Translation method and system based on sound field detection

By monitoring the sound field of the rear seats in the vehicle voice system and dynamically adjusting the voice acquisition time, the problem of fixed time capture is solved, and the adaptability and privacy protection of voice interaction is achieved, ensuring efficient resource utilization and data Safety.

CN118918890APending Publication Date: 2024-11-08IOL WUHAN INFORMATION TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410920951.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-10
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

The existing vehicle voice system uses a fixed time to capture subsequent voice sequences after capturing activation words, which cannot adapt to the conversation habits of different users and cannot support the communication needs of non-car computer operations, which has waste of resources and data privacy concerns.

Method used

By monitoring the sound field of the rear seat, we can determine whether the voice sequence contains preset activation words, and dynamically adjust the voice acquisition time according to the number of passengers and the length of the voice clips. Only translate and play the voice sequence when necessary, ensuring the adaptability of voice interaction and Privacy protection.

Benefits of technology

Adaptive voice sequence capture and translation based on passenger conversation status is realized, avoiding resource waste and data leakage, ensuring normal communication is not disturbed, and adapting to the conversation habits of different users.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118918890A_ABST
    Figure CN118918890A_ABST
Patent Text Reader

Abstract

The invention provides a translation method and system based on sound field detection, and belongs to the technical field of intelligent automobiles and human-computer interaction. The translation method is applied to a vehicle-mounted cabin system, and comprises the following steps: monitoring a first voice sequence generated at the current moment; when the first voice sequence comes from the sound field range of the rear seats, judging whether the first voice sequence contains a preset activation word or not; if the first voice sequence contains the preset activation word, translating the first voice sequence; and if the first voice sequence does not contain the preset activation word, ignoring the first voice sequence. The translation system is applied to a vehicle-mounted cabin environment, the vehicle-mounted cabin environment comprises front-row seats and rear-row seats, the front-row seats correspond to a first sound field range, and the rear-row seats correspond to a second sound field range. According to the technical scheme, the voice sound field source can be intelligently recognized, the current voice interaction intention can be distinguished, the normal conversation environment is prevented from being interfered, and meanwhile self-adaption staged learning can be achieved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent automobile and human-computer interaction, and in particular relates to a translation method, system and medium based on sound field detection. Background Art

[0002] The intelligent voice design in the vehicle environment has supported a new interaction method of "using words instead of hands". A common method is to wake up the voice system through certain activation words and let it execute voice operation instructions. For example, the user can wake up the voice system by issuing a voice command "Hello, Xiao X" (generally, the voice system replies in real time with "Hello", "Please speak", etc. to indicate that it has been awakened). Among them, "Hello, Xiao X" is called the "activation word". The activation word can be preset by the system (for example, "Hello, Xiao X") or customized by the user (for example, "Hello, Xiao X", "Hello Xiao Y", "Hello, Xiao Y").

[0003] In the above-mentioned prior art, after the vehicle system captures the activation word, it continues to capture the subsequent voice sequence within a fixed time period (e.g., 20 seconds) each time. However, in the actual vehicle environment, the number of users on board may be different each time, and different users have different speaking habits. The fixed time period does not conform to the actual situation. At the same time, the above-mentioned prior art only supports the activation of vehicle operation instructions, and does not support other communication needs of users, such as real-time communication between foreign passengers and drivers.

[0004] In addition, smart cars are usually equipped with multiple sound field sensors to collect and analyze voice sequences in different sound field ranges in real time, which takes up a long time of vehicle computer resources and wastes traffic resources. In the actual vehicle environment, the voice interaction process does not need to be started at any time, especially the user's normal conversation process may be frequently interrupted or interfered by the vehicle computer voice system, which also causes passengers to worry about their conversation records being illegally captured, used or even leaked. Summary of the invention

[0005] In order to solve the above technical problems, the present invention proposes a translation method, system and medium based on sound field detection.

[0006] In a first aspect of the present invention, it is proposed that the translation method is applied to a vehicle cabin system, the vehicle cabin system comprising at least two rows of seat units, and the method comprises:

[0007] Monitoring a first speech sequence generated at a current moment;

[0008] When the first voice sequence comes from the sound field range of the rear seat, determining whether the first voice sequence contains a preset activation word;

[0009] If the first speech sequence includes a preset activation word, translating the first speech sequence;

[0010] If the first speech sequence does not contain a preset activation word, the first speech sequence is ignored.

[0011] Furthermore, the method further comprises:

[0012] Detects the activation status of the rear seats;

[0013] When two rear seats are activated, detecting a second speech sequence from a sound field range of the rear seats;

[0014] when the second voice sequence comes from two passengers, ignoring the second voice sequence;

[0015] When the second speech sequence comes from only one passenger, the second speech sequence is translated.

[0016] The method further comprises:

[0017] If the first voice sequence is in a second language and contains a preset activation word, the first voice sequence is translated into a second voice sequence in the first language, and the second voice sequence is played in the sound field of the front seat;

[0018] The first language is the language used by the driver.

[0019] Further, when it is detected that the voice sequence generated at the current moment is from the sound field range of the rear seats and contains a preset activation word, the voice from the sound field range of the rear seats is collected within a first preset time length as the first voice sequence;

[0020] When two rear seats are activated, detecting a second speech sequence from a sound field range of the rear seats;

[0021] When the second voice sequence comes from two passengers, the first preset duration is adjusted.

[0022] Preferably, when the second voice sequence comes from two passengers, adjusting the first preset duration specifically includes:

[0023] Obtaining the length of the speech segment contained in the second speech sequence;

[0024] Based on the length of the voice segment, the first preset duration is adjusted.

[0025] In a second aspect of the present invention, a translation system based on sound field detection is provided, wherein the translation system is applied to a vehicle cabin environment, wherein the vehicle cabin environment includes a front row seat and a rear row seat, wherein the front row seat corresponds to a first sound field range, and the rear row seat corresponds to a second sound field range;

[0026] The translation system further comprises:

[0027] A sound field detection unit, used to detect the sound field range generated by the current speech sequence;

[0028] A seat detection unit, used to detect the activation status of the seat;

[0029] A sound source detection unit, configured to detect a type of sound source included in the current speech sequence when the seat detection unit detects that the two rear seats are activated and the sound field detection unit detects that the current speech sequence is located within the sound field range of the rear seats;

[0030] The translation unit is used to translate the current speech sequence when the sound source detection unit detects that the sound source type contained in the current speech sequence is a single sound source.

[0031] The translation unit is further configured to ignore the current speech sequence when the sound source detection unit detects that the sound source type contained in the current speech sequence is not a single sound source.

[0032] Furthermore, the translation system further includes:

[0033] An activation word detection unit, used to detect whether the current speech sequence contains a preset activation word;

[0034] When the activation word detection unit detects that the current speech sequence contains a preset activation word, the translation unit translates the current speech sequence.

[0035] Furthermore, the translation system further includes:

[0036] If the current voice sequence is in the second language and contains a preset activation word, the current voice sequence is translated into a second voice sequence in the first language, and the second voice sequence is played in the sound field of the front seat;

[0037] The first language is the language used by the driver.

[0038] Furthermore, the translation system further includes:

[0039] When a preset activation word is detected in the sound field of the rear seats, speech from the sound field of the rear seats is collected within a first preset time period as the current speech sequence;

[0040] When two rear seats are activated, detecting a second speech sequence from a sound field range of the rear seats;

[0041] When the second voice sequence comes from two passengers, the first preset duration is adjusted.

[0042] The translation method based on sound field detection described in the first aspect can be automatically implemented through various forms of electronic devices through computer program instructions; the computer program instructions can be stored in storage media of different forms and loaded into computer electronic devices for execution.

[0043] Therefore, in the third aspect of the present invention, a computer-readable storage medium is further provided for storing computer instructions. When the computer instructions are executed on an electronic device, the electronic device executes a translation method based on sound field detection as described in the first aspect.

[0044] When the translation method of the present invention is applied to the vehicle-mounted cockpit system, the voice sequence can be accurately and adaptively captured and translated according to the conversation status of the rear passengers, without affecting the normal communication of multiple passengers. At the same time, the captured voice sequence is only temporarily stored during this ride, which will not lead to data leakage and abuse; in addition, the voice collection time can also be adjusted based on actual conditions to meet the conversation habits of different people as much as possible. Therefore, the technical solution of the present invention can intelligently identify the source of the voice sound field and distinguish the current voice interaction intention, avoid interfering with the normal conversation environment, and can adaptively learn in stages.

[0045] Further advantages of the present invention will be further reflected in detail in the specific embodiments section in conjunction with the drawings of the specification. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0047] Figure 1 This is a main flow chart of a translation method based on sound field detection according to an embodiment of the present invention.

[0048] Figure 2 This is a main flow chart of a translation method based on sound field detection according to another embodiment of the present invention.

[0049] Figure 3 The schematic diagram of the principle of adaptively adjusting the preset voice collection time by the method shown in the figure

[0050] Figure 4 This is a schematic diagram of the functional units of a translation system based on sound field detection according to an embodiment of the present invention.

[0051] Figure 5 Schematic diagram of several typical passenger seating situations involved in the technical solution of the present invention

[0052] Figure 6 Schematic diagram of electronic equipment and medium for implementing the translation method based on sound field detection of the present invention DETAILED DESCRIPTION

[0053] The invention is further described below in conjunction with the accompanying drawings and specific implementation methods.

[0054] The technical solution of this application obviously involves the collection of personal data, especially personal voice data, in specific circumstances. However, all user data collected by this application is collected with the consent and authorization of the user, does not involve personal identification, and the collection, use and processing of relevant user data comply with the relevant laws, regulations and standards of relevant countries and regions.

[0055] Specifically, on the one hand, the technical solution of this application is applied to vehicle cockpits, especially to vehicle cockpits of commercial vehicles. According to relevant laws, regulations and standards, the vehicle cockpit at this time is a public place in a certain sense. At this time, in order to maintain public safety, corresponding data collection can be carried out in compliance with relevant national regulations and with prominent prompt signs. At the same time, the voice data collected by the technical solution of this application does not involve personal identity recognition;

[0056] On the other hand, when the in-vehicle intelligent system is turned on for the first time, before collecting user-related data and during the process of collecting user-related data, a prompt interface, pop-up window or voice prompt information can be displayed. The prompt interface, pop-up window or voice prompt information is used to prompt the user that their relevant data is currently being collected, so that the present application only starts to execute the relevant steps of obtaining user-related data after obtaining the user's confirmation operation on the prompt interface or pop-up window. Otherwise (that is, when the user's confirmation operation on the prompt interface or pop-up window is not obtained), the relevant steps of obtaining user-related data are terminated, that is, the user's relevant data is not obtained.

[0057] On this basis, various specific embodiments of the present invention are introduced as follows:

[0058] See also Figure 1 , Figure 1 The main flow chart of a translation method based on sound field detection in one embodiment.

[0059] Specifically, Figure 1 A translation method based on sound field detection is shown, the translation method is applied to a vehicle cabin system, the vehicle cabin system includes at least two rows of seat units, and the method includes the following steps:

[0060] Monitoring a first speech sequence generated at a current moment;

[0061] When the first voice sequence comes from the sound field range of the rear seat, determining whether the first voice sequence contains a preset activation word;

[0062] If the first speech sequence includes a preset activation word, translating the first speech sequence;

[0063] If the first speech sequence does not contain a preset activation word, the first speech sequence is ignored.

[0064] Each embodiment of the present invention can handle various different situations, and different situations involve different combinations of preset activation words and voice sequence sound field sources, which are introduced one by one below.

[0065] First, the construction of preset activation words is introduced.

[0066] In one embodiment, a preset activation word library is constructed with multiple possible foreign-related languages ​​(hereinafter collectively referred to as the first language), and the preset activation word library includes multiple first preset activation words. It can be understood that foreign-related is only a relative concept. In English-speaking countries, Chinese is a foreign-related language; and in Mandarin Chinese countries, English is a foreign-related language. The application scenario of the present invention is suitable for situations where the local driver does not master or is not proficient in foreign-related languages, and the seated passengers, especially the passengers in the back seat, are from foreign-related language countries.

[0067] Therefore, the first preset activation word is used to represent the expectation of the non-driving seat passenger to communicate with the driver in the driving seat;

[0068] As an example, the first preset activation word may be:

[0069] “Hello, driver”, “Hello, master” (examples where the first language is Chinese);

[0070] “Hello, driver”, “Hello, master” (examples where the first language is English);

[0071] In actual applications, the above-mentioned first preset activation word and its functional usage can be prompted in a prominent manner in a location visible / audible to the non-driving passenger, for example, by posting a prompt in a location visible to the non-driving passenger, by a pop-up prompt on the screen visible to the non-driving passenger, by voice broadcast through a microphone within the sound field range of the non-driving passenger, etc.

[0072] For example, when the driver detects that a passenger is sitting in the back row (of course, the passenger can also be sitting in the front row), if the driver determines that the passenger may be a foreigner, the driver will start to broadcast the following prompts:

[0073] If you are a foreigner and want to talk to the driver, please say "Hello, driver", Translation will be performed automatically.

[0074] Figure 1 The embodiment focuses on taking the passengers sitting in the rear row as an example, and the method can be further implemented as follows:

[0075] S1: When monitoring the real-time speech generated by the seated rear row passenger and containing the first preset activation word, continue to capture subsequent speech sequences within a first preset time length (e.g., an initialization value, such as 15 seconds, 20 seconds, etc.), and use the subsequent speech sequences captured within the first preset time length (e.g., 15 seconds, 20 seconds, etc.) as the first speech sequence;

[0076] S2: when the first speech sequence comes from the sound field range of the rear seats, translating the first speech sequence;

[0077] At this time, in the step S1, when the real-time voice generated by the passenger in the non-driving seat does not contain the first preset activation word, the voice sequence capture and translation process is not started, that is, only monitoring is performed without storage.

[0078] Of course, the above method is not limited to passengers in the back seats. Even if the passengers sit in the front seats (next to the driver, in the front passenger seat), the method can still be implemented as follows:

[0079] When the real-time speech generated by the passenger sitting in the front row non-driver's seat contains a first preset activation word, continue to capture subsequent speech sequences within a first preset time length (for example, an initialization value, 15 seconds, 20 seconds, etc.), and use the subsequent speech sequences captured within the first preset time length (for example, 15 seconds, 20 seconds, etc.) as the first speech sequence; and translate the first speech sequence in real time.

[0080] Of course, in most cases, there are multiple rear passengers in the vehicle at the same time. After getting on the vehicle, the multiple rear passengers have regular personal conversations. At this time, irrelevant interference should not be made to them to avoid voice interruption. Therefore, see Figure 2 The embodiments can solve this problem.

[0081] Specifically, Figure 2 The method and technical solution are as follows:

[0082] A translation method based on sound field detection, the translation method is applied to a vehicle cabin system, the vehicle cabin system includes at least two rows of seat units, and the method includes:

[0083] Detects the activation status of the rear seats;

[0084] When two rear seats are activated, detecting a second speech sequence from a sound field range of the rear seats;

[0085] When the second voice sequence comes from at least two passengers, ignoring the second voice sequence;

[0086] When the second speech sequence comes from only one passenger, the second speech sequence is translated.

[0087] Figure 2 In the improved embodiment, translation intention recognition can be realized without activation words, that is, as long as passengers get on the bus, if there are multiple people sitting in the back row and only one person speaks, it is determined that there is a passenger-driver communication intention, and the car computer performs real-time translation.

[0088] Of course, due to the existence of the prompt, the passenger is more likely to start the translation process based on the activation word after sitting down. In any case, the communication intention between a passenger and the driver can be identified as much as possible.

[0089] However, if it is detected that this is normal communication between multiple passengers, there is no need to perform voice analysis, that is, when the second voice sequence comes from at least two passengers, the second voice sequence is ignored. Ignoring means that the voice monitoring process will not be executed for a period of time afterwards to avoid disturbing passengers.

[0090] A further embodiment of the present invention also fully considers the personal privacy issue of voice data, that is, the data protection issue.

[0091] To this end, the above method further comprises:

[0092] Monitoring a first speech sequence from a passenger in a non-driving seat and temporarily storing the first speech sequence in a first memory;

[0093] When the first voice sequence does not include a first preset activation word, and / or when it is detected that the passenger leaves the non-driving seat, the first voice sequence is deleted from the first memory.

[0094] Therefore, although the technical solution of the present invention collects data, it does not involve personal user identity identification. At the same time, the collected data is only used for possible voice translation and interaction processes, and other irrelevant data is not collected. At the same time, the data usage period is limited to the period when the passenger is in the vehicle, and the data is automatically deleted after the passenger leaves, effectively avoiding data leakage and excessive storage.

[0095] Preferably, if the first voice sequence is in a second language and contains a preset activation word, the first voice sequence is translated into a second voice sequence in the first language, and the second voice sequence is played in the sound field of the front seats;

[0096] The first language is the language used by the driver.

[0097] In the prior art, after capturing the activation word, the vehicle computer system continues to capture subsequent voice sequences within a fixed time (e.g., 20 seconds) and determines whether the voice sequence contains the corresponding vehicle computer operation. If so, the vehicle computer operation is executed; otherwise, it is not executed and waits for the next activation.

[0098] However, in the actual operating vehicle environment, the voice conversation habits of seated passengers vary, some are good at talking, and some are brief. If the subsequent voice sequence is captured within a fixed time (for example, 20 seconds), it may waste vehicle computer traffic and resources, and may also mistakenly capture irrelevant voice sequences.

[0099] Therefore, as a further preference, the above method is further improved as follows:

[0100] When monitoring the real-time speech generated by the passenger in the non-driving seat to include the first preset activation word, continue to capture the subsequent speech sequence within a first preset time length (for example, an initialization value, such as 15 seconds, 20 seconds, etc.);

[0101] Adjusting the first preset duration according to the speech state at the end time of the first speech sequence specifically includes:

[0102] When the first speech sequence ends without speech, the first preset duration is reduced; otherwise, the first preset duration is increased.

[0103] Of course, when it is detected that the passenger leaves the non-driving seat, the first preset time length is initialized again.

[0104] In this way, dynamic voice sequence capture duration adjustment is achieved, which can better suit the voice habits of each passenger with different habits.

[0105] The above adjustment method takes into account individual habits. However, in a multi-occupant environment, group conversation habits also need to be considered. Figure 3 , Figure 3 The figure is a schematic diagram of the principle of adaptively adjusting the preset voice collection time by the method shown in the figure.

[0106] When it is detected that the voice sequence generated at the current moment is from the sound field range of the rear seats and contains a preset activation word, the voice from the sound field range of the rear seats is collected within a first preset time length as the first voice sequence;

[0107] When two rear seats are activated, detecting a second speech sequence from a sound field range of the rear seats;

[0108] When the second voice sequence comes from at least two passengers, the first preset duration is adjusted.

[0109] Preferably, when the second voice sequence comes from two passengers, adjusting the first preset duration specifically includes:

[0110] Obtaining the length of the speech segment contained in the second speech sequence;

[0111] Based on the length of the voice segment, the first preset duration is adjusted.

[0112] As a pseudo code embodiment in computer implementation form, the above process can be implemented as follows:

[0113] Assume that the initialization value of the first preset duration is T0;

[0114] When the second voice sequence comes from at least two passengers, the process of adjusting the first preset duration T0 is as follows:

[0115] The second speech sequence is expressed as S={V1, V2, V3, ... Vn}; wherein Vi represents the ith speech segment; |Vi| represents its length (unit: second).

[0116] Calculate the average speech length LPi of each passenger Pi in the second speech sequence:

[0117]

[0118] Among them, i belongs to Pi means that the above formula only selects the voice sequence Vi of passenger Pi, and NumVi represents the total number of voice segments uttered by passenger Pi.

[0119] Then the first preset time length T0 is adjusted as follows:

[0120]

[0121] Where max{LPi} represents the maximum value of all LPi (because there are at least two passengers Pi);

[0122]

[0123] Indicates rounding up.

[0124] To better understand the above adjustment process, let's take the case of two rear passengers (P1, P2) as an example and perform the following calculations:

[0125] Assume that within the initial first preset duration T0 (assuming 15 seconds), the voice conversation segment detected in the back row is:

[0126] P1→V1 (indicates that the starting segment V1 is generated by P1, assuming the duration is 5 seconds);

[0127] P2→V2, V3; (indicates that the next segments V2 and V3 are generated by P2, with durations of 3 seconds and 4 seconds respectively);

[0128] P1→V4, V5, V6 (indicates that the conversation ends after P1 generates three segments V4, V5, and V6, with durations of 4 seconds, 5 seconds, and 9 seconds respectively);

[0129] Therefore, at this time S = {V1, V2, V3, V4, V5, V6}.

[0130] The average speech lengths LP1 and LP2 of passengers P1 and P2 are calculated as follows:

[0131]

[0132] Max{LPi}=5.75;

[0133] Max{LPi} rounded up is 5.75 rounded up, then T0=6.

[0134] Therefore, in the next monitoring, the speech from the rear seat sound field range is collected within the adjusted first preset time length (6 seconds) as the first speech sequence.

[0135] The above adjustment process is to monitor only a single voice sequence of a single person as much as possible, which is in line with the actual scenario of possible interaction instructions.

[0136] Based on the above embodiments, the above processes are combined into the following more preferred embodiments:

[0137] Preferred embodiment 1:

[0138] A translation method based on sound field detection, the translation method is applied to a vehicle cabin system, the vehicle cabin system includes at least two rows of seat units, characterized in that the method comprises:

[0139] S0: Initialize the first preset duration to an initialization value;

[0140] S1: when it is detected that the voice sequence generated at the current moment is from the sound field range of the rear seats and contains a preset activation word, collecting the voice from the sound field range of the rear seats within a first preset time length as the first voice sequence;

[0141] S3: translating the first speech sequence;

[0142] S4: when the first speech sequence ends without speech, reducing the first preset duration; otherwise, increasing the first preset duration, and returning to step S0;

[0143] Preferred embodiment 2:

[0144] A translation method based on sound field detection, the translation method is applied to a vehicle cabin system, the vehicle cabin system includes at least two rows of seat units, characterized in that the method comprises:

[0145] A1: Detect the activation status of the rear seats;

[0146] A2: When two rear seats are activated, a second voice sequence from the sound field range of the rear seats is detected within a first preset time period;

[0147] A3: When the second voice sequence comes from two passengers, ignore the second voice sequence and proceed to step A5;

[0148] A4: When the second voice sequence comes from only one passenger, translating the second voice sequence;

[0149] A5: Adjust the first preset duration;

[0150] Wherein, step A5 comprises:

[0151] The second speech sequence is expressed as S={V1, V2, V3, ... Vn}; wherein Vi represents the ith speech segment; |Vi| represents its length (unit: second).

[0152] Calculate the average speech length LPi of each passenger Pi in the second speech sequence:

[0153]

[0154] Among them, i belongs to Pi means that the above formula only selects the voice sequence Vi of passenger Pi, and NumVi represents the total number of voice segments uttered by passenger Pi.

[0155] Then the first preset time length T0 is adjusted as follows:

[0156]

[0157] Where max{LPi} represents the maximum value of all LPi (because there are at least two passengers Pi);

[0158]

[0159] Indicates rounding up.

[0160] Both of the above embodiments further include:

[0161] When it is monitored that the non-driver seat is not activated, the first memory of the sound field sensor configuration corresponding to the non-driver seat that is not activated is formatted, and the first preset value is initialized.

[0162] Based on the above method embodiments, the following briefly introduces the corresponding product embodiments, that is, the system and scenario environment for implementing the method embodiments of the present invention. Figure 4-Figure 5 .

[0163] Figure 4 A translation system based on sound field detection is shown, and the translation system is applied to a vehicle cabin environment. The vehicle cabin environment includes front seats and rear seats. The front seats correspond to a first sound field range, and the rear seats correspond to a second sound field range.

[0164] The translation system further comprises:

[0165] A sound field detection unit, used to detect the sound field range generated by the current speech sequence;

[0166] A seat detection unit, used to detect the activation status of the seat;

[0167] A sound source detection unit, configured to detect a type of sound source included in the current speech sequence when the seat detection unit detects that the two rear seats are activated and the sound field detection unit detects that the current speech sequence is located within the sound field range of the rear seats;

[0168] The translation unit is used to translate the current speech sequence when the sound source detection unit detects that the sound source type contained in the current speech sequence is a single sound source.

[0169] The translation unit is further configured to ignore the current speech sequence when the sound source detection unit detects that the sound source type contained in the current speech sequence is not a single sound source.

[0170] The translation system further comprises:

[0171] An activation word detection unit, used to detect whether the current speech sequence contains a preset activation word;

[0172] When the activation word detection unit detects that the current speech sequence contains a preset activation word, the translation unit translates the current speech sequence.

[0173] If the current voice sequence is in the second language and contains a preset activation word, the current voice sequence is translated into a second voice sequence in the first language, and the second voice sequence is played in the sound field of the front seat;

[0174] The first language is the language used by the driver.

[0175] When a preset activation word is detected in the sound field of the rear seats, speech from the sound field of the rear seats is collected within a first preset time period as the current speech sequence;

[0176] When two rear seats are activated, detecting a second speech sequence from a sound field range of the rear seats;

[0177] When the second voice sequence comes from two passengers, the duration adjustment unit adjusts the first preset duration.

[0178] Further, it can be understood that the system embodiment corresponds to the method embodiment, and the steps and principles corresponding to the aforementioned method embodiment can be combined in the system embodiment, which will not be repeated here.

[0179] Figure 5 The following scenarios schematically describe several typical application scenarios and principles of the present invention. The driver's seat is marked as 0, the co-driver's seat is marked as 1, and the three rear seats are marked as 2-3-4. The driver's seat is activated by default (the activation state is not indicated in the figure).

[0180] exist Figure 5 Figure (1) of the diagram describes the state where only the front passenger seat is occupied (activated). At this time, the passenger is sitting in the front passenger seat and there is no one in the back seat;

[0181] If the driver needs to determine that the passenger may be a foreigner, after the passenger sits down, an activation word prompt will be played in the front passenger seat. For example, the following prompt will be played:

[0182] If you are a foreigner and want to talk to the driver, please say "Hello, driver", Translation will be performed automatically.

[0183] At this time, when it is detected that the speech sequence generated at the current moment contains the preset activation word, the speech collected within the first preset time length is taken as the first speech sequence and directly translated;

[0184] exist Figure 5 Figure (2)(3)) describes the state of two people sitting in the back row (activated).

[0185] At this time, after the passengers take their seats, the above activation word prompts are played in the rear sound field.

[0186] Next, it depends on the conversation between the two people.

[0187] When it is detected that the speech sequence generated at the current moment contains a preset activation word, the speech collected within the first preset time length is used as the first speech sequence and directly translated;

[0188] Otherwise, ignoring the first speech sequence;

[0189] and, detecting the activation status of the rear seats;

[0190] When two rear seats are activated, detecting a second speech sequence from a sound field range of the rear seats;

[0191] when the second voice sequence comes from two passengers, ignoring the second voice sequence;

[0192] When the second speech sequence comes from only one passenger, the second speech sequence is translated.

[0193] Preferably, if the first voice sequence is in a second language and contains a preset activation word, the first voice sequence is translated into a second voice sequence in the first language, and the second voice sequence is played in the sound field of the front seats;

[0194] The first language is the language used by the driver.

[0195] When it is detected that the speech sequence generated at the current moment is from the sound field range of the rear seats and contains a preset activation word, the speech from the sound field range of the rear seats is collected within a first preset time length as the first speech sequence;

[0196] When two rear seats are activated, detecting a second speech sequence from a sound field range of the rear seats;

[0197] When the second voice sequence comes from two passengers, the first preset duration is adjusted.

[0198] The specific adjustment process and principle are described in the above-mentioned embodiment and will not be repeated here.

[0199] The translation method based on sound field detection mentioned in the above embodiment can be automatically implemented through various forms of electronic devices through computer program instructions; the computer program instructions can be stored in different forms of storage media and loaded into computer electronic devices for execution.

[0200] Figure 6 The embodiment of the present invention adaptively shows a schematic diagram of the relationship between an electronic device and a computer-readable storage medium. Figure 6The electronic device shown may include: a processor 1002, a memory 1001, a communication component 1003, an audio component 1006, a power component 1005 and a display component 1004. The processor 1002, the memory 1001, the communication component 1003, the audio component 1006, the power component 1005 and the display component 1004 are connected to a computer-readable storage medium via a universal serial bus (USB) interface 130, and computer program instructions are read and loaded into the computer electronic device for execution.

[0201] When the translation method of the present invention is applied to the vehicle-mounted cockpit system, the voice sequence can be accurately and adaptively captured and translated according to the conversation status of the rear passengers, without affecting the normal communication of multiple passengers. At the same time, the captured voice sequence is only temporarily stored during this ride, which will not lead to data leakage and abuse; in addition, the voice collection time can also be adjusted based on actual conditions to meet the conversation habits of different people as much as possible. Therefore, the technical solution of the present invention can intelligently identify the source of the voice sound field and distinguish the current voice interaction intention, avoid interfering with the normal conversation environment, and can adaptively learn in stages.

[0202] The embodiments of the present invention can also be divided into embodiments in the form of method flow, embodiments in the form of system functional architecture, and embodiments in the form of computer program code pseudo language. It can be understood that, if there is no special explanation, the same or corresponding steps between the embodiments in different forms correspond to each other. For example, in the embodiment in the form of system functional architecture, "XX unit, used to implement XX function" actually corresponds to "XX step, used to implement XX function" in the embodiment in the form of method flow.

[0203] Therefore, after the embodiment in the form of method flow is elaborated in detail, the corresponding embodiment in the form of system function architecture does not need to be repeated again, but those skilled in the art can directly and unambiguously determine that the embodiment in the form of system function architecture also corresponds to all the contents of the embodiment in the form of method flow. The same is true for the embodiment in the form of computer program code pseudo language.

Claims

1. A translation method based on sound field detection, the translation method is applied to a vehicle cabin system, the vehicle cabin system includes at least two rows of seat units, characterized in that: The method comprises: Monitoring a first speech sequence generated at a current moment; When the first voice sequence comes from the sound field range of the rear seat, determining whether the first voice sequence contains a preset activation word; If the first speech sequence includes a preset activation word, translating the first speech sequence; If the first speech sequence does not contain a preset activation word, the first speech sequence is ignored.

2. A translation method based on sound field detection as claimed in claim 1, characterized in that: The method further comprises: Detects the activation status of the rear seats; When two rear seats are activated, detecting a second speech sequence from a sound field range of the rear seats; when the second voice sequence comes from two passengers, ignoring the second voice sequence; When the second speech sequence comes from only one passenger, the second speech sequence is translated.

3. The translation method based on sound field detection as claimed in claim 1, characterized in that: The method further comprises: If the first voice sequence is in a second language and contains a preset activation word, the first voice sequence is translated into a second voice sequence in the first language, and the second voice sequence is played in the sound field of the front seat; The first language is the language used by the driver.

4. The translation method based on sound field detection as claimed in claim 2, characterized in that: The method further comprises: When it is detected that the speech sequence generated at the current moment is from the sound field range of the rear seats and contains a preset activation word, the speech from the sound field range of the rear seats is collected within a first preset time length as the first speech sequence; When two rear seats are activated, detecting a second speech sequence from a sound field range of the rear seats; When the second voice sequence comes from two passengers, the first preset duration is adjusted.

5. A translation method based on sound field detection as claimed in claim 4, characterized in that: When the second voice sequence comes from two passengers, adjusting the first preset duration specifically includes: Obtaining the length of the speech segment contained in the second speech sequence; Based on the length of the voice segment, the first preset duration is adjusted.

6. A translation system based on sound field detection, the translation system is applied to a vehicle cabin environment, the vehicle cabin environment includes a front row seat and a rear row seat, the front row seat corresponds to a first sound field range, the rear row seat corresponds to a second sound field range, characterized in that: The translation system further comprises: a sound field detection unit, which is used to detect the sound field range generated by the current speech sequence; A seat detection unit, used to detect the activation status of the seat; A sound source detection unit, configured to detect a type of sound source included in the current speech sequence when the seat detection unit detects that the two rear seats are activated and the sound field detection unit detects that the current speech sequence is located within the sound field range of the rear seats; The translation unit is used to translate the current speech sequence when the sound source detection unit detects that the sound source type contained in the current speech sequence is a single sound source.

7. A translation system based on sound field detection as claimed in claim 6, characterized in that: The translation unit is further configured to ignore the current speech sequence when the sound source detection unit detects that the sound source type contained in the current speech sequence is not a single sound source.

8. A translation system based on sound field detection as claimed in claim 6, characterized in that: The translation system further comprises: An activation word detection unit, used to detect whether the current speech sequence contains a preset activation word; When the activation word detection unit detects that the current speech sequence contains a preset activation word, the translation unit translates the current speech sequence.

9. A translation system based on sound field detection as claimed in claim 6, characterized in that: The translation system further comprises: If the current voice sequence is in the second language and contains a preset activation word, the current voice sequence is translated into a second voice sequence in the first language, and the second voice sequence is played in the sound field of the front seat; The first language is the language used by the driver.

10. A translation system based on sound field detection as claimed in claim 6, characterized in that: The translation system further comprises: When a preset activation word is detected in the sound field of the rear seats, speech from the sound field of the rear seats is collected within a first preset time period as the current speech sequence; When two rear seats are activated, detecting a second speech sequence from a sound field range of the rear seats; When the second voice sequence comes from two passengers, the first preset duration is adjusted.