Awakening method of voice recognition device, voice recognition device and carrying device thereof

Abstract

The embodiment of the present application provides a wake-up method for a voice recognition device, a voice recognition device, and a carrier thereof. According to the solution provided in the embodiment of the present application, when the voice recognition device enters a sleep state, the processor and other modules enter a sleep state. When the vibration signal sensed by the vibration sensor is received, the processor in the sleep state is awakened; the processor determines the signal characteristics of the vibration signal; if the vibration signal is determined to be a valid signal based on the signal characteristics of the vibration signal, the voice wake-up module in the sleep state is awakened to perform voice recognition; otherwise, the processor is restored to a sleep state. The vibration signal in the environment is detected in advance by the vibration sensor, and the normal working state of the processor is started and maintained only when the vibration signal is a valid signal, and the processor is restored to a sleep state when the vibration signal is an invalid signal, which effectively saves the power consumption of the system.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of computer technology, and in particular to a wake-up method for a voice recognition device, a voice recognition device, and a carrier thereof. Background Art

[0002] With the in-depth development of speech recognition, speech recognition devices are also widely used in various systems. In order to reduce system power consumption, speech recognition devices often enter a dormant state when there is no process to be processed.

[0003] To quickly restart a speech recognition device from sleep mode, all processors in a typical device are not put into sleep mode. For example, a single processor core with lower performance is kept active to perform the wake-up function. Alternatively, when no lower performance core is available, a processor core with higher performance is kept active to perform the wake-up function. This still wastes system power.

[0004] Based on this, a wake-up solution for a voice recognition device that can save more system power consumption is needed. Summary of the Invention

[0005] In view of this, an embodiment of the present application provides a wake-up solution for a voice recognition device to at least partially solve the above-mentioned problem.

[0006] According to a first aspect of an embodiment of the present application, a wake-up method for a voice recognition device is provided, wherein the voice recognition device includes: a vibration sensor, a processor, and a voice wake-up module. The method includes: when the processor receives an activation signal sent by a vibration signal sensed by the vibration sensor, changing the sleep state to the working state; determining the signal characteristics of the vibration signal; if the vibration signal is determined to be a valid signal based on the signal characteristics of the vibration signal, waking up the voice wake-up module in the sleep state to perform voice recognition; otherwise, restoring the processor to the sleep state.

[0007] According to a second aspect of an embodiment of the present application, a speech recognition device is provided, comprising: a vibration sensor, which senses a vibration signal generated in an environment and generates an activation signal; a processor, which receives the activation signal generated by the vibration sensor and changes a sleep state to a working state; determines a signal feature of the vibration signal; if the vibration signal is determined to be a valid signal based on the signal feature of the vibration signal, wakes up a speech wake-up module in a sleep state; otherwise, restores the processor to a sleep state; and the speech wake-up module performs speech recognition after being awakened from the sleep state.

[0008] According to a third aspect of an embodiment of the present application, a transport device is provided, on which the speech recognition device as described in the third aspect is configured.

[0009] According to a fourth aspect of an embodiment of the present application, a computer storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the wake-up method of the speech recognition device as described in the first aspect is implemented.

[0010] According to a fifth aspect of an embodiment of the present application, a computer program product is provided, comprising computer instructions, wherein the computer instructions instruct a computing device to execute operations corresponding to the wake-up method of the speech recognition device as described in the first aspect.

[0011] According to the solution provided in the embodiment of the present application, when the voice recognition device enters the sleep state, the processor and other modules also enter the sleep state. When the activation signal sent by the vibration sensor is received due to the vibration signal sensed, the sleep state is changed to the working state; the signal characteristics of the vibration signal are determined; if the vibration signal is determined to be a valid signal based on the signal characteristics of the vibration signal, the voice wake-up module in the sleep state is awakened to perform voice recognition; otherwise, the processor is restored to the sleep state. In the embodiment of the present application, the vibration signal in the environment is detected in advance by the vibration sensor, and the normal working state of the processor is started and maintained only when the vibration signal is a valid signal. When the vibration signal is an invalid signal, the processor is restored to the sleep state, effectively saving the power consumption of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments recorded in the embodiments of the present application. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0013] Figure 1 A flowchart of a method for waking up a voice recognition device provided in an embodiment of the present application;

[0014] Figure 2 A schematic diagram of an acceleration component provided in an embodiment of the present application;

[0015] Figure 3 A schematic diagram of a sampling space provided in an embodiment of the present application;

[0016] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present application, the technical solutions 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 embodiments of the present application, all other embodiments obtained by ordinary technicians in this field should fall within the scope of protection of the embodiments of the present application.

[0018] In current voice wake-up devices, in order to enable faster system wake-up, a single, lower-performance processor core is typically used to run the wake-up function. However, when some hardware or platforms do not have a similar lower-performance processor core, the system must run on a higher-performance processor, which significantly wastes system power. Therefore, embodiments of the present application provide a more power-efficient wake-up solution for voice recognition devices.

[0019] The following further describes the specific implementation of the embodiment of the present application in conjunction with the accompanying drawings of the embodiment of the present application. Figure 1 As shown, Figure 1 A flowchart of a method for waking up a voice recognition device provided in an embodiment of the present application specifically includes:

[0020] S101: When the processor receives an activation signal sent by the vibration sensor in response to a vibration signal sensed by the vibration sensor, the processor changes the sleep state to the working state.

[0021] As mentioned above, in conventional methods, it is necessary to keep at least some processor cores in an active state. However, in the scenario involved in the embodiments of the present application, the speech recognition device is already in a dormant state, at which point all processors and corresponding other modules in the speech recognition device are also in a dormant state.

[0022] For example, after getting off a vehicle, if the user is likely to return soon, the user may set the vehicle's control center to sleep. At this point, all other modules in the control center, including the processor and the voice wake-up module in the smart speaker, can simultaneously enter sleep mode.

[0023] However, at this time, the vibration sensor (e.g., gyroscope sensor) included in the vehicle is still operating normally. The vibration sensor is a simple and easy-to-use positioning control device based on free space movement, which can accurately determine the orientation of the vehicle. In addition, it can sensitively sense the vibration signals generated by the surrounding environment, including vibration signals generated by air vibration and vibration signals caused by the shaking of the vehicle in which the gyroscope is located. For example, the vibration sensor can perform sampling at a preset sampling frequency (e.g., a sampling frequency of 150 Hz), with each sampling obtaining a sampling point. Each sampling point contains the acceleration characteristics of the vibration sensor and related information of the vibration signal, and multiple sampling points within a continuous preset time duration are regarded as a sampling space (e.g., each 1 second interval is regarded as a sampling space, and at a sampling frequency of 150 Hz, there are 150 sampling points per second). Based on the information of the multiple sampling points in the sampling space, the device simulates and obtains the vibration signal corresponding to the sampling space.

[0024] The vibration signal generated by the surrounding environment can be a voice signal generated when the user speaks, or a sound signal generated when the user opens the car door and performs other corresponding operations when entering the vehicle, or a signal representing the vibration of the vehicle caused by the shaking of the vehicle when the user gets in the vehicle, or a mixed signal (that is, including a sound signal and a signal representing the vibration of the vehicle) generated by the user patting or knocking certain parts of the vehicle.

[0025] The vibration sensor detects the vibration signal and sends a corresponding activation signal to the processor. At this time, the processor changes from a dormant state to a normal working state to process the received vibration signal. For example, the vibration sensor can wake up the processor by activating a wake-up pin on the processor to which it is connected (for example, by a level change from low to high on the pin).

[0026] S103: Determine the signal characteristics of the vibration signal.

[0027] The vibration signal can be regarded as a time-related signal sequence, and the corresponding signal characteristics of the vibration signal can be obtained by analyzing methods such as Fast Fourier Transform (FFT).

[0028] Specifically, the signal characteristics of the vibration signal include the acceleration characteristics of the vibration sensor, or the frequency domain characteristics and time domain characteristics of the vibration signal.

[0029] The time domain features include waveform indicators, pulse indicators, kurtosis indicators, margin indicators, peak value, vibration duration, etc.; the frequency domain features include center of gravity frequency, mean square frequency, root mean square frequency, frequency variance, frequency standard deviation, frequency area, etc. The frequency domain features and time domain features can be obtained based on the aforementioned FFT method.

[0030] The acceleration characteristics of a vibration sensor include the acceleration of the vibration signal corresponding to each sampling space. They may also include three acceleration components corresponding to the three axes of the vibration sensor's pre-established spatial coordinate system, namely, the three acceleration components on the x, y, and z axes. Furthermore, the acceleration characteristics of a vibration sensor may also include the three acceleration components corresponding to each sampling point.

[0031] S105 , if the vibration signal is determined to be a valid signal according to the signal characteristics of the vibration signal, then waking up the voice wake-up module in the sleep state to perform voice recognition; otherwise, restoring the processor to the sleep state.

[0032] Specifically, whether the vibration signal is a valid signal is determined by determining whether the signal characteristics of the vibration signal meet certain preset conditions. The preset conditions can be preset differently based on different scenarios.

[0033] For example, if the automatic control module is activated based on the sound of the user closing the door when getting in the car, then the sound signal generated by the user closing the car door is expected to be regarded as a valid signal. Since the sound vibration generated by closing the car door is usually short and low-frequency, a shorter pulse time, a shorter vibration duration, and a larger peak-to-peak value can be selected as time domain features, and a lower frequency variance and a lower frequency area can be selected as the conditions that the vibration signal needs to meet.

[0034] For example, if the user wakes up the corresponding intelligent voice function through voice commands after getting in the car, for example, the user wakes up the smart speaker through voice commands after getting in the car, then the user's voice command is expected to be regarded as a valid signal. Since the frequency of the sound vibration generated by the user's voice command is usually within the normal sound range and the duration is medium, medium vibration duration and large peak-to-peak value can be selected as time domain features, while low frequency variance and medium frequency range can be selected as the conditions that the vibration signal must meet.

[0035] For another example, if the shaking generated when the user gets on the car is used as an indicator that the user has gotten on the car and the system needs to work normally, since the vibration signal corresponding to the shaking generated when the user gets on the car actually lasts for a long time and has a very low frequency, a longer vibration duration can be selected as the time domain feature, and a very low frequency area can be selected as the condition that the vibration signal needs to meet.

[0036] For example, in a scenario where the system needs to operate normally, a mixed signal generated by a user tapping or knocking on certain vehicle components is used to indicate that the user has boarded the vehicle. Since the sound signal and the signal representing the shaking are significantly different, the mixed signal can be distinguished using the aforementioned FFT signal processing to obtain the sound signal and the signal representing the shaking, respectively. Furthermore, a shorter pulse time, a larger peak-to-peak value, and a larger frequency variance can be used as the preset conditions that the mixed signal needs to meet. Furthermore, specific conditions that the sound signal and the signal representing the shaking can also be set separately.

[0037] In other words, in different scenarios, the selection criteria for valid signals are different, so the selection items and ranges of the time domain features and frequency domain features are based on the scenarios and all the differences are not listed here one by one.

[0038] Furthermore, the acceleration characteristics of vibration sensors vary in different scenarios. For example, in a voice wake-up scenario, the acceleration caused by the vibration signal from the air vibrations generated by the user's speech is generally not very large. Therefore, a second preset condition with a lower acceleration threshold, such as 0.30 m / s², can be set. This means that when the acceleration of the vibration sensor exceeds 0.30 m / s², the vibration signal is considered valid.

[0039] In the scenario where the device is woken up by the sound of the door closing when the user gets in the car, the vibration signal generated at this time is generally large and short-lived. Therefore, a second preset condition including a larger acceleration threshold can be set, such as 2m / s2. That is, when the acceleration of the vibration sensor exceeds 2m / s2, the vibration signal is determined to be a valid signal (that is, the user may have opened the door and got in the car).

[0040] If a vibration signal is determined to be a valid signal, it means that the system needs to prepare to enter normal working state. Therefore, the processor will be activated first, that is, the processor will maintain normal working state and send relevant activation signals to the corresponding voice wake-up module according to the requirements of the scene, so that the corresponding voice wake-up module also enters normal working state from sleep state to prepare for subsequent voice recognition.

[0041] In addition, it should be noted that if the accuracy of the vibration sensor is high, the aforementioned vibration signal can also be used for voice recognition when the voice wake-up module performs subsequent voice recognition, thereby achieving faster voice recognition.

[0042] If a vibration signal is not a valid signal, the processor will re-enter the sleep state and will not send an activation signal to the voice wake-up module.

[0043] According to the solution provided in the embodiment of the present application, when the voice recognition device enters the sleep state, the processor and other modules also enter the sleep state. When the activation signal sent by the vibration sensor is received due to the vibration signal sensed by the vibration sensor, the processor in the sleep state is awakened; the signal characteristics of the vibration signal are determined; if the vibration signal is determined to be a valid signal based on the signal characteristics of the vibration signal, the voice wake-up module in the sleep state is awakened to perform voice recognition; otherwise, the processor is restored to the sleep state. In the embodiment of the present application, the vibration signal in the environment is detected in advance by the vibration sensor, and the normal working state of the processor is started and maintained only when the vibration signal is a valid signal. When the vibration signal is an invalid signal, the processor is restored to the sleep state, effectively saving the power consumption of the system.

[0044] In one embodiment, when the processor determines the signal characteristics of the vibration signal, it may only need to use the acceleration characteristics of the vibration sensor as the signal characteristics of the vibration signal. Since the acceleration characteristics of the vibration sensor can be directly obtained by signal acquisition using a vibration sensor with corresponding functions (such as a gyroscope sensor), this method can more quickly and efficiently obtain the signal characteristics of the vibration signal.

[0045] In one embodiment, based on the application scenario, further determination of effective signals may be performed based on acceleration characteristics of the vibration sensor.

[0046] Specifically, the acceleration characteristics of the vibration sensor can be decomposed to obtain the three acceleration components corresponding to the acceleration characteristics of the vibration sensor in the three directions of the preset space (i.e., the three acceleration components in the aforementioned x, y, and z directions). Figure 2 As shown, Figure 2 A schematic diagram of an acceleration component provided in an embodiment of the present application.

[0047] If only one of the three acceleration components in the x, y, and z directions exceeds the acceleration threshold (e.g., 0.25 m / s2), the vibration signal is considered invalid. Only when two or three acceleration components exceed the acceleration threshold is the vibration signal considered valid. The specific value of the acceleration threshold can be determined empirically based on the needs of the scenario. Due to the differences in the vibration signals generated in each scenario, this method can further accurately determine the valid signal in each scenario, avoid invalid wake-ups, and further save power consumption.

[0048] In one embodiment, if there are at least two acceleration components that exceed the acceleration threshold, then it is possible to further determine whether the vibration signal is a valid signal based on the information of the sampling point. For example, at a sampling frequency of 150 Hz, there will be 150 sampling points in every 1 second of sampling space. At this time, multiple sampling points in the sampling space (i.e., 150 sampling points) and the three acceleration components corresponding to each sampling point can be obtained. Then, based on the acceleration components of the sampling point, it is determined whether a sampling point is a valid sampling point. Figure 3 As shown, Figure 3 This is a schematic diagram of a sampling space provided in an embodiment of the present application. A sampling space includes multiple sampling points arranged in sequence according to sampling time. The sampling duration of each sampling space is equal, for example, 2s.

[0049] For each sampling point, if at least two acceleration components of a sampling point exceed the acceleration threshold, the sampling point is determined to be a valid sampling point. Furthermore, whether the vibration signal is a valid signal can be determined based on the number of valid sampling points.

[0050] For example, when the values ​​of more than two acceleration components in the x, y, and z axes are greater than 0.25 m / s2, the number of valid sampling points among the 150 sampling points collected within 1000 milliseconds is determined. If the number of valid sampling points exceeds a preset value or exceeds a preset ratio, for example, the number of valid sampling points exceeds 120, or the proportion exceeds 80%, the vibration signal is considered to be a valid signal; otherwise, it is considered to be an invalid signal.

[0051] In one embodiment, based on the acceleration characteristics of the vibration sensor, the frequency domain characteristics and time domain characteristics of the vibration signal can also be used to comprehensively judge whether the vibration signal is a valid signal. For example, the center of gravity frequency, frequency standard deviation, and frequency area in the frequency domain characteristics, and the waveform index, pulse index, peak value, vibration duration, etc. in the time domain characteristics can be used to comprehensively judge whether a vibration signal is a valid signal. By using the frequency domain characteristics, time domain characteristics of the vibration signal and the acceleration characteristics of the vibration sensor, the judgment of the valid signal can be made more accurate.

[0052] In one embodiment, a specific voice signal emitted by the user (i.e., a voice signal containing a wake-up word) is typically used as the wake-up signal for the voice recognition device to perform voice recognition. In this case, since the user's speech has very few low-frequency bands and very few high-frequency bands, the judgment condition for a valid signal can be set as follows: the vibration signal has a common frequency of normal human speech, and the length of the vibration signal does not exceed the pronunciation duration corresponding to the wake-up word. In this case, the vibration signal is a valid voice signal.

[0053] At this time, the processor can maintain its normal working state and send an activation signal to the voice wake-up module in the smart speaker, so that the voice wake-up module in the smart speaker can enter the normal working state and process the aforementioned voice signals (including voice recognition, human-computer dialogue, etc.), and then realize voice interaction with the user based on the processor, voice wake-up module and even the server.

[0054] In the voice wake-up scenario, a person usually speaks at a speed of 100 to 200 words per minute, or about one word per 0.45 seconds. Generally, the voice wake-up words are more than two words, so the duration of an effective vibration signal should usually not be less than 1 second. At the same time, excluding the bass and treble parts, the frequency of normal human speech is approximately between [80 Hz, 400 Hz], and the vibration of the sound signal generated at this time is not large.

[0055] Therefore, the preset conditions can be set as follows: the frequency of the vibration signal is between [80HZ, 400HZ], the duration of the vibration signal is not less than 1 second, and at least two acceleration components corresponding to the vibration sensor in three directions exceed 0.25m / s2. When the vibration signal meets the aforementioned preset conditions, it can be determined that the vibration signal is a valid signal. Furthermore, the processor can send an activation signal to the voice wake-up module to activate the voice wake-up module to exit the sleep state and perform subsequent voice recognition.

[0056] Under this implementation, the smart speaker and processor can all enter a sleep state when the driver or passenger does not need them. Only when the vibration sensor detects a valid voice signal will the processor maintain a normal working state and wake up the corresponding function of the smart speaker. If the detected voice signal is an invalid signal (in fact, the driver and passengers may often have occasional communication), the voice wake-up module in the smart speaker will not be woken up, and the processor will re-enter the sleep state, thereby significantly saving power consumption.

[0057] In one embodiment, before the processor determines whether the vibration signal is a valid signal, it can also perform noise filtering on the vibration signal, that is, perform spectrum analysis on the vibration signal, and treat the vibration signal whose amplitude does not exceed a preset amplitude and / or whose vibration time does not exceed a preset duration as noise. For example, in the scenario of voice wake-up, the vibration signal with an amplitude not exceeding 1mm and a vibration time not exceeding 0.2s is regarded as noise and filtered out, thereby filtering out the jitter in the vibration signal, which is conducive to improving the recognition accuracy of the valid signal in the subsequent steps. For example, it can be Figure 3 The vibration signal in the image is filtered out accordingly to remove the background noise signal with smaller amplitude and relatively flat in the image, so as to improve the subsequent recognition accuracy.

[0058] In a second aspect of the embodiments of the present application, an electronic device is further provided, such as Figure 4 As shown, Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The specific embodiment of the present application does not limit the specific implementation of the electronic device.

[0059] like Figure 4 As shown, the electronic device may include: a gyro sensor 1000 and a processor 1002 The gyro sensor 1000 senses a vibration signal generated in the environment and generates an activation signal;

[0060] Processor 1002 receives the activation signal generated by the gyroscope sensor, changes the sleep state to the working state, and determines the characteristics corresponding to the vibration signal; determines whether the vibration signal is a valid signal based on the characteristics corresponding to the vibration signal; if so, maintains the normal working state; otherwise, re-enters the sleep state.

[0061] The voice wake-up module 1003 performs voice recognition after being woken up from the dormant state.

[0062] The electronic device provided in the embodiments of the present application can be used as an accessory in various other entities, such as mobile phones, smart home appliances, smart vehicles, etc.

[0063] The electronic device may further include other related components for implementing the aforementioned functions, such as a communication interface 1004 , a memory 1006 , and a communication bus 1008 .

[0064] in:

[0065] The voice wake-up module 1001 , the processor 1002 , the voice wake-up module 1003 , the communication interface 1004 , and the memory 1006 communicate with each other via the communication bus 1008 .

[0066] The communication interface 1004 is used to communicate with other electronic devices or servers.

[0067] The processor 1002 is configured to execute the program 1010 , and specifically may execute the relevant steps in the above-mentioned processor wake-up method embodiment.

[0068] Specifically, the program 1010 may include program codes, which include computer operation instructions.

[0069] Processor 1002 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application. The one or more processors included in the smart device may be processors of the same type, such as one or more CPUs, or may be processors of different types, such as one or more CPUs and one or more ASICs.

[0070] The memory 1006 is used to store the program 1010. The memory 1006 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0071] The program 1010 can be specifically used to enable the processor 1002 to perform operations corresponding to the processor wake-up method in the first aspect.

[0072] The specific implementation of each step in program 1010 can refer to the corresponding description of the corresponding steps and units in the above-mentioned processor wake-up method embodiment, and will not be repeated here. Those skilled in the art will clearly understand that for the convenience and brevity of description, the specific working process of the above-described devices and modules can refer to the corresponding process description in the above-mentioned method embodiment, and will not be repeated here.

[0073] The present application discloses TS8, a speech recognition device, comprising:

[0074] A vibration sensor, which senses a vibration signal generated in the environment and generates an activation signal;

[0075] a processor that receives an activation signal generated by the vibration sensor, changes the dormant state to an active state, determines a signal characteristic of the vibration signal, and wakes up a voice wake-up module in a dormant state if the vibration signal is determined to be a valid signal based on the signal characteristic of the vibration signal; otherwise, restores the processor to a dormant state;

[0076] The voice wake-up module performs voice recognition after being awakened from the dormant state.

[0077] TS9. In the speech recognition device as described in TS8, the processor obtains three acceleration components corresponding to the acceleration characteristics of the vibration sensor in three directions of a preset space; when at least two acceleration components exceed an acceleration threshold, the vibration signal is determined to be a valid signal.

[0078] TS10. The speech recognition device as described in TS9, wherein the processor determines that the vibration signal is a valid signal when the frequency of the vibration signal is between [80 Hz, 400 Hz], the duration of the vibration signal is not less than 1 second, and at least two acceleration components corresponding to the vibration sensor in three directions exceed 0.25 m / s2.

[0079] In a third aspect of the embodiments of the present application, a transport device is further provided, on which the voice recognition device as described above is configured.

[0080] The embodiment of the present application further discloses TS11, a transport device, on which a speech recognition device as described in any one of TS8 to TS10 is configured.

[0081] For example, a smart speaker including a voice wake-up module can be configured in a car. The car's control center can provide the functions of the processor of the voice recognition device (that is, the car's control center serves as the processor of the voice recognition device), and the gyroscope contained in the car serves as the vibration sensor of the voice recognition device.

[0082] When the user gets out of the car or temporarily sets it to sleep mode, the control center enters sleep mode, and the gyroscope on the car can sense the sound signals generated by the surrounding environment, such as opening and closing the car door, the voice signals generated by the user speaking, etc., and generate an activation signal, so that the control center can determine whether it is necessary to wake up the voice wake-up module in the smart speaker.

[0083] In a fourth aspect of the embodiments of the present application, a computer storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the wake-up method of the voice recognition device as described in the first aspect is implemented.

[0084] According to a fifth aspect of an embodiment of the present application, a computer program product is provided, comprising computer instructions, wherein the computer instructions instruct a computing device to perform operations corresponding to the wake-up method of the speech recognition device as described in the first aspect.

[0085] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory media such as modulated data signals and carrier waves.

[0086] Through the description of the above implementation methods, it can be known that those skilled in the art can clearly understand that the embodiments of this specification can be implemented by means of software plus a necessary general hardware platform. Based on this understanding, the technical solutions of the embodiments of this specification, or the part that contributes to the prior art, can be embodied in the form of a software product, which can be stored in a storage medium such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or certain parts of the embodiments of this specification.

[0087] The systems, methods, modules, or units described in the above embodiments may be implemented by a computer chip or entity, or by a product having certain functions. A typical implementation device is a computer, which may be in the form of a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email transceiver, game console, tablet computer, wearable device, or any combination of these devices.

[0088] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the method embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The method embodiment described above is merely illustrative, wherein the modules described as separate components may or may not be physically separated, and the functions of each module can be implemented in the same one or more software and / or hardware when implementing the embodiment scheme of this specification. It is also possible to select some or all of the modules according to actual needs to achieve the purpose of the embodiment scheme. A person of ordinary skill in the art can understand and implement it without paying any creative work.

[0089] The above is only a specific implementation of the embodiment of this specification. 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 embodiment of this specification. These improvements and modifications should also be regarded as the scope of protection of the embodiment of this specification.

Claims

1. A method for waking up a speech recognition device, the speech recognition device comprising: Vibration sensor, processor and voice wake-up module, the method includes: When the processor receives an activation signal sent by the vibration sensor in response to a vibration signal sensed by the vibration sensor, the processor changes the dormant state to the working state; Determining a signal characteristic of the vibration signal; the signal characteristic including an acceleration characteristic of the vibration sensor; If the vibration signal is determined to be a valid signal according to the signal characteristics of the vibration signal, waking up the voice wake-up module in a dormant state to perform voice recognition; otherwise, restoring the processor to a dormant state; Wherein, determining that the vibration signal is a valid signal according to the signal characteristics of the vibration signal includes: Obtaining three acceleration components corresponding to the acceleration characteristics of the vibration sensor in three directions of a preset space; When at least two acceleration components exceed an acceleration threshold, determining that the vibration signal is a valid signal includes: Determine a sampling space corresponding to when at least two acceleration components exceed an acceleration threshold, and obtain a plurality of sampling points in the sampling space and three acceleration components corresponding to each sampling point; When at least two acceleration components of a sampling point exceed the acceleration threshold, determining the sampling point as a valid sampling point; When the number of valid sampling points among the plurality of sampling points exceeds a preset value, the vibration signal is determined to be a valid signal.

2. The method according to claim 1, wherein The processor determines a signal characteristic of the vibration signal, including: The processor determines frequency domain characteristics and time domain characteristics of the vibration signal; The frequency domain characteristics, the time domain characteristics of the vibration signal and the acceleration characteristics of the vibration sensor are determined as signal characteristics of the vibration signal.

3. The method according to claim 2, wherein: Determining, according to a signal feature of the vibration signal, that the vibration signal is a valid signal includes: When the frequency of the vibration signal is between [80 Hz, 400 Hz], the duration of the vibration signal is not less than 1 second, and at least two acceleration components corresponding to the vibration sensor in three directions exceed 0.25 m / s 2 , it is determined that the vibration signal is a valid signal.

4. The method according to claim 1, wherein Before determining that the vibration signal is a valid signal, the method further includes: Noise signals whose amplitude does not exceed a preset amplitude and / or whose vibration duration does not exceed a preset duration are filtered out from the vibration signal.

5. A speech recognition device, comprising: A vibration sensor, which senses a vibration signal generated in the environment and generates an activation signal; The processor receives the activation signal generated by the vibration sensor and changes the dormant state to the working state; determines the signal characteristics of the vibration signal; the signal characteristics include the acceleration characteristics of the vibration sensor; if the vibration signal is determined to be a valid signal based on the signal characteristics of the vibration signal, wakes up the voice wake-up module in the dormant state; otherwise, restores the processor to the dormant state; wherein, to determine the signal characteristics of the vibration signal, the processor is further configured to: Obtaining three acceleration components corresponding to the acceleration characteristics of the vibration sensor in three directions of a preset space; When at least two acceleration components exceed an acceleration threshold, determining that the vibration signal is a valid signal; The voice wake-up module performs voice recognition after being awakened from the dormant state.

6. A computer storage medium having a computer program stored thereon, wherein when the program is executed, the wake-up method of the speech recognition device according to any one of claims 1 to 4 is implemented.

7. A computer program product, comprising computer instructions, wherein the computer instructions instruct a computing device to execute operations corresponding to the wake-up method for a speech recognition device according to any one of claims 1 to 4.

Images