Voice wake-up method and device and student card

Abstract

The invention discloses a voice wake-up method, a voice wake-up device and a student card. The method comprises the following steps: receiving voice information of a target object; the matching condition of the voice information and a target voiceprint is determined through the low-power-consumption voice wake-up chip, and the target voiceprint is used for representing voice features which are pre-stored in a database and correspond to the target object; under the condition that the voice information and the target voiceprint are matched successfully, the voice information is recognized through a high-power-consumption voice processing chip, and text reply information corresponding to the voice information is generated according to a recognition result; and converting the text reply information into voice reply information, and playing the voice reply information. The technical problems of high power consumption and low recognition precision of voice equipment due to the fact that a voice wake-up function of a student card cannot accurately recognize voice information of a user in the prior art and continuous monitoring is usually carried out in the background are solved.

Description

Technical Field

[0001] The present application relates to the field of speech recognition technology, and more specifically, to a voice wake-up method, device and student card. Background Art

[0002] Student cards need to interact with users through voice in daily use, for example, through ASR (Automatic Speech Recognition) and TTS (Text-to-Speech) combined with cloud-based intelligent brain to achieve intelligent responses to questions and terminal device command operations.

[0003] In the related technology, the accuracy of the relevant algorithms in the voice wake-up devices of student cards that can identify the target user and the target sentence is not high and needs to be improved; most of the voice wake-up devices will continue to run in the background to monitor the user's voice in real time, waiting to receive the key words in the user's sentence to wake up, and then continue to read all the subsequent user's voice. Although this function ensures the interaction between the voice function and the user, the continuous running of monitoring in the background causes high power consumption of the device and continuous waste of device computing resources.

[0004] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0005] The embodiments of the present application provide a voice wake-up method, device and student card to at least solve the technical problems that the voice wake-up function of the student card in the related technology cannot accurately recognize the user's voice information, and usually continues to run monitoring in the background, resulting in high power consumption and low recognition accuracy of the voice device.

[0006] According to one aspect of an embodiment of the present application, a voice wake-up method is provided, including: receiving voice information of a target object; determining a match between the voice information and a target voiceprint through a low-power voice wake-up chip, wherein the target voiceprint is used to represent a sound feature corresponding to the target object pre-stored in a database; when the voice information matches the target voiceprint, recognizing the voice information through a high-power voice processing chip, and generating text reply information corresponding to the voice information based on the recognition result; converting the text reply information into voice reply information, and playing the voice reply information.

[0007] Optionally, the method further includes: storing a target user identifier for identifying a target object in a database, wherein a mapping relationship exists between the target user identifier and the target voiceprint.

[0008] Optionally, the matching status of the voice information and the target voiceprint is determined by a low-power voice wake-up chip, including: determining the time domain amplitude of the voice information by a low-power voice wake-up chip, wherein the time domain amplitude is used to represent the amplitude change of the voice information over time; converting the time domain amplitude into a first frequency domain amplitude by a low-power voice wake-up chip, wherein the first frequency domain amplitude is used to represent the signal strength of the voice information in frequency; matching the first frequency domain amplitude with the second frequency domain amplitude by a low-power voice wake-up chip to obtain the matching status of the voice information and the target voiceprint, wherein the second frequency domain amplitude is the frequency domain amplitude corresponding to the target voiceprint.

[0009] Optionally, the first frequency domain amplitude is matched with the second frequency domain amplitude to obtain the matching status of the speech information and the target voiceprint, including: determining a first amplitude of the first frequency domain amplitude at a preset frequency, and determining a second amplitude of the second frequency domain amplitude at a preset frequency; determining the similarity between the first frequency domain amplitude and the second frequency domain amplitude based on the first amplitude and the second amplitude; determining the matching status based on the similarity and a preset threshold.

[0010] Optionally, the voice information is recognized by a high-power voice processing chip, and a reply text corresponding to the voice information is generated based on the recognition result, including: converting the voice information into a corresponding voice feature vector; using a deep neural network model to recognize the voice feature vector to obtain a recognition result, wherein the recognition result is used to represent the probability distribution of the voice information on each voice unit; determining a voice unit sequence corresponding to the voice information based on the recognition result, wherein the voice unit sequence is used to represent a sequence of basic elements constituting the voice information; and determining text reply information based on the voice unit sequence.

[0011] Optionally, the method also includes: analyzing the text reply information to obtain text analysis results; determining a speech synthesis model based on the text analysis results, and determining sound parameters of the text reply information based on the speech synthesis model, wherein the sound parameters are used to represent audio spectrum features corresponding to the text reply information; and generating voice reply information based on the sound parameters.

[0012] Optionally, the method also includes: determining the voice amplitude of the voice information, wherein the voice amplitude is used to indicate the volume of the voice information; determining the display state of the breathing light based on the voice amplitude and a preset sound ripple amplitude, wherein the preset sound ripple amplitude is used to indicate the voice amplitude threshold pre-set in the breathing light control chip.

[0013] According to another aspect of an embodiment of the present application, a voice wake-up device is also provided, including: a receiving module for receiving voice information of a target object; a determining module for determining a match between the voice information and a target voiceprint through a low-power voice wake-up chip, wherein the target voiceprint is used to represent a sound feature corresponding to the target object pre-stored in a database; an identifying module for identifying the voice information through a high-power voice processing chip when the voice information matches the target voiceprint, and generating text reply information corresponding to the voice information based on the recognition result; and a playing module for converting the text reply information into voice reply information and playing the voice reply information.

[0014] According to another aspect of the embodiment of the present application, a student card is also provided, including: a memory and a processor, wherein the memory is used to store program instructions; the processor is connected to the memory and is used to execute the above-mentioned voice wake-up method.

[0015] According to another aspect of the embodiments of the present application, a non-volatile storage medium is further provided, wherein the non-volatile storage medium includes a stored computer program, wherein the device where the non-volatile storage medium is located executes the above-mentioned voice wake-up method by running the computer program.

[0016] According to another aspect of the embodiments of the present application, a computer program product is also provided, including computer instructions, which implement the above-mentioned voice wake-up method when executed by a processor.

[0017] In an embodiment of the present application, voice information of a target object is received; a match between the voice information and the target voiceprint is determined through a low-power voice wake-up chip, wherein the target voiceprint is used to represent the sound features corresponding to the target object pre-stored in a database; when the voice information matches the target voiceprint, the voice information is recognized through a high-power voice processing chip, and text reply information corresponding to the voice information is generated based on the recognition result; the text reply information is converted into voice reply information, and the voice reply information is played, thereby achieving the purpose of reducing the power consumption of the voice wake-up device and improving the accuracy of voice recognition based on the low-power voice wake-up chip and the high-power core processing chip, thereby achieving the technical effect of efficient and energy-saving voice interaction, and further solving the technical problems of high power consumption and low recognition accuracy of the voice device due to the fact that the voice wake-up function of the student card in the related technology cannot accurately recognize the user's voice information, and usually continues to run monitoring in the background. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0019] Figure 1 is a hardware structure diagram of a computer terminal for implementing a voice wake-up method according to an embodiment of the present application;

[0020] Figure 2 is a flow chart of a voice wake-up method according to an embodiment of the present application;

[0021] Figure 3 It is a structural diagram of a two-level architecture voice wake-up device according to an embodiment of the present application;

[0022] Figure 4 is a schematic diagram of a voice wake-up process according to an embodiment of the present application;

[0023] Figure 5 It is a structural diagram of a voice wake-up device according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0025] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] First, some nouns or terms that appear in the process of explaining the embodiments of the present application are subject to the following explanations:

[0027] ASR (Automatic Speech Recognition): Automatic speech recognition technology that can automatically recognize and transcribe speech into text, allowing computers to understand and process speech information.

[0028] TTS (Text-to-Speech): Text-to-speech technology that can convert written text into audible speech output.

[0029] In order to solve the problem of poor voice recognition ability in the related art, the embodiment of the present application provides a voice wake-up method, which can be run on Figure 1 In the computer terminal shown, the computer terminal is described below.

[0030] The voice wake-up method embodiment provided in the embodiment of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 FIG. 1 shows a hardware structure block diagram of a computer terminal for implementing a voice wake-up method. Figure 1 As shown, the computer terminal 10 may include one or more (102a, 102b, ..., 102n are used to illustrate) processors (the processor may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, and a transmission module 106 for communication functions connected via a wired and / or wireless network. In addition, it may also include: a display, a keyboard, a cursor control device, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, and a BUS bus. Those skilled in the art can understand that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0031] It should be noted that the one or more processors and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 10. As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0032] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the voice wake-up method in the embodiment of the present application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, realizing the above-mentioned voice wake-up method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0033] The transmission module 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission module 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission module 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0034] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 10 .

[0035] It should be noted that, in some optional embodiments, the above Figure 1 The computer terminal shown may include hardware elements (including circuits), software elements (including computer code stored on a computer-readable medium), or a combination of hardware elements and software elements. It should be noted that Figure 1 This is merely one example of a particular embodiment and is intended to illustrate the types of components that may be present in the computer terminal described above.

[0036] In the above operating environment, an embodiment of the present application provides an embodiment of a voice wake-up method. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0037] Figure 2 is a flow chart of a voice wake-up method according to an embodiment of the present application. Figure 2 As shown, the method comprises the following steps:

[0038] Step S202: receiving voice information of a target object.

[0039] In the above step S202, the target object (student) activates the voice wake-up system by speaking specific keywords or commands. The system receives voice information from the target object through a microphone and transmits the voice information to the system for recognition and processing.

[0040] Step S204, determining the matching status between the voice information and the target voiceprint through the low-power voice wake-up chip, wherein the target voiceprint is used to represent the sound feature corresponding to the target object pre-stored in the database.

[0041] In the above step S204, the low-power voice wake-up chip is responsible for preliminarily processing the received voice information and matching it with the target voiceprint stored in the database, wherein the target voiceprint is a pre-stored sound feature corresponding to the target object, and contains the unique voiceprint information of the target object. For example, the received voice information is filtered using the Kalman filter algorithm to remove other noise signals in the environment to retain the voice signal of the target object; further, the time domain signal corresponding to the voice information is converted into a frequency domain signal through the Fourier transform technology, and the voiceprint features in the voice information are extracted, and finally the matching between the received voice information and the target voiceprint is determined by the similarity calculation method (such as the Euclidean distance).

[0042] Step S206, when the voice information matches the target voiceprint, the voice information is recognized by the high-power voice processing chip, and text reply information corresponding to the voice information is generated according to the recognition result.

[0043] In the above step S206, once the voice information of the target object matches the target voiceprint stored in the database, the interface of the low-power voice wake-up chip will communicate with the interface of the high-power core processing chip, indicating that the system is awakened and ready to interact with the target object by voice. Furthermore, the high-power core processing chip can process the voice information of the target object through a voice recognition algorithm (such as an ASR algorithm) to generate corresponding text reply information.

[0044] Step S208, converting the text reply message into a voice reply message, and playing the voice reply message.

[0045] In the above step S208, the text reply information can be converted into audible voice reply information through text-to-speech conversion technology (such as TTS algorithm), and played to the target object through a speaker.

[0046] Through the above steps S202 to S208, the purpose of reducing the power consumption of the voice wake-up device and improving the accuracy of voice recognition is achieved based on the low-power voice wake-up chip and the high-power core processing chip, thereby achieving the technical effect of efficient and energy-saving voice interaction, and further solving the technical problems of high power consumption and low recognition accuracy of the voice device due to the fact that the voice wake-up function of the student card in the related technology cannot accurately recognize the user's voice information, and usually continues to run monitoring in the background.

[0047] Figure 3 is a structural diagram of a two-level architecture voice wake-up device according to an embodiment of the present application, such as Figure 3 As shown, the two-level architecture wake-up device includes: microphone, low-power voice wake-up chip, high-power core processing module (including high-power core and processing chip, ASR / TTS algorithm, large model intelligent question-answering system), breathing light control chip, speaker, and battery. The specific functions are as follows:

[0048] The microphone is used to convert the voice information of the target object in the air environment into a level signal, and connects with the low-power voice wake-up chip and the high-power core processing chip to transmit the level signal for use.

[0049] The battery is used to provide power for low-power voice wake-up chips, high-power core processing modules, etc.

[0050] The low-power voice wake-up chip is connected to the microphone through an interface and is embedded with a voiceprint recognition system. It also contains a database that stores the waveform of the target object's keyword voiceprint (i.e. the above-mentioned target voiceprint). It is responsible for comparing the voiceprint waveform of the target object's voice in the environment with the voiceprint waveform stored in the database to decide whether to start the large-model intelligent question-and-answer system in the core processing chip.

[0051] The high-power core processing module includes high-power core processing chip hardware, ASR / TTS, and a large-model intelligent question-and-answer system. The high-power core processing chip is connected to a microphone, a low-power voice wake-up chip, and a breathing light control chip, on which a large-model intelligent question-and-answer system, ASR, and TTS run. The voice information input by the target object is converted into text and a reply is generated. The text reply information of the question-and-answer system is converted into voice reply information and then output to other hardware through an interface.

[0052] The breathing light control chip is connected to the microphone, high-power core processing chip and breathing light. It mainly controls the opening and closing of the breathing light according to the input voice amplitude to improve the human-computer interaction experience.

[0053] The speaker is connected to the core processing chip and plays the speech generated by the large-model intelligent question-and-answer system for conversations with the target object.

[0054] The following combination Figure 3The above steps S202 to S208 are described in detail with respect to each component and function.

[0055] In the embodiment of the present application, a target user identifier for identifying a target object is stored in the database, wherein a mapping relationship exists between the target user identifier and the target voiceprint.

[0056] Specifically, the database stores voiceprint data associated with each target object, including the target object's voice characteristics, commonly used keywords, etc., and each target object's voiceprint data is associated with a unique target user identifier, that is, each target object has a one-to-one corresponding target user identifier, which is used to retrieve and match specific voiceprint data in the database, ensuring the personalization and uniqueness of the voiceprint data.

[0057] In the above step S204, the matching status of the voice information and the target voiceprint is determined by the low-power voice wake-up chip, including: determining the time domain amplitude of the voice information by the low-power voice wake-up chip, wherein the time domain amplitude is used to represent the amplitude change of the voice information over time; converting the time domain amplitude into a first frequency domain amplitude by the low-power voice wake-up chip, wherein the first frequency domain amplitude is used to represent the signal strength of the voice information in frequency; matching the first frequency domain amplitude with the second frequency domain amplitude by the low-power voice wake-up chip to obtain the matching status of the voice information and the target voiceprint, wherein the second frequency domain amplitude is the frequency domain amplitude corresponding to the target voiceprint.

[0058] In the embodiment of the present application, preliminary voiceprint matching is performed through a low-power voice wake-up chip, which can achieve efficient voiceprint recognition without consuming a lot of energy. It should be noted that under normal circumstances, the high-power core processing chip and voice question and answer function of the learning card are in a dormant state, and the microphone hardware interface is connected to the low-power voice wake-up chip interface. The microphone reads the voice information of the target object and sends it to the low-power voice wake-up chip. A fixed frequency is set in the low-power voice wake-up chip to monitor the voice of the target object, and the frequency of voice collection is controlled by the interface with the microphone hardware. In the low-power voice wake-up chip, a voiceprint recognition system is constructed to identify the identity and keywords of the target object, ensuring that the high-power core processing module is only woken up when the target object and the target voiceprint match, such as Figure 4 As shown, the overall process is as follows:

[0059] First, the voice information of the target object is monitored, and the received voice information is filtered using the Kalman filter algorithm to remove other noise signals in the environment in order to retain the voice signal of the target object.

[0060] Secondly, the time domain signal corresponding to the voice information (i.e., the above-mentioned time domain amplitude) is converted into a frequency domain signal (i.e., the above-mentioned first frequency domain amplitude) through Fourier transform technology to extract the voiceprint waveform features in the voice information. Among them, the time domain amplitude refers to the amplitude change of the voice signal in the time dimension, which reflects the change of the signal strength over time. Therefore, the analysis of the time domain amplitude helps to capture the dynamic characteristics of the voice signal, including the characteristics of the start, duration and end of the voice. The frequency domain amplitude represents the intensity or energy of the voice signal at different frequencies. Therefore, the analysis of the frequency domain amplitude can reveal the spectral characteristics of the voice signal, which is crucial for distinguishing the voices of different speakers. The two provide different perspectives to understand and process the voice information of the target object. The analysis of the time domain amplitude focuses on the change of the signal over time, while the analysis of the frequency domain amplitude focuses on the frequency components of the signal. For example, in voiceprint recognition, frequency domain analysis can help extract the unique frequency characteristics of the speaker, while time domain analysis may focus on the dynamic changes and rhythm patterns of the sound.

[0061] Finally, the converted frequency domain amplitude (first frequency domain amplitude) is compared with the frequency domain amplitude (second frequency domain amplitude) corresponding to the target voiceprint stored in the specific sentence database to verify whether the received voice information matches the sound characteristics of the target object.

[0062] Furthermore, the first frequency domain amplitude is matched with the second frequency domain amplitude to obtain the matching status of the speech information and the target voiceprint, including: determining a first amplitude of the first frequency domain amplitude at a preset frequency, and determining a second amplitude of the second frequency domain amplitude at a preset frequency; determining the similarity between the first frequency domain amplitude and the second frequency domain amplitude based on the first amplitude and the second amplitude; determining the matching status based on the similarity and a preset threshold.

[0063] In the embodiment of the present application, by comparing the amplitude of the frequency domain amplitude and calculating the similarity, the system can accurately evaluate the matching between the voice information and the target voiceprint, thereby achieving efficient voiceprint recognition. The specific process can be as follows:

[0064] The obtained first frequency domain amplitude and the second frequency domain amplitude corresponding to the target object in the database are digitized for comparison, for example, by uniformly sampling the points on the curve and storing the frequency and amplitude of the sampling points as digital data.

[0065] Secondly, a mathematical method is used to calculate the similarity or consistency between the first frequency domain amplitude and the second frequency domain amplitude, such as Euclidean distance, cosine similarity, correlation coefficient, etc. Taking Euclidean distance as an example, it measures the distance between two vectors, and the specific calculation formula is as follows:

[0066]

[0067] In the formula, xi Indicates the first amplitude of the first frequency domain wave at the preset frequency, y i Indicates the second amplitude of the second frequency domain wave at the preset frequency.

[0068] Through the above calculation, the similarity or consistency between the first frequency domain amplitude and the second frequency domain amplitude can be obtained, wherein the smaller the Euclidean distance d is, the more similar the first frequency domain amplitude and the second frequency domain amplitude are.

[0069] Finally, the match is determined based on the similarity and the preset threshold. The preset threshold is pre-set according to the system design and performance requirements, and is used to determine whether the similarity is high enough to determine the voiceprint match. If the calculated similarity exceeds the preset threshold, the system will consider that the voice information matches the target voiceprint successfully, indicating that the received voice information is indeed from the target object, and the high-power core processing module will be started; if the similarity is lower than the threshold, the system will consider that the match has failed, which may be due to the mismatch between the voice information and the target voiceprint or the presence of noise interference, and will return to monitoring the target object's voice operation.

[0070] In the above step S206, the voice information is recognized by a high-power voice processing chip, and a reply text corresponding to the voice information is generated based on the recognition result, including: converting the voice information into a corresponding voice feature vector; using a deep neural network model to recognize the voice feature vector to obtain a recognition result, wherein the recognition result is used to represent the probability distribution of the voice information on each voice unit; determining a voice unit sequence corresponding to the voice information based on the recognition result, wherein the voice unit sequence is used to represent a sequence of basic elements constituting the voice information; and determining text reply information based on the voice unit sequence.

[0071] In the embodiment of the present application, the high-power core processing module of the student card includes high-power core processing chip hardware, ASR / TTS, and large-model intelligent question-answering system. Once the voice information of the target object matches the target voiceprint stored in the database, the interface of the low-power voice wake-up chip will communicate with the interface of the high-power core processing chip, and the high-power core processing chip will start to recognize and process the voice information of the target object and generate corresponding text reply information.

[0072] When the low-power voice wake-up chip activates the voice question and answer module, in order to switch to the voice question and answer mode, it is necessary to close the interface between the microphone and the low-power voice wake-up chip, and open the connection interface with the core processing chip so that the high-power core processing chip can receive the voice information of the target object. The high-power core processing chip takes over the control of the microphone device and starts reading the voice input data of the target object at a preset frequency. After reading the voice information through the microphone, the high-power core processing chip also uses the Kalman filter algorithm to filter other noises in the environment to improve the accuracy and stability of speech recognition. The voice data after noise filtering is sent to the control chip for subsequent processing, which may include speech recognition, command execution and other operations. The specific process can be as follows:

[0073] First, the speech information is converted into a speech feature vector. Specifically, feature extraction is performed on the speech signal after noise is filtered out, for example, using methods such as Mel frequency cepstral coefficients and filter bank banks. The extracted speech features are preprocessed, such as normalization or standardization, to obtain speech feature vectors to ensure that they meet the input requirements of the model.

[0074] Secondly, a high-power speech processing model, such as a Gaussian mixture model, a hidden Markov model, a deep neural network, etc., is used to perform recognition processing on the speech feature vector to obtain a recognition result, which is used to represent the probability distribution of the speech information of the target object on each speech unit (the basic building block of speech information, such as phonemes, syllables, etc.), that is, the recognition confidence of the model for different speech units.

[0075] A decoding strategy, such as maximum a posteriori probability decoding and greedy decoding, is used to determine the most likely speech unit sequence based on the recognition results. The speech unit sequence is used to represent the order of basic elements (such as phonemes, syllables, etc.) that constitute speech information.

[0076] Finally, the recognized speech unit sequence is mapped to the corresponding text or vocabulary to generate a text recognition result, which is converted into text output and passed as input to the large model intelligent question-answering system to generate text reply information.

[0077] Further, the text reply information is analyzed to obtain text analysis results; a speech synthesis model is determined based on the text analysis results, and sound parameters of the text reply information are determined based on the speech synthesis model, wherein the sound parameters are used to represent audio spectrum features corresponding to the text reply information; and speech reply information is generated based on the sound parameters.

[0078] In the embodiment of the present application, the process of converting the above reply text into voice reply information by using a text-to-speech (TTS) algorithm in the student card high-power core processing module can be as follows:

[0079] The generated reply text is analyzed, such as word segmentation, grammatical analysis, semantic understanding and other processing steps, to obtain the text analysis results. According to the text analysis results, a suitable speech synthesis model is selected, which is a rule-based, splicing-based or neural network-based synthesis method. According to the selected speech synthesis model, the sound parameters of the reply text are determined, and the sound parameters define the physical characteristics of the speech, such as pitch, duration, intensity, spectrum envelope, etc. Using the acoustic model, the sound parameters are converted into corresponding audio waveforms, such as by using a vocoder to generate an actual speech waveform, and the generated audio waveform can also be processed subsequently, such as denoising, enhancement, timbre adjustment, etc., to improve the naturalness and clarity of the speech. In addition, the generated speech waveform can also be adjusted according to the timbre set in the database to meet the preferences of the target object or the needs of a specific scene. Finally, the generated speech is connected to the interface of the breathing light control chip, and the speech is output to the breathing light control chip and played to the target object through the speaker.

[0080] Optionally, the above method also includes: determining the voice amplitude of the voice information, wherein the voice amplitude is used to indicate the volume of the voice information; determining the display state of the breathing light based on the voice amplitude and a preset sound ripple amplitude, wherein the preset sound ripple amplitude is used to indicate the voice amplitude threshold pre-set in the breathing light control chip.

[0081] In the embodiment of the present application, by measuring the voice amplitude of the target object and comparing it with the preset threshold, the system can intelligently control the display state of the breathing light, thereby providing intuitive user feedback and enhancing the overall interactive experience. Specifically, the breathing light control chip hardware interface connects the breathing light control chip and the high-power core processing chip, and sets the sound ripple amplitude threshold (preset sound ripple amplitude) in the breathing light control chip to determine the volume (voice amplitude) of the target object's input voice. When the input voice amplitude is greater than the preset sound ripple amplitude, the control chip controls the breathing light circuit to turn on, and the breathing light is on to prompt the target object that the system is receiving voice input; when the input voice amplitude is less than the preset sound ripple amplitude, the breathing light goes out to prompt the target object that it is in a waiting state or has completed receiving voice input. In addition, during the voice reply process, the system can also adjust the brightness of the breathing light according to the voice amplitude of the returned voice reply information to provide additional feedback. For example, when the voice amplitude of the returned voice reply information is larger, the breathing light will be brighter, otherwise it will be darker.

[0082] In the embodiment of the present application, by adopting a two-level architecture, namely, including a low-power voice wake-up chip and a high-power core processing chip, the low-power voice wake-up chip matches the voiceprint waveform of key sentences in the voiceprint recognition system and the database. When the matching degree reaches the set threshold, the high-power core processing chip is awakened for ASR and TTS processing, realizing the combination of low-power monitoring and high-precision recognition, and effectively solving the problems of high power consumption and low recognition accuracy of existing voice devices. Among them, by combining the Fourier transform algorithm and the Kalman filter algorithm, the accuracy and anti-noise ability of voiceprint recognition are significantly improved, so that the wake-up command of the target object can be responded quickly and accurately in a complex environment. At the same time, the use of multi-level wake-up mechanism and dynamic power management strategy effectively reduces the energy consumption of the device, prolongs the battery life, and improves the target object experience. In addition, the addition of the breathing light control mechanism not only enhances the interactivity of the device, but also enables the target object to intuitively perceive the working status of the device, further improving the target object friendliness of the product, and has significant technical progress and application value.

[0083] According to an embodiment of the present application, a voice wake-up device is provided. It should be noted that the voice wake-up device of the embodiment of the present application can be used to execute the voice wake-up method provided by the embodiment of the present application. The voice wake-up device provided by the embodiment of the present application is introduced below.

[0084] Figure 5 is a structural diagram of a voice wake-up device provided according to an embodiment of the present application. Figure 5 As shown, the device comprises:

[0085] A receiving module 50 is used to receive the voice information of the target object;

[0086] A determination module 52, used to determine the matching between the voice information and the target voiceprint through the low-power voice wake-up chip, wherein the target voiceprint is used to represent the sound feature corresponding to the target object pre-stored in the database;

[0087] The recognition module 54 is used to recognize the voice information through the high-power voice processing chip when the voice information matches the target voiceprint, and generate text reply information corresponding to the voice information according to the recognition result;

[0088] The playback module 56 is used to convert the text reply information into voice reply information and play the voice reply information. Through the receiving module 50, the determining module 52, the identifying module 54 and the playing module 56 in the above-mentioned voice wake-up device, the purpose of reducing the power consumption of the voice wake-up device and improving the accuracy of voice recognition based on the low-power voice wake-up chip and the high-power core processing chip is achieved, thereby achieving the technical effect of efficient and energy-saving voice interaction, and further solving the technical problems that the voice wake-up function of the student card in the related technology cannot accurately recognize the user's voice information, and usually continues to run monitoring in the background, resulting in high power consumption of the voice device and low recognition accuracy.

[0089] In the voice wake-up device provided in the embodiment of the present application, the determination module is also used to determine the time domain amplitude of the voice information through a low-power voice wake-up chip, wherein the time domain amplitude is used to represent the amplitude change of the voice information over time; the time domain amplitude is converted into a first frequency domain amplitude through the low-power voice wake-up chip, wherein the first frequency domain amplitude is used to represent the signal strength of the voice information in frequency; the first frequency domain amplitude is matched with the second frequency domain amplitude through the low-power voice wake-up chip to obtain the matching status of the voice information and the target voiceprint, wherein the second frequency domain amplitude is the frequency domain amplitude corresponding to the target voiceprint.

[0090] In the voice wake-up device provided in the embodiment of the present application, the determination module is also used to determine the voice amplitude of the voice information, wherein the voice amplitude is used to indicate the volume of the voice information; the display state of the breathing light is determined based on the voice amplitude and the preset sound ripple amplitude, wherein the preset sound ripple amplitude is used to indicate the voice amplitude threshold value pre-set in the breathing light control chip.

[0091] In the voice wake-up device provided in the embodiment of the present application, the determination module is also used to determine a first amplitude of the first frequency domain amplitude at a preset frequency, and to determine a second amplitude of the second frequency domain amplitude at a preset frequency; determine the similarity between the first frequency domain amplitude and the second frequency domain amplitude based on the first amplitude and the second amplitude; determine the matching situation based on the similarity and a preset threshold.

[0092] In the voice wake-up device provided in the embodiment of the present application, the recognition module is also used to convert voice information into a corresponding voice feature vector; use a deep neural network model to recognize the voice feature vector to obtain a recognition result, wherein the recognition result is used to represent the probability distribution of the voice information on each voice unit; determine the voice unit sequence corresponding to the voice information based on the recognition result, wherein the voice unit sequence is used to represent the basic element sequence constituting the voice information; determine the text reply information based on the voice unit sequence.

[0093] In the voice wake-up device provided in the embodiment of the present application, the playback module is also used to analyze the text reply information to obtain a text analysis result; determine a speech synthesis model based on the text analysis result, and determine the sound parameters of the text reply information based on the speech synthesis model, wherein the sound parameters are used to represent the audio spectrum characteristics corresponding to the text reply information; generate voice reply information based on the sound parameters.

[0094] In the voice wake-up device provided in the embodiment of the present application, the determination module is also used to determine the voice amplitude of the voice information, wherein the voice amplitude is used to indicate the volume of the voice information; the display state of the breathing light is determined based on the voice amplitude and the preset sound ripple amplitude, wherein the preset sound ripple amplitude is used to indicate the voice amplitude threshold value pre-set in the breathing light control chip.

[0095] An embodiment of the present application also provides a student card, including: a memory and a processor, wherein the memory is used to store program instructions; the processor is connected to the memory and is used to execute the above-mentioned voice wake-up method.

[0096] It should be noted that the above student card is used to execute Figure 2 The voice wake-up method shown in the figure, therefore the relevant explanations in the above voice wake-up method are also applicable to this student card and will not be repeated here.

[0097] An embodiment of the present application also provides a non-volatile storage medium, which includes a stored computer program, wherein the device where the non-volatile storage medium is located executes the above-mentioned voice wake-up method by running the computer program.

[0098] It should be noted that the above non-volatile storage medium is used to execute Figure 2 The voice wake-up method shown, therefore the relevant explanations in the above voice wake-up method are also applicable to the non-volatile storage medium, and will not be repeated here.

[0099] The embodiment of the present application also provides a computer program product, including computer instructions, which implement the above-mentioned voice wake-up method when executed by a processor.

[0100] It should be noted that the above-mentioned computer program product is used to execute Figure 2 The voice wake-up method shown, therefore the relevant explanations in the above voice wake-up method are also applicable to the computer program product, and will not be repeated here.

[0101] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0102] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0103] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0104] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0105] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0106] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, RandomAccess Memory), mobile hard disk, disk or CD-ROM and other media that can store program codes.

[0107] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A voice wake-up method, characterized in that: include: Receive voice information from the target object; Determining the matching status between the voice information and the target voiceprint through a low-power voice wake-up chip, wherein the target voiceprint is used to represent the sound feature corresponding to the target object pre-stored in a database; When the voice information matches the target voiceprint, the voice information is recognized by a high-power voice processing chip, and a text reply message corresponding to the voice information is generated according to the recognition result; The text reply message is converted into a voice reply message, and the voice reply message is played.

2. The method according to claim 1, characterized in that The method further comprises: The database stores a target user identifier for identifying the target object, wherein a mapping relationship exists between the target user identifier and the target voiceprint.

3. The method according to claim 1, characterized in that The matching condition between the voice information and the target voiceprint is determined by a low-power voice wake-up chip, including: Determining the time domain amplitude of the voice information by using the low-power voice wake-up chip, wherein the time domain amplitude is used to represent the amplitude change of the voice information over time; The time domain amplitude is converted into a first frequency domain amplitude by the low-power voice wake-up chip, wherein the first frequency domain amplitude is used to represent the signal strength of the voice information in frequency; The first frequency domain amplitude is matched with the second frequency domain amplitude through the low-power voice wake-up chip to obtain the matching status of the voice information and the target voiceprint, wherein the second frequency domain amplitude is the frequency domain amplitude corresponding to the target voiceprint.

4. The method according to claim 3, characterized in that Matching the first frequency domain amplitude with the second frequency domain amplitude to obtain a matching condition between the voice information and the target voiceprint includes: Determine a first amplitude of the first frequency domain amplitude at a preset frequency, and determine a second amplitude of the second frequency domain amplitude at the preset frequency; Determining the similarity between the first frequency domain amplitude and the second frequency domain amplitude according to the first amplitude and the second amplitude; The matching condition is determined according to the similarity and a preset threshold.

5. The method according to claim 1, characterized in that The voice information is recognized by a high-power voice processing chip, and a reply text corresponding to the voice information is generated according to the recognition result, including: Converting the speech information into a corresponding speech feature vector; Using a deep neural network model to recognize the speech feature vector to obtain the recognition result, wherein the recognition result is used to represent the probability distribution of the speech information on each speech unit; Determining a speech unit sequence corresponding to the speech information according to the recognition result, wherein the speech unit sequence is used to represent a basic element sequence constituting the speech information; The text reply information is determined according to the speech unit sequence.

6. The method according to claim 1, characterized in that The method further comprises: Analyze the text reply information to obtain a text analysis result; Determining a speech synthesis model according to the text analysis result, and determining sound parameters of the text reply information according to the speech synthesis model, wherein the sound parameters are used to represent audio spectrum features corresponding to the text reply information; The voice reply information is generated according to the sound parameters.

7. The method according to claim 1, characterized in that The method further comprises: Determining the voice amplitude of the voice information, wherein the voice amplitude is used to represent the volume of the voice information; The display state of the breathing light is determined according to the voice amplitude and the preset sound ripple amplitude, wherein the preset sound ripple amplitude is used to represent the voice amplitude threshold value pre-set in the breathing light control chip.

8. A voice wake-up device, characterized in that: include: A receiving module, used for receiving the voice information of the target object; A determination module, used to determine the matching status between the voice information and the target voiceprint through a low-power voice wake-up chip, wherein the target voiceprint is used to represent the sound feature corresponding to the target object pre-stored in a database; A recognition module, configured to recognize the voice information through a high-power voice processing chip when the voice information matches the target voiceprint, and generate text reply information corresponding to the voice information according to the recognition result; The playing module is used to convert the text reply information into voice reply information and play the voice reply information.

9. A student card, characterized in that: include: A memory and a processor, wherein the memory is used to store program instructions; The processor is connected to the memory and is used to execute the voice wake-up method described in any one of claims 1 to 7.

10. A non-volatile storage medium, characterized in that: The non-volatile storage medium includes a stored computer program, wherein the device where the non-volatile storage medium is located executes the voice wake-up method according to any one of claims 1 to 7 by running the computer program.

11. A computer program product comprising computer instructions, characterized in that: When the computer instructions are executed by a processor, the voice wake-up method described in any one of claims 1 to 7 is implemented.

Images