Significance word detection method and device, computer device, readable storage medium and program product

Abstract

The application relates to a meaningful word detection method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: receiving a voice signal output by a user, and converting the voice signal into a detection text; performing semantic purification processing on the detection text to generate a pure semantic text; performing total amount threshold determination on the pure semantic text, and determining that the voice signal is a semantic effective signal in the case of successful total amount threshold determination; performing an incremental rate determination on the pure semantic text in the case of failed total amount threshold determination, and determining that the voice signal is a semantic effective signal in the case of successful incremental rate determination; and performing key word instant determination on the pure semantic text in the case of failed incremental rate determination, and determining that the voice signal is a semantic effective signal in the case of successful key word instant determination. The method can improve the accuracy of meaningful word detection.

Description

Technical Field

[0001] This application relates to the field of natural language processing technology, and in particular to a method, apparatus, computer device, computer-readable storage medium, and computer program product for detecting meaningful words. Background Technology

[0002] With the widespread application of Automatic Speech Recognition (ASR) systems in intelligent customer service scenarios, it is necessary to accurately determine when a user has completed a valid speech input in order to respond in a timely manner. The accuracy and real-time nature of the judgment process directly affect the user experience and system efficiency.

[0003] In related technologies, the energy threshold method is used to judge speech activity based on the energy level of the speech signal. However, it is not sensitive to speech content (such as coughing, throat clearing and other high-energy meaningless sounds are easily misjudged as valid speech), the threshold is difficult to set (the energy levels of different users and environments vary greatly and need to be dynamically adjusted), and it cannot understand semantics. It only relies on signal features and ignores the meaning of speech content, resulting in inaccurate final detection results. Summary of the Invention

[0004] Therefore, it is necessary to provide a semantic word detection method, apparatus, computer equipment, computer-readable storage medium, and computer program product that can improve the detection accuracy in response to the above-mentioned technical problems.

[0005] Firstly, this application provides a semantic word detection method, including:

[0006] Receives the voice signal output by the user and converts the voice signal into detected text;

[0007] The detected text is semantically cleaned to generate clean semantic text;

[0008] A total threshold is determined for the pure semantic text. If the total threshold determination is successful, the speech signal is determined to be a semantically valid signal.

[0009] If the total threshold determination fails, the incremental rate of the pure semantic text is determined. If the incremental rate determination is successful, the speech signal is determined to be a semantically valid signal.

[0010] If the incremental rate determination fails, the pure semantic text is subjected to real-time keyword determination. If the real-time keyword determination is successful, the speech signal is determined to be a semantically valid signal.

[0011] In one embodiment, the step of performing semantic cleansing on the detected text to generate clean semantic text includes:

[0012] The detected text is segmented into words to obtain the processed detected text;

[0013] The processed detection text is matched with filler words in the preset filler word database. All successfully matched filler words in the processed detection text are removed to generate clean semantic text.

[0014] In one embodiment, the determination of the total amount threshold for the clean semantic text includes:

[0015] Based on the clean semantic text, determine the cumulative number of effective words in the current dialogue round;

[0016] If the cumulative number of valid words is greater than or equal to the preset word count threshold, the total threshold determination is successful.

[0017] If the cumulative number of valid words is less than the preset word count threshold, the total threshold determination fails.

[0018] In one embodiment, the incremental rate determination of the clean semantic text includes:

[0019] Calculate the vocabulary increment between the cumulative effective vocabulary in the current dialogue round and the historical effective vocabulary;

[0020] If the word increment is greater than or equal to a preset increment threshold, the increment rate determination is successful.

[0021] If the increment of the vocabulary is less than the preset increment threshold, the increment rate determination fails.

[0022] In one embodiment, the real-time keyword determination of the clean semantic text includes:

[0023] The pure semantic text is matched with a preset keyword database;

[0024] If a matching keyword exists, the keyword is immediately determined to be successful.

[0025] If all keywords in the preset keyword database have been matched and no keyword has been successfully matched, the keyword is immediately judged as unsuccessful.

[0026] In one embodiment, the method further includes:

[0027] If the decibel level of the speech signal is less than the first decibel threshold, increase the preset word count threshold and the preset increment threshold.

[0028] If the decibel level of the speech signal exceeds the second decibel threshold, the preset word count threshold and the preset increment threshold are reduced.

[0029] Secondly, this application also provides a semantic word detection device, comprising:

[0030] The conversion module is used to receive the voice signal output by the user and convert the voice signal into detected text;

[0031] The generation module is used to perform semantic cleansing on the detected text to generate clean semantic text;

[0032] The determination module is used to determine the total amount threshold of the pure semantic text, and if the total amount threshold determination is successful, determine that the speech signal is a semantically valid signal;

[0033] The determination module is also used to perform incremental rate determination on the pure semantic text when the total threshold determination fails, and to determine the speech signal as a semantically valid signal when the incremental rate determination succeeds.

[0034] The determination module is further configured to perform real-time keyword determination on the clean semantic text in the event of failure of incremental rate determination, and to determine the speech signal as a semantically valid signal in the event of success of real-time keyword determination.

[0035] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0036] Receives the voice signal output by the user and converts the voice signal into detected text;

[0037] The detected text is semantically cleaned to generate clean semantic text;

[0038] A total threshold is determined for the pure semantic text. If the total threshold determination is successful, the speech signal is determined to be a semantically valid signal.

[0039] If the total threshold determination fails, the incremental rate of the pure semantic text is determined. If the incremental rate determination is successful, the speech signal is determined to be a semantically valid signal.

[0040] If the incremental rate determination fails, the pure semantic text is subjected to real-time keyword determination. If the real-time keyword determination is successful, the speech signal is determined to be a semantically valid signal.

[0041] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0042] Receives the voice signal output by the user and converts the voice signal into detected text;

[0043] The detected text is semantically cleaned to generate clean semantic text;

[0044] A total threshold is determined for the pure semantic text. If the total threshold determination is successful, the speech signal is determined to be a semantically valid signal.

[0045] If the total threshold determination fails, the incremental rate of the pure semantic text is determined. If the incremental rate determination is successful, the speech signal is determined to be a semantically valid signal.

[0046] If the incremental rate determination fails, the pure semantic text is subjected to real-time keyword determination. If the real-time keyword determination is successful, the speech signal is determined to be a semantically valid signal.

[0047] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0048] Receives the voice signal output by the user and converts the voice signal into detected text;

[0049] The detected text is semantically cleaned to generate clean semantic text;

[0050] A total threshold is determined for the pure semantic text. If the total threshold determination is successful, the speech signal is determined to be a semantically valid signal.

[0051] If the total threshold determination fails, the incremental rate of the pure semantic text is determined. If the incremental rate determination is successful, the speech signal is determined to be a semantically valid signal.

[0052] If the incremental rate determination fails, the pure semantic text is subjected to real-time keyword determination. If the real-time keyword determination is successful, the speech signal is determined to be a semantically valid signal.

[0053] The aforementioned semantic word detection method, apparatus, computer equipment, computer-readable storage medium, and computer program product first receive the user-output speech signal and convert it into detection text; then, they perform semantic cleansing on the detection text to generate clean semantic text; finally, they perform a total threshold determination on the clean semantic text, and if the total threshold determination is successful, the speech signal is determined to be a semantically valid signal; if the total threshold determination fails, they perform an incremental rate determination on the clean semantic text, and if the incremental rate determination is successful, the speech signal is determined to be a semantically valid signal; and if the incremental rate determination fails, they perform real-time keyword determination on the clean semantic text, and if the real-time keyword determination is successful, the speech signal is determined to be a semantically valid signal. Thus, through a triple determination mechanism (cumulative word count, incremental rate, and keyword matching) combined with semantic cleansing preprocessing, they can effectively distinguish between valid semantic content and meaningless filler words, noise, and other interference, thereby significantly reducing the false trigger rate, improving the accuracy of speech endpoint detection and interruption, and ensuring the semantic consistency of the system response. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1 This is a diagram illustrating the application environment of a semantic word detection method in one embodiment.

[0056] Figure 2 This is a flowchart illustrating a semantic word detection method in one embodiment;

[0057] Figure 3 This is a structural block diagram of a semantic word detection device in one embodiment;

[0058] Figure 4 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0060] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0061] The semantic word detection method provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or located on the cloud or other network servers. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, drones, low-altitude aircraft, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, projection devices, etc. Portable wearable devices can include smartwatches, smart bracelets, head-mounted devices, etc. Head-mounted devices can be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. Server 104 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0062] In one embodiment, in interactive scenarios, it is not only necessary to accurately convert speech into text, but also to determine in real time when the user has completed a meaningful expression in order to respond in a timely manner or interrupt naturally. This directly affects user experience, interaction efficiency and system performance.

[0063] Most existing methods are based on audio signal features without incorporating semantic information from the speech content, leading to discrepancies between detection and the user's true intent. High-precision models have slow responses, while fast detection methods have low accuracy, making it difficult to balance both. They also struggle to adapt to the different speaking habits (speech speed, pauses) of various users and the diverse needs of different application scenarios (short commands vs. complex narratives). Complex deep learning models require high computational power, limiting their application in mobile terminals, embedded devices, and other scenarios. Furthermore, they are susceptible to interference from filler words (such as "um" and "ah"), environmental noise, and non-semantic speech, resulting in inaccurate system responses.

[0064] Therefore, a semantic word detection technology is needed. This technology should have semantic perception capabilities to make judgments based on the semantics of speech content, high real-time millisecond-level detection response, high computational efficiency to adapt to edge device deployment, strong adaptability to match different user habits and application scenarios, and low false trigger rate, effectively filtering meaningless speech segments and accurately identifying valid semantic content.

[0065] Among them, semantic words are the effective semantic content that the system needs to identify and respond to.

[0066] In one exemplary embodiment, such as Figure 2 As shown, a semantic word detection method is provided, which is then applied to... Figure 1 Taking terminal 102 as an example, the explanation includes the following steps 202 to 210. Wherein:

[0067] Step 202: Receive the voice signal output by the user and convert the voice signal into detection text.

[0068] For example, the application scenario is determined, and a specified instruction for a user-specific configuration template is selected or received according to the application scenario. The specific configuration template is loaded, all internal state variables are initialized, the user-output voice signal is obtained from the ASR (Automatic Speech Recognition) engine, and the obtained voice signal is converted into detection text.

[0069] The speech signal also includes the acquired timestamp and confidence level.

[0070] Step 204: Perform semantic cleansing on the detected text to generate clean semantic text.

[0071] Optionally, the transformed detection text can be semantically cleaned to obtain clean semantic text.

[0072] Step 206: Perform a total threshold determination on the pure semantic text. If the total threshold determination is successful, the speech signal is determined to be a semantically valid signal.

[0073] For example, based on the clean semantic text, the number of effective words accumulated by the user in the current dialogue round is determined, and a total threshold is determined. If the total threshold determination is successful, the speech signal is determined to be a semantically effective signal.

[0074] Step 208: If the total threshold determination fails, perform incremental rate determination on the pure semantic text. If the incremental rate determination succeeds, determine that the speech signal is a semantically valid signal.

[0075] Optionally, if the total amount threshold of the pure semantic text fails to be determined, the incremental rate of the pure semantic text is determined. If the incremental rate determination is successful, the speech signal is determined to be a semantically valid signal.

[0076] Step 210: If the incremental rate determination fails, perform real-time keyword determination on the pure semantic text. If the real-time keyword determination succeeds, determine that the speech signal is a semantically valid signal.

[0077] For example, if the total threshold determination of the pure semantic text fails and the incremental rate determination also fails, the pure semantic text is subject to real-time keyword determination. If the real-time keyword determination is successful, the speech signal is determined to be a semantically valid signal.

[0078] In one embodiment, if the keyword determination of the pure semantic text also fails, the voice signal is determined to be an invalid signal, and the determination of the voice signal continues in the next dialogue round.

[0079] In another embodiment, the final judgment result of the speech signal includes the trigger source (which judgment stage determines the result, for example, the trigger source is the total threshold judgment), the timestamp when the judgment result is triggered, the clean semantic text, and the confidence score of the judgment result.

[0080] The confidence score is calculated from the trigger source weight and the trigger intensity. Each trigger source includes the trigger source weight, and the trigger intensity refers to the intensity value calculated based on the dialogue parameters and preset trigger parameters in the current trigger source scenario. The specific calculation method can be set by the user according to the actual situation, and this application embodiment does not limit it.

[0081] In one embodiment, when the system is configured in long speech mode, a segmented detection strategy is adopted. For example, the speech signal is judged once every preset time interval (such as 5 seconds), that is, the meaning word is detected. The determination of the cumulative number of effective words is valid within one dialogue round.

[0082] In another embodiment, the semantic word detection method in this application supports the detection of different languages ​​and adjusts the threshold according to different language types.

[0083] In the aforementioned semantic word detection method, the user-output speech signal is received and converted into detection text. Semantic cleansing is then performed on the detection text to generate clean semantic text. A total quantity threshold is applied to the clean semantic text; if the threshold is successfully reached, the speech signal is determined to be semantically valid. If the threshold fails, an incremental rate is applied; if the incremental rate is successfully applied, the speech signal is determined to be semantically valid. If the incremental rate fails, keyword instantaneous detection is performed on the clean semantic text; if the keyword instantaneous detection is successful, the speech signal is determined to be semantically valid. Thus, by combining a triple-determination mechanism (cumulative word count, incremental rate, and keyword matching) with semantic cleansing preprocessing, effective semantic content can be effectively distinguished from meaningless filler words, noise, and other interference, thereby significantly reducing the false trigger rate, improving the accuracy of speech endpoint detection and interruption, and ensuring the semantic consistency of the system response.

[0084] In an exemplary embodiment, semantic purification processing is performed on the detected text to generate clean semantic text, including: segmenting the detected text into words to obtain processed detected text; matching the processed detected text with filler words in a preset filler word database; removing all successfully matched filler words from the processed detected text to generate clean semantic text.

[0085] In practice, the detection text is segmented to obtain the processed detection text. The processed detection text is then matched with filler words in the preset filler word database. All successfully matched filler words in the processed detection text are removed, and the number and position information of the removed filler words are recorded to generate clean semantic text.

[0086] The preset filler word database contains common meaningless words in this application scenario (e.g., “um”, “ah”, “this”, “that”, “then”, etc.). The matching method can be to calculate the text similarity between the processed detection text and the filler words in the preset filler word database, or it can be other matching methods. This application embodiment does not limit this.

[0087] In the above embodiments, by removing filler words from the detected text in advance and filtering out meaningless words, the computational workload in the subsequent judgment process is reduced, and the efficiency of meaningful word detection is accelerated.

[0088] In an exemplary embodiment, determining the total threshold for clean semantic text includes: determining the cumulative number of valid words in the current dialogue round based on the clean semantic text; if the cumulative number of valid words is greater than or equal to a preset word count threshold, the total threshold determination is successful; if the cumulative number of valid words is less than the preset word count threshold, the total threshold determination fails.

[0089] In practice, based on the clean semantic text, the cumulative number of valid words in the current dialogue round is determined, which is the total number of words in the clean semantic text. The cumulative number of valid words is compared with a preset word count threshold. If the cumulative number of valid words is greater than or equal to the preset word count threshold, the total threshold determination is successful, triggering the semantic validity flag and determining the speech signal as a semantically valid signal. If the cumulative number of valid words is less than the preset word count threshold, the total threshold determination fails.

[0090] The current dialogue round is a complete detection cycle from when the user "starts to speak" or "a valid dialogue round begins" until the system makes a "semantic valid" judgment and responds (or the user's long pause is judged as abandonment). It can be set by the user according to the specific situation, and this application embodiment does not limit it.

[0091] In the above embodiments, the total threshold determination is applicable to users who express themselves completely and speak at a steady pace, avoiding unnecessary activation of more complex incremental analysis and keyword scanning in simple and clear dialogues, thus saving computing resources.

[0092] In an exemplary embodiment, incremental rate determination of clean semantic text includes: calculating the word increment between the cumulative number of effective words in the current dialogue round and the historical number of effective words; if the word increment is greater than or equal to a preset increment threshold, the incremental rate determination is successful; if the word increment is less than the preset increment threshold, the incremental rate determination fails.

[0093] In practice, the word increment between the cumulative number of valid words in the current dialogue round and the historical number of valid words is calculated. If the word increment is greater than or equal to the preset increment threshold, the increment rate determination is successful, triggering the semantic validity flag and confirming the speech signal as a semantically valid signal. If the word increment is less than the preset increment threshold, the increment rate determination fails.

[0094] For example, incremental rate determination is only performed when the number of decoding attempts is not equal to 0. Otherwise, incremental rate determination is skipped and keyword instant determination is performed or the word increment is calculated based on the cumulative number of valid words and the preset number of historical valid words.

[0095] Each dialogue round corresponds to one decoding result, namely the number of historical valid words, which is stored in a circular buffer.

[0096] In the above embodiments, when the previous conditions are not met, the system focuses on the user's recent speaking rate and information density. This is suitable for users who speak quickly or like to think while speaking, improving the immediacy and fluency of the interaction. The system no longer waits for the total number of words to reach the target, but can keenly capture the "momentum" of the user effectively advancing the conversation, achieving more intelligent and humanized interruption and response.

[0097] In an exemplary embodiment, keyword instant determination of clean semantic text includes: matching the clean semantic text with a preset keyword database; if a keyword is successfully matched, the keyword instant determination is successful; if all keywords in the preset keyword database are matched and no keyword is successfully matched, the keyword instant determination fails.

[0098] In practice, the clean semantic text is matched with a preset keyword database. If a keyword is successfully matched, the keyword is immediately determined to be successful, triggering the semantic validity flag and confirming that the voice signal is a semantically valid signal. If all keywords in the preset keyword database have been matched and no keyword is successfully matched, the keyword is immediately determined to be unsuccessful.

[0099] The preset keyword database is stored according to scenario categories (such as customer service, instruction, and consultation), and supports synonym and near-synonym mapping and weight configuration (different keywords have different priorities). The matching method is fast matching using a prefix tree structure, but other matching methods with matching functions can also be used. This application embodiment does not limit this.

[0100] In the above embodiments, a high-speed and accurate matching is achieved through a preset meaningful vocabulary database (such as a prefix tree structure). Once a match is successful, a valid flag is immediately triggered, which improves the efficiency of the judgment.

[0101] In an exemplary embodiment, the semantic word detection method further includes: increasing the preset word quantity threshold and the preset increment threshold when the decibel level of the speech signal is less than a first decibel threshold; and decreasing the preset word quantity threshold and the preset increment threshold when the decibel level of the speech signal is greater than a second decibel threshold.

[0102] In actual implementation, when the system is configured in volume detection collaborative mode, the audio signal is used as an auxiliary factor for judgment. When the decibel level of the voice signal is continuously lower than the first decibel threshold, the preset word quantity threshold and the preset increment threshold are increased; when the decibel level of the voice signal suddenly exceeds the second decibel threshold, the preset word quantity threshold and the preset increment threshold are decreased.

[0103] In the above embodiments, regardless of whether the environment is noisy or quiet, the system can adjust itself to strive to provide users with a relatively stable and reliable interaction success rate and response speed, thereby reducing the direct impact of the environment on the experience and improving the accuracy of the final detection results.

[0104] To illustrate the semantic word detection method in this application in detail, an embodiment is described below. For example, this application describes a semantic word detection method in a specific scenario.

[0105] First, determine the application scenario, select or receive the user's specified instructions for a specific configuration template based on the application scenario, load the specific configuration template, initialize all internal state variables, obtain the user's output voice signal from the ASR (Automatic Speech Recognition) engine, and convert the obtained voice signal into detection text.

[0106] The converted detection text undergoes semantic cleansing to obtain clean semantic text. Based on the clean semantic text, the cumulative number of valid words in the current dialogue round is determined, and a total threshold is applied. If the total threshold is successfully applied, the speech signal is determined to be a semantically valid signal.

[0107] If the total threshold determination of the pure semantic text fails, an incremental rate determination is performed on the pure semantic text. If the incremental rate determination is successful, the speech signal is determined to be a semantically valid signal. If both the total threshold determination and the incremental rate determination fail, real-time keyword determination is performed on the pure semantic text. If the real-time keyword determination is successful, the speech signal is determined to be a semantically valid signal.

[0108] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0109] Based on the same inventive concept, this application also provides a semantic word detection device for implementing the semantic word detection method described above. The solution provided by this device is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more semantic word detection device embodiments provided below can be found in the limitations of the semantic word detection method described above, and will not be repeated here.

[0110] In one exemplary embodiment, such as Figure 3 As shown, a semantic word detection device is provided, including: a conversion module 301, a generation module 302, and a judgment module 303, wherein:

[0111] The conversion module is used to receive the voice signal output by the user and convert the voice signal into detected text;

[0112] The generation module is used to perform semantic cleansing on the detected text to generate clean semantic text.

[0113] The determination module is used to determine the total amount threshold of the pure semantic text. If the total amount threshold determination is successful, the speech signal is determined to be a semantically valid signal.

[0114] The determination module is further configured to perform incremental rate determination on the pure semantic text if the total threshold determination fails, and to determine the speech signal as a semantically valid signal if the incremental rate determination succeeds.

[0115] The determination module is further configured to perform real-time keyword determination on the clean semantic text in the event of failure of incremental rate determination, and to determine the speech signal as a semantically valid signal in the event of success of real-time keyword determination.

[0116] In one exemplary embodiment, the above-described generation module is further configured to:

[0117] The detected text is segmented into words to obtain the processed detected text;

[0118] The processed detection text is matched with filler words in the preset filler word database. All successfully matched filler words in the processed detection text are removed to generate clean semantic text.

[0119] In one exemplary embodiment, the determination module is further configured to:

[0120] Based on the clean semantic text, determine the cumulative number of effective words in the current dialogue round;

[0121] If the cumulative number of valid words is greater than or equal to the preset word count threshold, the total threshold determination is successful.

[0122] If the cumulative number of valid words is less than the preset word count threshold, the total threshold determination fails.

[0123] In one exemplary embodiment, the determination module is further configured to:

[0124] Calculate the vocabulary increment between the cumulative effective vocabulary in the current dialogue round and the historical effective vocabulary;

[0125] If the increase in vocabulary is greater than or equal to the preset increment threshold, the increment rate determination is successful.

[0126] If the increase in vocabulary is less than the preset increment threshold, the increment rate determination fails.

[0127] In one exemplary embodiment, the determination module is further configured to:

[0128] Match the clean semantic text with a preset keyword database;

[0129] If a matching keyword exists, the keyword is immediately determined to be successful.

[0130] If all keywords in the preset keyword database have been matched and no keyword has been successfully matched, the keyword is immediately judged as unsuccessful.

[0131] In one exemplary embodiment, the above-described apparatus further includes an adjustment module for:

[0132] If the decibel level of the speech signal is less than the first decibel threshold, increase the preset word count threshold and the preset increment threshold.

[0133] If the decibel level of the speech signal exceeds the second decibel threshold, reduce the preset word count threshold and the preset increment threshold.

[0134] Each module in the aforementioned semantic word detection device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0135] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 4As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When executed by the processor, the computer program implements a semantic word detection method.

[0136] The display unit of this computer device is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of this computer device can be a touch layer covering the display screen, or buttons, a trackball, or a touchpad set on the casing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0137] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0138] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0139] Receives the voice signal output by the user and converts the voice signal into detected text;

[0140] The detected text is semantically cleaned to generate clean semantic text;

[0141] A total threshold is determined for the pure semantic text. If the total threshold determination is successful, the speech signal is determined to be a semantically valid signal.

[0142] If the total threshold determination fails, the incremental rate of the pure semantic text is determined. If the incremental rate determination is successful, the speech signal is determined to be a semantically valid signal.

[0143] If the incremental rate determination fails, the pure semantic text is subjected to real-time keyword determination. If the real-time keyword determination is successful, the speech signal is determined to be a semantically valid signal.

[0144] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0145] Receives the voice signal output by the user and converts the voice signal into detected text;

[0146] The detected text is semantically cleaned to generate clean semantic text;

[0147] A total threshold is determined for the pure semantic text. If the total threshold determination is successful, the speech signal is determined to be a semantically valid signal.

[0148] If the total threshold determination fails, the incremental rate of the pure semantic text is determined. If the incremental rate determination is successful, the speech signal is determined to be a semantically valid signal.

[0149] If the incremental rate determination fails, the pure semantic text is subjected to real-time keyword determination. If the real-time keyword determination is successful, the speech signal is determined to be a semantically valid signal.

[0150] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps:

[0151] Receives the voice signal output by the user and converts the voice signal into detected text;

[0152] The detected text is semantically cleaned to generate clean semantic text;

[0153] A total threshold is determined for the pure semantic text. If the total threshold determination is successful, the speech signal is determined to be a semantically valid signal.

[0154] If the total threshold determination fails, the incremental rate of the pure semantic text is determined. If the incremental rate determination is successful, the speech signal is determined to be a semantically valid signal.

[0155] If the incremental rate determination fails, the pure semantic text is subjected to real-time keyword determination. If the real-time keyword determination is successful, the speech signal is determined to be a semantically valid signal.

[0156] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0157] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0158] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0159] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for detecting semantic words, characterized in that, The method includes: Receives the voice signal output by the user and converts the voice signal into detected text; The detected text is semantically cleaned to generate clean semantic text; A total threshold is determined for the pure semantic text. If the total threshold determination is successful, the speech signal is determined to be a semantically valid signal. If the total threshold determination fails, the incremental rate of the pure semantic text is determined. If the incremental rate determination is successful, the speech signal is determined to be a semantically valid signal. If the incremental rate determination fails, the pure semantic text is subjected to real-time keyword determination. If the real-time keyword determination is successful, the speech signal is determined to be a semantically valid signal.

2. The method according to claim 1, characterized in that, The step of performing semantic cleansing on the detected text to generate clean semantic text includes: The detected text is segmented into words to obtain the processed detected text; The processed detection text is matched with filler words in the preset filler word database. All successfully matched filler words in the processed detection text are removed to generate clean semantic text.

3. The method according to claim 1, characterized in that, The determination of the total amount threshold for the pure semantic text includes: Based on the clean semantic text, determine the cumulative number of effective words in the current dialogue round; If the cumulative number of valid words is greater than or equal to the preset word count threshold, the total threshold determination is successful. If the cumulative number of valid words is less than the preset word count threshold, the total threshold determination fails.

4. The method according to claim 3, characterized in that, The incremental rate determination of the pure semantic text includes: Calculate the vocabulary increment between the cumulative effective vocabulary in the current dialogue round and the historical effective vocabulary; If the word increment is greater than or equal to a preset increment threshold, the increment rate determination is successful. If the increment of the vocabulary is less than the preset increment threshold, the increment rate determination fails.

5. The method according to claim 1, characterized in that, The instant keyword determination of the pure semantic text includes: The pure semantic text is matched with a preset keyword database; If a matching keyword exists, the keyword is immediately determined to be successful. If all keywords in the preset keyword database have been matched and no keyword has been successfully matched, the keyword is immediately judged as unsuccessful.

6. The method according to claim 1, characterized in that, The method further includes: If the decibel level of the speech signal is less than the first decibel threshold, increase the preset word count threshold and the preset increment threshold. If the decibel level of the speech signal exceeds the second decibel threshold, the preset word count threshold and the preset increment threshold are reduced.

7. A semantic word detection device, characterized in that, The device includes: The conversion module is used to receive the voice signal output by the user and convert the voice signal into detected text; The generation module is used to perform semantic cleansing on the detected text to generate clean semantic text; The determination module is used to determine the total amount threshold of the pure semantic text, and if the total amount threshold determination is successful, determine that the speech signal is a semantically valid signal. The determination module is also used to perform incremental rate determination on the pure semantic text when the total threshold determination fails, and to determine the speech signal as a semantically valid signal when the incremental rate determination succeeds. The determination module is further configured to perform real-time keyword determination on the clean semantic text in the event of failure of incremental rate determination, and to determine the speech signal as a semantically valid signal in the event of success of real-time keyword determination.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images