the spirit of the present invention fall within the scope of protection claimed by the present invention.

Claims

1. A human-computer interaction method based on dialogue prediction, characterized in that, The human-computer interaction method based on dialogue prediction includes: Obtain user voice information in the first moment; Obtain the phoneme information to be recognized based on the user's voice information; Acquire basic vehicle information and user interaction information within a preset time period prior to the first moment; A dynamic language model is obtained based on basic vehicle information and user interaction information. The text information corresponding to the phoneme information to be identified is obtained based on the phoneme information to be identified and the dynamic language model; Generate human-computer interaction command information based on the text information; The process of obtaining a dynamic language model based on basic vehicle information and user interaction information includes: Obtain user intent information based on user interaction information; Generate dynamically related text information based on the user intent information; the generation of dynamically related text information based on the user intent information includes: obtaining the vehicle and user interaction functions based on the user intent information; obtaining a second dialogue database associated with the vehicle and user interaction functions, each second dialogue database including multiple word groups; Obtain information about the interactive features in the current vehicle; Obtain the first dialogue database associated with each function that can interact with the user, and each first dialogue database includes multiple phrases; The phrases in each of the first speech databases are compared with each other and with the phrases in the second speech database. If any phrase has the same phoneme information as at least one other phrase, then the phrase is called a phrase with the same pronunciation. Collect words with the same pronunciation and group them to form multiple sets of words. Each set of words contains words with the same phoneme information. Register the dynamically associated text information and each set of word groups to the dynamic language model; The step of obtaining a dynamic language model based on basic vehicle information and user interaction information further includes: The phrases in each of the first speech databases are compared with each other and with the phrases in the second speech database. If any phrase does not have the same phoneme information as at least one other phrase, then the phrase is called a phrase with different pronunciations. Register each word with a different pronunciation to a candidate dynamic language model.

2. The human-computer interaction method based on dialogue prediction as described in claim 1, characterized in that, When the dynamic language model is obtained based on vehicle basic information and user interaction information, the step of obtaining the text information corresponding to the phoneme information based on the phoneme information and the dynamic language model includes: The phoneme information of each word group in each second speech database is obtained as the first phoneme information; The similarity of the phoneme information to be identified with each first phoneme information is calculated to determine whether there is a similarity exceeding a preset threshold. If so, the word group corresponding to the first phoneme information with a similarity exceeding the preset threshold is obtained as the first candidate word group. The phoneme information of each group of words is obtained as the third phoneme information; The similarity between the phoneme information to be identified and each third phoneme information is calculated to determine whether any similarity exceeds a preset threshold. If not, then... The first candidate word group is used as the text information corresponding to the phoneme information to be identified.

3. The human-computer interaction method based on dialogue prediction as described in claim 2, characterized in that, When the dynamic language model is obtained based on vehicle basic information and user interaction information, the step of obtaining the text information corresponding to the phoneme information based on the phoneme information and the dynamic language model includes: The similarity of the phoneme information to be identified with each third phoneme information is calculated to determine whether there is a similarity exceeding a preset threshold. If so, the set of word groups corresponding to the third phoneme information with similarity exceeding the preset threshold is obtained as the candidate word group set. Each phrase in the set of candidate phrases is selected as a second candidate phrase; The time when user interaction information is generated, and the time when information on interactive functions in the current vehicle corresponding to the second candidate phrase is generated; A first weight is generated based on the time generated from the first moment and the user interaction information; Based on the time generated by the first moment and the time generated by the interactive function information of the current vehicle corresponding to each second candidate word group, a second weight is generated for the time generated by the interactive function information of the current vehicle corresponding to each second candidate word group. Obtain a historical usage database, which includes at least one pre-device selected word group and the usage weight corresponding to each pre-device selected word; Obtain the usage weight corresponding to the pre-selected word group that is the same as the first candidate word group; Obtain the usage weight corresponding to the same pre-selected word groups for each second candidate word group; Based on the first weight, the second weight, and each used weight, obtain one of the first candidate word groups or each of the second candidate word groups as the text information corresponding to the phoneme information to be identified.

4. The human-computer interaction method based on dialogue prediction as described in claim 3, characterized in that, The step of obtaining the text information corresponding to the phoneme information to be identified by one of the first candidate word groups or each of the second candidate word groups based on the first weight, the second weight, and each used weight includes: The final weight of the first candidate word group is obtained based on the usage weight of the first candidate word group and the first weight; Perform the following operations for each second alternative phrase: The final weight of the second candidate word group is obtained based on the usage weight of the second candidate word group and the second weight corresponding to the interactive function information of the current vehicle. The first or second candidate word group with the highest final weight among the first candidate word group and the second candidate word group is used as the text information corresponding to the phoneme information to be identified.

5. The human-computer interaction method based on dialogue prediction as described in claim 4, characterized in that, When the dynamic language model is obtained based on vehicle basic information and user interaction information, the step of obtaining the text information corresponding to the phoneme information based on the phoneme information and the dynamic language model includes: The phoneme information of each word group in each second speech database is obtained as the first phoneme information; The similarity of the phoneme information to be identified with each of the first phoneme information is calculated to determine whether any similarity exceeds a preset threshold. If not, then... The phoneme information of each group of words is obtained as the third phoneme information; The similarity between the phoneme information to be identified and each third phoneme information is calculated to determine whether any similarity exceeds a preset threshold. If not, then... The phoneme information of each different pronunciation phrase in each different pronunciation phrase in the candidate dynamic language model is obtained as the fourth phoneme information; The similarity between the phoneme information to be identified and each of the fourth phonemes is calculated to determine whether any similarity exceeds a preset threshold. If so, then... The pronunciation phrases corresponding to the fourth phoneme information with a similarity exceeding a preset threshold are obtained as the text information corresponding to the phoneme information to be identified.

6. A human-computer interaction device based on dialogue prediction, used in the human-computer interaction method based on dialogue prediction as described in any one of claims 1 to 5, characterized in that, The human-computer interaction device based on dialogue prediction includes: User voice information acquisition module, the user voice information acquisition module is used to acquire user voice information at the first moment; A phoneme information acquisition module is used to acquire phoneme information to be identified based on user voice information. A vehicle basic information acquisition module, which is used to acquire vehicle basic information prior to the first moment. User interaction information acquisition module, the user interaction information acquisition module is used to acquire user interaction information before the first moment; A dynamic language model acquisition module, which is used to acquire a dynamic language model based on basic vehicle information and user interaction information; A text information acquisition module is used to acquire text information corresponding to the phoneme information to be identified based on the phoneme information to be identified and the dynamic language model. A human-computer interaction command information acquisition module is used to generate human-computer interaction command information based on the text information.

Images