Speech recognition optimization method and home appliance device

Abstract

The application provides a voice recognition optimization method and a household appliance, comprising: in response to detecting an audio signal meeting a preset trigger condition, determining whether the audio signal meets a preset voice recognition condition; if the audio signal does not meet the preset voice recognition condition, obtaining a current working state parameter of the household appliance; according to a preset hierarchical adjustment strategy, determining a target power adjustment amplitude corresponding to the current working state parameter; controlling a working component to reduce output power according to the target power adjustment amplitude, so as to reduce environmental noise; in a state where the working component is at the reduced output power, collecting a voice instruction and performing voice recognition on the voice instruction to obtain a voice recognition result. In this way, when a voice interaction demand is detected but the environmental noise is too large, the working component of the device is actively controlled to reduce the output power, which can improve the voice recognition rate of the high-noise household appliance in a high working state, and ensures the accuracy and stability of user interaction.

Description

Technical Field

[0001] This application relates to the field of smart home technology, and in particular to a voice recognition optimization method and home appliances. Background Technology

[0002] With the popularization of smart home technology, voice interaction functions have been widely used in various home appliances such as range hoods and blenders. However, when these devices are operating under high load conditions such as high-power smoke extraction or high-speed blending, their internal working components (such as high-power motors or fans) will generate extremely high levels of environmental noise, resulting in a sharp decrease in the signal-to-noise ratio.

[0003] Current solutions primarily rely on multi-microphone arrays combined with beamforming algorithms for signal enhancement, or on increasing chip computing power for noise suppression after wake-up detection. However, multi-microphone solutions significantly increase device hardware costs and standby power consumption, and the optimization effect of relying solely on backend signal processing algorithms in high-noise scenarios has reached its limit, making it difficult to avoid a precipitous drop in speech recognition rate. Furthermore, existing solutions often suffer from the logical contradiction of the wake-up command failing first; that is, the wake-up word itself cannot be recognized under extremely high noise, causing the subsequent noise reduction process to fail to start. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide a speech recognition optimization method and a home appliance. By actively controlling the working components of the device to reduce the output power according to a preset graded adjustment strategy when a speech interaction demand is detected but the ambient noise is too high, the environmental noise generated by the device operation can be physically reduced from the source, thereby improving the signal-to-noise ratio of the speech signal collected by the sound acquisition component. This improves the speech recognition rate of high-noise home appliances in high-power operation mode and ensures the accuracy and stability of user interaction.

[0005] In a first aspect, the present invention provides a speech recognition optimization method applied to home appliances, the home appliances including working components that generate environmental noise during operation; the method includes: In response to the detection of an audio signal that meets the preset trigger conditions, it is determined whether the audio signal meets the preset speech recognition conditions.

[0006] If the audio signal does not meet the preset voice recognition conditions, obtain the current working status parameters of the home appliance.

[0007] Based on the preset graded adjustment strategy, determine the target power adjustment range corresponding to the current operating status parameters.

[0008] The control unit adjusts its output power according to the target power level to reduce environmental noise.

[0009] When the working component is in a state of reduced output power, voice commands are collected and voice recognition is performed on the voice commands to obtain the voice recognition results.

[0010] In an optional implementation, after obtaining the speech recognition result, the method further includes: Determine whether the speech recognition results contain multi-turn dialogue intent.

[0011] If the speech recognition result does not contain a multi-turn dialogue intent, or if no subsequent audio input is detected within the preset waiting time, the control unit increases the output power at a preset recovery slope until it is restored to the output power before the reduction.

[0012] In an optional implementation, prior to the step of detecting an audio signal that satisfies a preset trigger condition, the method further includes: The sound acquisition component controlling the home appliances is in a real-time detection state to collect ambient sound signals.

[0013] Calculate the real-time energy value of the ambient sound signal.

[0014] Determine whether the real-time energy value exceeds the preset energy threshold, and whether the duration of exceeding the preset energy threshold reaches the preset time threshold.

[0015] If the real-time energy value exceeds the preset energy threshold, and the duration of exceeding the preset energy threshold reaches the preset time threshold, the ambient sound signal will be determined as an audio signal that meets the preset triggering conditions.

[0016] In an optional implementation, in response to detecting an audio signal that meets a preset trigger condition, determining whether the audio signal meets a preset speech recognition condition includes: Acquire audio signals and extract spectral features.

[0017] Calculate the matching degree between the spectral features and the preset human voice feature library, and determine whether the matching degree is higher than the preset matching threshold.

[0018] If the matching degree is higher than the preset matching threshold, the audio signal is determined to be a valid human voice signal.

[0019] Calculate the current signal-to-noise ratio (SNR) of the valid human voice signal and determine whether the current SNR is lower than the preset SNR threshold.

[0020] If the current signal-to-noise ratio is lower than the preset signal-to-noise ratio threshold, it is determined that the audio signal does not meet the preset speech recognition conditions.

[0021] In an optional implementation, the current operating status parameters include at least one of the current operating level of the home appliance and the current ambient noise level.

[0022] The steps for determining the target power adjustment range corresponding to the current operating state parameters based on a preset graded adjustment strategy include: Identify the noise intensity level corresponding to the current operating status parameters.

[0023] If the current operating status parameter is identified as corresponding to the first noise intensity level, the target power adjustment range is determined to be the first ratio.

[0024] If the current operating status parameter is identified as corresponding to the second noise intensity level, the target power adjustment range is determined to be the second ratio.

[0025] The higher the current operating level or the greater the current ambient noise level, the higher the corresponding noise intensity level; the second noise intensity level is higher than the first noise intensity level, and the second ratio is greater than the first ratio.

[0026] In an optional implementation, the step of controlling the working component to reduce the output power according to the target power adjustment range includes: Based on the target power adjustment range, calculate the target operating parameters of the working components.

[0027] Based on a preset smooth transition strategy, the actual operating parameters of the working components are controlled to change towards the target operating parameters until the target operating parameters are reached; wherein, the preset smooth transition strategy limits the power change rate of the working components to be less than or equal to a preset change rate threshold.

[0028] In an optional implementation, before the step of controlling the working component to reduce the output power according to the target power adjustment range, the method further includes: Obtain the operating safety status parameters of home appliances; the operating safety status parameters should include at least the temperature data and load data of the working components.

[0029] Determine whether the operating safety status parameters are within the preset safety protection range.

[0030] If the operating safety status parameters are within the preset safety protection range, operations that reduce output power are prohibited, and the current output power of the working components is maintained unchanged.

[0031] In an optional implementation, the step of acquiring voice commands includes: Monitor whether the output power of the working components has decreased to the target power value.

[0032] If the power level is reduced to the target value, the control appliance will issue an interactive prompt signal; the interactive prompt signal is used to guide the user to start inputting voice commands.

[0033] Within a preset time window after issuing the interactive prompt signal, the sound acquisition component of the home appliance is activated to collect voice commands.

[0034] In a second aspect, the present invention provides a home appliance, comprising: a home appliance body, a controller and a working component respectively disposed inside the home appliance body; the working component is communicatively connected to the controller; the controller is configured to execute a voice recognition optimization method as described in any of the foregoing embodiments.

[0035] In an optional implementation, the home appliance also includes a sound acquisition component that is communicatively connected to the controller.

[0036] This application provides a voice recognition optimization method and a home appliance. In response to the detection of a human voice signal that does not meet the recognition conditions, a preset graded power adjustment strategy is executed based on the device's current operating state parameters. This proactively controls the working components to temporarily reduce output power, thereby physically reducing environmental noise at the sound source and improving the signal-to-noise ratio of voice commands under high-power conditions. This solves the problem of a sharp drop in voice recognition rate and wake-up failure in high-noise scenarios. Simultaneously, by combining spectral feature matching, signal-to-noise ratio judgment, and a multi-turn dialogue detection mechanism, accurate recognition of valid human voice intent and automatic power recovery are achieved, avoiding unnecessary power consumption and long-term damage to device functions caused by accidental touches due to pure noise. Furthermore, with the addition of smooth transition control, operational safety verification, and interactive timing prompts, abnormal noises and safety hazards caused by sudden changes in device state are prevented, and the problem of command loss caused by speed reduction delays is solved. Thus, without increasing the cost of multi-microphone arrays, a dynamic balance is achieved between core device functions, operational safety, and a high-quality voice interaction experience.

[0037] Other features and advantages of this application will be set forth in the following description and will be apparent in part from the description or may be learned by practicing the application. The objectives and other advantages of this application are realized and obtained through the structures particularly pointed out in the description, claims and drawings.

[0038] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0040] Figure 1 This is a flowchart of a speech recognition optimization method provided in an embodiment of this application; Figure 2This is a flowchart of the output power recovery method provided in an embodiment of this application; Figure 3 A flowchart of an audio signal determination method that satisfies preset triggering conditions provided in an embodiment of this application; Figure 4 A flowchart of the preset speech recognition condition judgment provided in the embodiments of this application; Figure 5 A flowchart of the target power adjustment range determination method provided in the embodiments of this application; Figure 6 This is a flowchart of a method for reducing output power provided in an embodiment of this application; Figure 7 This application provides a flowchart for determining the safety status of home appliances in an embodiment. Figure 8 This is a flowchart of a method for collecting voice commands provided in an embodiment of this application; Figure 9 A schematic diagram of a home appliance provided in an embodiment of this application.

[0041] Icons: 1-Home appliance body; 2-Controller; 3-Working part; 4-Sound acquisition part. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0043] To help those skilled in the art better understand this application, a brief introduction to its application scenarios and design concepts is provided.

[0044] Smart home appliances such as range hoods, blenders, and vacuum cleaners often generate noise levels as high as 75dB or even 85dB when operating at high or ultra-high power settings, posing a serious challenge to voice interaction.

[0045] Existing technologies primarily address noise by adding features, such as using multi-microphone arrays with beamforming algorithms, or increasing chip computing power for noise suppression after wake-up detection. However, multi-microphone array solutions increase hardware costs by 30%-50%, and high-performance computing modules can increase device standby power consumption from 1W to over 3W, which does not align with the design trend of low-cost, low-power home appliances.

[0046] When the ambient noise exceeds 75dB, the effect of simply relying on the backend algorithm to separate human voice from noise reaches its limit. The accuracy of speech recognition will drop sharply from 85% in low-noise environments to below 40%, which cannot meet the needs of normal interaction.

[0047] Existing solutions for boosting computing power after wake-up rely on the device successfully recognizing the wake word first. However, in noisy environments, the recognition success rate of the wake word itself is often less than 60%, causing subsequent computing power boosts and noise reduction processes to fail to be triggered, resulting in a logical dead loop where the wake-up command fails first.

[0048] Based on this, this application provides a speech recognition optimization method and a home appliance. When a potential voice interaction need is detected, the output power of the device's noise source (such as a motor or fan) is actively and temporarily reduced through control logic, directly reducing the environmental noise level at the physical level. The noise level is then automatically restored after recognition is completed. By physically suppressing the noise source, this application can stably improve the speech recognition accuracy in high-noise scenarios to over 85%, and the wake-up success rate to over 90%. Furthermore, this application does not require expensive multi-microphone arrays and high-computing-power chips; it can be implemented using only a single microphone and basic algorithms, significantly reducing hardware costs, and keeping standby power consumption below 1W.

[0049] To facilitate understanding of this embodiment, the embodiments of this application will be described in detail below.

[0050] This application provides a speech recognition optimization method, referring to... Figure 1 The speech recognition optimization method provided in this application embodiment is applied to home appliances, which include working parts that generate environmental noise during operation.

[0051] Here, the household appliance includes a working component that generates environmental noise during operation (e.g., a fan in a range hood, a motor in a blender, a motor in a vacuum cleaner, etc.), a sound acquisition component for collecting sound (e.g., a single microphone or a microphone array), and a controller. The controller is communicatively connected to the working component and the sound acquisition component to perform the following method, which includes: Step S101: In response to detecting an audio signal that meets the preset triggering conditions, determine whether the audio signal meets the preset speech recognition conditions.

[0052] Here, home appliances are typically in low-power monitoring mode. Preset trigger conditions are used to initially filter valid sounds in the environment, preventing invalid sounds from frequently waking up the device.

[0053] Preset trigger conditions can be set based on sound energy and duration. For example, the sound acquisition component monitors ambient sound in real time. When the energy value of the detected sound signal is greater than or equal to a preset energy threshold (e.g., 45dB), and the duration of the high-energy state reaches a preset time threshold (e.g., 100ms), it is determined that an audio signal meeting the preset trigger conditions has been detected. To adapt to different home environments, the preset energy threshold can be dynamically calibrated. That is, the controller periodically acquires a baseline value of ambient background noise and adds a fixed number of decibels to the baseline value as the current preset energy threshold.

[0054] In addition to energy detection, preset trigger conditions can also incorporate multimodal information. For example, in devices equipped with cameras, visual detection of a user approaching or looking at the device can be used as an auxiliary trigger condition. Alternatively, human body sensing signals from infrared or ultrasonic sensors can be used as trigger conditions.

[0055] After detecting the audio signal, it is necessary to further determine whether it is suitable for direct speech recognition, that is, whether it meets the preset speech recognition conditions, in order to distinguish human voice from pure noise, and to determine whether the human voice is clear.

[0056] Specifically, the spectral features of the audio signal are extracted and compared with a pre-set human voice spectral feature database. This database contains typical fundamental frequency ranges (e.g., 85Hz-300Hz) and harmonic ranges (e.g., 2kHz-4kHz) for different ages and genders. The matching degree is calculated. If the matching degree is lower than a preset matching threshold (e.g., 30%), it is determined to be pure noise (such as wind noise or mechanical noise) and is ignored. If the matching degree is higher than the preset matching threshold (e.g., 60%), it is determined to be a valid human voice signal.

[0057] Calculate the current signal-to-noise ratio (SNR) of the valid human voice signal. The preset speech recognition condition is typically set to an SNR greater than or equal to a preset SNR threshold (e.g., 5dB). If the current SNR is lower than this threshold, it indicates that the environment is too noisy, making it difficult for the speech recognition engine to accurately interpret the command. In this case, it is determined that the preset speech recognition condition is not met, and noise reduction processing is required.

[0058] Step S102: If the audio signal does not meet the preset voice recognition conditions, obtain the current working status parameters of the home appliance.

[0059] Here, when it is determined that the speech recognition environment needs to be optimized, the first step is to understand the current operating status of the equipment. The current operating status parameters are indicators that reflect the equipment's operating load and noise level.

[0060] Current operating status parameters can include the current operating level of the home appliance, such as the logical levels of a range hood: low, high, and extra high. Different levels correspond to preset motor speeds and known noise baselines.

[0061] Current operating status parameters can also include the current ambient noise level of the home appliance. This can be obtained in real time through the device's built-in noise sensor, or by estimating the energy in non-human voice frequency bands using a sound acquisition component. For example, directly reading the current sound pressure level value (e.g., 75dB, 82dB).

[0062] In addition, the current operating status parameters may also include physical parameters such as the real-time rotational speed of the working components, drive current, and PWM (pulse width modulation) duty cycle, which can also indirectly reflect the current noise level.

[0063] Step S103: Determine the target power adjustment range corresponding to the current working state parameters according to the preset graded adjustment strategy.

[0064] Here, the preset hierarchical adjustment strategy can be pre-stored in the controller's memory, manifested as a lookup table, mapping curve, or calculation formula.

[0065] Specifically, it includes: 1. Adjustment based on power setting: If the current operating status parameters indicate that the device is at the high setting (corresponding to a noise level of approximately 75-80dB), determine the target power adjustment range as a reduction of 15%-20% of the current power. If it is at the ultra-high setting (corresponding to a noise level of approximately 80-85dB), determine the target power adjustment range as a reduction of 20%-25% of the current power.

[0066] 2. Noise Level-Based Adjustment: Divide the current ambient noise level into different noise intensity ranges. If the noise level is in the first range (e.g., 75dB-80dB), set the target power adjustment range to the first percentage; if the noise level is in the second range with higher intensity (e.g., above 80dB), set the target power adjustment range to the second percentage, which is larger.

[0067] 3. Adjustment based on continuous functions: The adjustment range is calculated using linear or nonlinear formulas, for example: target reduction = coefficient K × (current noise value - target identification noise threshold), to achieve stepless dynamic adjustment.

[0068] Step S104: Control the working component to reduce the output power according to the target power adjustment range in order to reduce environmental noise.

[0069] Here, to optimize the user experience and prevent sharp electromagnetic whistling or sudden wind noise from abrupt changes in motor speed, a smooth transition control strategy is preferred. The controller limits the rate of power change, for example, limiting the power change every 100 milliseconds to no more than 5% of the total power. This allows ambient noise to gently decrease to a suitable level for speech recognition (e.g., 65dB-70dB) within a short time (e.g., 300ms), at which point the signal-to-noise ratio will be significantly improved.

[0070] Before reducing power, a safety check is performed. The controller reads the device's operating safety status parameters (such as motor temperature, current load, and whether dense smoke or high heat is detected). If the device is in an emergency operating state (e.g., the range hood detects high temperatures due to dry burning or extremely high smoke concentration), for safety reasons, the power reduction operation can be prohibited, prioritizing smoke extraction or heat dissipation functions.

[0071] Step S105: With the working component in a reduced output power state, collect voice commands and perform voice recognition on the voice commands to obtain voice recognition results.

[0072] Here, once the power of the working component is reduced and stabilized at the target level, the ambient noise is physically suppressed. At this time, the sound acquisition component collects the user's voice commands (such as turning on the lights or reducing the airflow).

[0073] To improve the success rate of interaction and solve the problem of users interrupting due to the delay caused by the power reduction, after confirming that the power has been reduced to the appropriate level, the home appliance can be controlled to issue an interactive prompt signal (such as playing a "beep" prompt sound or flashing an indicator light) to guide the user to start speaking after the prompt signal.

[0074] The collected voice commands are sent to a speech recognition engine for processing. This engine can be a local offline recognition module, a cloud-based recognition service, or a combined edge-cloud solution. Because physical noise at the front end has been suppressed, the signal-to-noise ratio of the voice commands meets the recognition requirements.

[0075] In addition, after receiving the voice recognition result, or after the preset timeout period, the controller should control the working parts to restore the output power to the level before the reduction, so as to ensure that the home appliance can continue to complete its original working task (such as continuing to powerfully exhaust smoke).

[0076] In one embodiment, reference is made to Figure 2 After obtaining the speech recognition result in step S105, the method further includes the following steps S201-S202.

[0077] Step S201: Determine whether the speech recognition result contains multi-turn dialogue intent.

[0078] Here, the controller performs semantic analysis on the speech recognition results. Multi-turn dialogue intent refers to the voice command indicating that the user may have subsequent command input, or that the current command requires further parameter confirmation.

[0079] Specifically, if the speech recognition result is an incomplete sentence (e.g., interrupted after “adjust the fan speed to…”), or contains a question tone (e.g., “What is the current oil temperature?”), it is determined that it contains a multi-turn dialogue intent.

[0080] If the voice recognition result triggers a specific interaction mode of the device, such as "enter recipe navigation mode" or "start multi-step cooking assistance", it will default to entering a multi-turn dialogue state and determine that it contains a multi-turn dialogue intent.

[0081] If the voice recognition result only contains a wake word (such as "Hello range hood") without a specific control command, this usually means that the user is about to issue a command, and therefore it is judged to contain a multi-turn dialogue intent. In this case, the power should not be restored immediately.

[0082] In step S202, if the speech recognition result does not contain a multi-turn dialogue intent, or if no subsequent audio input is detected within a preset waiting time, the control unit increases the output power at a preset recovery slope until it is restored to the output power before the reduction.

[0083] Here, if the voice recognition result is clearly a single control command (such as turning on the lights or turning off the display) and does not contain any subsequent intent, the power recovery process is immediately triggered to shorten the time the device is in a low suction state.

[0084] To prevent misjudgment or user abandonment of the interaction midway, a preset waiting time (e.g., 2 seconds) is set. If the sound acquisition component does not detect a new valid audio input (or does not detect a human voice) within 2 seconds after the previous recognition is completed, the interaction process is forcibly terminated, triggering power recovery.

[0085] Furthermore, as a feasible implementation variation, if the meaning of the voice recognition result itself is to adjust the gear of the working component (for example, the user says "adjust to the low gear" in the high gear), then the action of "restoring the output power before the reduction" can be replaced or modified by the specific execution of the gear change command, that is, the controller directly adjusts the working component to the new gear requested by the user's voice command.

[0086] In one embodiment, reference is made to Figure 3 Before the step of detecting an audio signal that meets the preset trigger condition in step S101, the method further includes the following steps S301-S304.

[0087] Step S301: Control the sound acquisition component of the home appliance to be in real-time detection state to acquire ambient sound signals.

[0088] Here, home appliances (such as range hoods) are typically in normal operation or standby mode. To achieve low-power operation, the sound acquisition component (such as a single microphone module) works in low-power mode in conjunction with the voice activity detection module. In this mode, only the most basic sound path is kept open, and high-energy-consuming complex signal processing algorithms are not activated, so that the overall power consumption of the module can be controlled at an extremely low level, such as 0.5W or even lower, thereby meeting the energy-saving and environmental protection requirements of home appliances.

[0089] Step S302: Calculate the real-time energy value of the ambient sound signal.

[0090] Here, the sound acquisition component continuously receives sounds from the external environment, i.e., ambient sound signals. The voice activity detection module samples the acquired ambient sound signals in real time and calculates their energy level (e.g., decibel value in dB).

[0091] Step S303: Determine whether the real-time energy value exceeds the preset energy threshold, and whether the duration of exceeding the preset energy threshold reaches the preset time threshold.

[0092] Here, a preset energy threshold is used to determine whether the sound is loud enough. This preset energy threshold is not fixed but dynamically calibrated based on the current background noise baseline of the appliance. For example, the current standby noise level of the device is monitored, and the preset energy threshold is set to be a certain margin higher than this standby noise level (e.g., 5-10 dB higher). In a typical embodiment, the preset energy threshold can be set to 45 dB.

[0093] A preset time threshold is used to determine whether the sound is continuous, because human speech commands usually have a certain duration, while transient noise is typically very short. For example, the preset time threshold can be set to 100ms. It detects whether the energy of the ambient sound signal remains above 45dB for a continuous 100ms.

[0094] Step S304: If the real-time energy value exceeds the preset energy threshold and the duration of exceeding the preset energy threshold reaches the preset time threshold, the ambient sound signal is determined as an audio signal that meets the preset triggering conditions.

[0095] Here, when an ambient sound signal simultaneously meets both of the above conditions, it is determined that the signal is highly likely to be a voice command issued by the user, i.e., it is identified as a potential voice interaction request. At this time, the ambient sound signal is marked as an audio signal that meets the preset triggering conditions.

[0096] Once the triggering conditions are met, the system immediately exits low-power mode, activates the high-precision detection mode of the voice activity detection module, and wakes up the backend voice recognition module to enter the recognition state, enabling further finer spectral analysis and human voice confirmation of the audio signal. This tiered wake-up mechanism ensures fast response to user commands while effectively avoiding false wake-ups and unnecessary power consumption caused by transient noise in the environment (such as clapping or collision sounds).

[0097] In one embodiment, reference is made to Figure 4 Step S101 includes the following steps S401-S405.

[0098] Step S401: Acquire audio signals and extract spectral features.

[0099] Here, after being awakened, the sound acquisition unit will continuously acquire audio signals, and the voice activity detection module will switch from low-power mode to high-precision detection mode. In high-precision detection mode, the voice activity detection module performs time-frequency analysis on the audio signal, for example, by extracting the spectral features of the audio signal through Fast Fourier Transform.

[0100] Spectral characteristics reflect the energy distribution of sound at different frequencies. Typically, human voice signals and mechanical noise signals differ significantly in the frequency domain. For example, the fundamental frequency range of human voice is usually concentrated between 85Hz and 300Hz, while the fundamental frequency of some mechanical noises (such as the high-frequency whistling of a fan) may be concentrated above 1000Hz.

[0101] Step S402: Calculate the matching degree between the spectral features and the preset human voice feature library, and determine whether the matching degree is higher than the preset matching threshold.

[0102] Here, the controller has a pre-installed voice feature library. The pre-installed voice feature library contains typical human voice spectrum data of users of different ages (children, adults, and the elderly) and genders. This data covers the core frequency range of human voices, specifically including the fundamental frequency range of 85Hz to 300Hz and the harmonic range of 2kHz to 4kHz.

[0103] A feature matching degree calculation algorithm is used to compare the spectral features of the audio signal acquired and extracted in real time with data in a preset human voice feature database, and calculate the degree of similarity between the two, i.e., the matching degree. A preset matching threshold is used to determine whether the signal has sufficient human voice features.

[0104] Step S403: If the matching degree is higher than the preset matching threshold, the audio signal is determined to be a valid human voice signal.

[0105] Here, the preset matching threshold is set to 60%. If the calculated matching degree is higher than or equal to 60%, and the duration of this high matching degree state reaches a certain length (e.g., greater than or equal to 150ms), then the audio signal is determined to contain a valid human voice component and is identified as a valid human voice signal.

[0106] Conversely, if the calculated matching degree is low (e.g., less than 30%), it indicates that although the audio signal has high energy, it does not conform to the spectral characteristics of human voice and is very likely pure environmental noise (such as simple wind noise or object collision sound). In this case, it is determined to be a false trigger, the voice activity detection module will return to low power mode, and the voice recognition module will return to standby mode.

[0107] Step S404: Calculate the current signal-to-noise ratio of the valid human voice signal and determine whether the current signal-to-noise ratio is lower than the preset signal-to-noise ratio threshold.

[0108] Here, after confirming that it is a human voice, the ratio of the energy of the effective human voice signal to the energy of the current background noise is calculated, which is the current signal-to-noise ratio.

[0109] The preset signal-to-noise ratio (SNR) threshold is set based on the performance benchmark of the speech recognition engine. In this embodiment, the preset SNR threshold is set to 5 dB. This is because in noisy environments with an SNR below 5 dB, conventional speech recognition algorithms often struggle to accurately extract speech features, resulting in a significant drop in recognition rate.

[0110] Step S405: If the current signal-to-noise ratio is lower than the preset signal-to-noise ratio threshold, determine that the audio signal does not meet the preset speech recognition conditions.

[0111] Here, if the calculated signal-to-noise ratio is below 5dB, it means that although the human voice exists, it is severely masked by environmental noise and cannot be heard clearly. In this case, the audio signal is determined not to meet the preset speech recognition conditions. This determination result will directly trigger the subsequent noise source power adjustment process, that is, generate a dynamic noise reduction command and send it to the noise source control module to request physical-level noise reduction support.

[0112] Conversely, if the current signal-to-noise ratio is higher than or equal to 5dB, it means that although there is noise in the environment, the human voice is relatively clear and the existing algorithm is sufficient to handle it. Therefore, it is determined that the conditions for speech recognition are met and speech recognition can be performed directly without controlling the working components to reduce power.

[0113] In one embodiment, the current operating status parameter includes at least one of the current operating level of the home appliance and the current ambient noise level.

[0114] Here, the current operating status parameters reflect the current operating load and noise level of the equipment, specifically including at least one of the current operating level of the home appliance (such as the setting of the range hood) and the current ambient noise value collected by the built-in sensor.

[0115] Reference Figure 5 Step S103 includes the following steps S501-S503.

[0116] Step S501: Identify the noise intensity level corresponding to the current working status parameters.

[0117] Here, after receiving the current operating status parameters, the controller matches them with a preset hierarchical strategy table to determine the noise level under which the device is currently operating.

[0118] If the current operating status parameter is the current ambient noise value, real-time noise data is collected using the built-in noise sensor, with a collection accuracy of ±1dB. The collected values ​​are then mapped to preset noise ranges. For example, the noise range of 75dB to 80dB is defined as the first noise intensity level, and the noise range of 80dB to 85dB is defined as the second noise intensity level.

[0119] If the current operating status parameter corresponds to the current operating level, the operating mode set by the user is directly identified. For example, the high-power mode of the range hood is identified as the first noise level, and the ultra-high-power mode is identified as the second noise level.

[0120] Step S502: If the current working state parameter is identified as corresponding to the first noise intensity level, the target power adjustment range is determined to be the first ratio.

[0121] Here, when the device is at the first noise level (e.g., in a high-power operating state, or when the real-time noise value is between 75dB and 80dB), it is determined that although the noise affects speech recognition, it can meet the interaction requirements without significantly sacrificing device performance.

[0122] At this point, the controller determines the target power adjustment range as the first proportion. Based on experimental data and equipment characteristics, the first proportion is preferably set to reduce the current output power by 15% to 20%. By implementing this adjustment, it is expected that the real-time noise level can be suppressed from 75-80dB to the range of 65-70dB. At this noise level, the human voice signal-to-noise ratio can typically be improved by more than 5dB.

[0123] Step S503: If the current working state parameter is identified as corresponding to the second noise intensity level, the target power adjustment range is determined to be the second ratio.

[0124] The higher the current operating level or the greater the current ambient noise level, the higher the corresponding noise intensity level; the second noise intensity level is higher than the first noise intensity level, and the second ratio is greater than the first ratio.

[0125] Here, when the device is at the second noise level (e.g., in the highest noise level, or when the real-time noise level is between 80dB and 85dB), it is determined that the ambient noise is extremely high and a greater suppression level is required to ensure that the voice commands are clearly captured.

[0126] At this point, the controller determines the target power adjustment range as the second ratio. The second ratio is preferably set to reduce the current output power by 20% to 25%. By implementing this larger adjustment, it is expected that the real-time noise level can be suppressed from 80-85 dB to the range of 68-72 dB.

[0127] The higher the current operating level or the greater the current ambient noise level, the higher the corresponding noise intensity level, and the greater the second ratio will be than the first ratio. This dynamic grading adjustment mechanism ensures that the voice recognition accuracy can be maintained at a stable high level (e.g., above 85%) under different operating conditions, while preserving the core working capabilities of the home appliance at the current operating level (e.g., smoke extraction capability) to the greatest extent possible.

[0128] In one embodiment, reference is made to Figure 6 Step S104 includes the following steps S601-S602.

[0129] Step S601: Calculate the target operating parameters of the working components based on the target power adjustment range.

[0130] Here, the controller translates the abstract power adjustment percentage into specific electrical control commands. Based on the determined target power adjustment range (e.g., a 20% reduction), the controller, in conjunction with the current electrical characteristics of the appliance, calculates the target operating parameters required for the working components to achieve that reduction.

[0131] These target operating parameters depend on the drive type of the working components. For a brushless DC motor using PWM control, the target operating parameter could be the target PWM duty cycle. For a variable frequency motor, it could be the target drive frequency. For a conventional motor, it could be the target drive voltage or current value. For example, if the current PWM duty cycle is 80% and the target power adjustment is a 20% reduction, the calculated target operating parameter might correspond to a 64% duty cycle.

[0132] Step S602: Based on a preset smooth transition strategy, control the actual operating parameters of the working component to change towards the target operating parameters until the target operating parameters are reached; wherein, the preset smooth transition strategy limits the power change rate of the working component to be less than or equal to a preset change rate threshold.

[0133] Here, after obtaining the target operating parameters, the controller does not directly abruptly change the current operating parameters to the target value. Instead, it controls the actual operating parameters to approach the target value along the trajectory through interpolation or step-by-step adjustment.

[0134] This process follows a pre-defined smooth transition strategy, the core of which lies in limiting the rate of change. Based on experimental verification and acoustic characteristic analysis, in order to prevent electromagnetic whistling, mechanical gear clashing noise, or sudden changes in air pressure within the duct from occurring during the deceleration process of working components (such as the range hood fan), the pre-defined smooth transition strategy limits the power change rate to a specific threshold. Specifically, the pre-defined change rate threshold is preferably set to a power change of no more than 5% per 100 milliseconds.

[0135] For example, if the power needs to be reduced from 100% to 80%, the controller will not issue the 80% command all at once. Instead, it will issue an intermediate command every 100 milliseconds (such as 95%, 90%, 85%, etc.), so that the power decreases smoothly in a step-like manner. Although a transition process is introduced, the response time of the entire power adjustment can still be controlled within 300 milliseconds due to the use of an efficient control algorithm.

[0136] Through this smooth transition control, the changes in the operating status of home appliances become gentle and continuous, so that the resulting changes in environmental noise appear to the user as a natural gradual decrease in volume, rather than an abrupt and precipitous change. This greatly reduces the user's perception of changes in the operating status of the equipment, thereby optimizing the voice recognition rate while ensuring a smooth and natural interactive experience.

[0137] In one embodiment, reference is made to Figure 7 Before the step of controlling the working component to reduce the output power according to the target power adjustment range in step S104, the method further includes the following steps S701-S703.

[0138] Step S701: Obtain the operating safety status parameters of the home appliance; the operating safety status parameters include at least the temperature data and load data of the working components.

[0139] Here, at the moment before executing the noise reduction action, the controller first reads the real-time data from the sensors inside the home appliance, namely the operating safety status parameters.

[0140] For working components (such as motors), temperature data is collected by NTC thermistors embedded in the motor windings or housing to monitor whether the motor is on the verge of overheating. Load data is collected by current sampling circuits or Hall effect sensors to monitor the motor's operating current, voltage, or torque, and to determine whether there is a stall or overload condition.

[0141] Furthermore, for specific types of home appliances, the operational safety status parameters can be expanded to include environmental safety data. For example, for range hoods, parameters may include real-time smoke concentration collected by smoke sensors or gas leak values ​​collected by gas sensors; for blenders, parameters may include liquid temperature within the blender or food resistance data.

[0142] Step S702: Determine whether the operating safety status parameters are within the preset safety protection range.

[0143] Here, the controller compares the collected parameters with preset safety thresholds. The preset safety protection range refers to the numerical range within which the device operates under abnormal conditions, high load limits, or emergency conditions. Within this range, the device's primary task is to maintain operation to ensure safety, rather than optimizing the user experience.

[0144] The specific judgment logic includes: Determine if the temperature of the working component is higher than the preset temperature warning value (e.g., 110°C). If it is higher than this value, the motor usually relies on the airflow generated by the high speed for self-cooling. Reducing the power may lead to insufficient heat dissipation and burn out the motor.

[0145] Determine if there are abnormal fluctuations in load data. For example, if the current value exceeds 120% of the rated current, or if the motor is found to be experiencing high resistance and stalling.

[0146] Taking a range hood as an example, determine whether the smoke concentration exceeds the preset fire warning threshold. If an extremely high concentration of smoke is detected (which may mean that the cookware has caught fire or that there is a peak in stir-frying), the equipment must maintain full speed or even exceed the frequency of smoke extraction, and the fan speed must not be reduced at this time.

[0147] In step S703, if the operating safety status parameters are within the preset safety protection range, the operation of reducing output power is prohibited, and the current output power of the working component remains unchanged.

[0148] Here, if any parameter falls into the preset safety protection range, the current device safety priority is determined to be higher than the voice interaction priority.

[0149] The controller will trigger interception logic, directly discarding or temporarily suspending the previously calculated target power adjustment range, and forcibly prohibiting the sending of power reduction commands to the operating components. The operating components will continue to maintain the current output power (i.e., maintain high or ultra-high speed operation) to ensure continuous heat dissipation, smoke extraction, or overcoming load resistance.

[0150] To avoid users mistakenly believing the device is unresponsive to voice commands (due to low recognition rates caused by the lack of noise reduction), while disabling power reduction, the controller can instruct the appliance to emit specific feedback signals. For example, it might play a prompt saying "Current operating load is too high; voice control is temporarily unavailable," or use a specific flashing pattern on the indicator light to inform the user that the device is in protection mode. The disabling mechanism will be lifted and normal voice recognition optimization will resume once the operating safety parameters have returned to normal from the preset safety protection range.

[0151] In one embodiment, reference is made to Figure 8 In step S105, the step of acquiring voice commands includes the following steps S801-S803.

[0152] Step S801: Monitor whether the output power of the working component has decreased to the target power value.

[0153] Here, after the controller issues a command to reduce power, it monitors the actual operating status of the working components in real time through an internal feedback loop.

[0154] Monitoring can be performed using either feedback signal detection or time estimation detection. In feedback signal detection, the controller reads real-time feedback parameters from the motor driver (such as the rotational speed from a Hall sensor and the current value from a current sampling circuit) to determine whether its actual operating parameters have stabilized within the calculated target operating parameter range. In time estimation detection, based on a preset smooth transition strategy, the physical response time required to reduce power from the current level to the target level is predicted. After issuing a power reduction command, the controller starts a timer; when the timer reaches the target response time, it is determined that the output power of the working component has been reduced to the target power value.

[0155] In step S802, if the power level is reduced to the target value, the control device issues an interactive prompt signal; the interactive prompt signal is used to guide the user to start inputting voice commands.

[0156] Here, once it is confirmed that the working component has stabilized in a low-power state, that is, the ambient noise has been physically suppressed to a level suitable for speech recognition (e.g., 65-70dB), the controller immediately triggers the interactive feedback logic.

[0157] The purpose of controlling home appliances to send interactive prompts is to establish a synchronized rhythm for human-computer interaction, clearly informing the user that "the device is ready and you can speak now."

[0158] Interactive prompts can take many forms, including but not limited to: Control the buzzer to emit a short "beep" sound, or play a specific prompt voice (such as "please speak") through the speaker.

[0159] The indicator lights, light rings, or displays on the control panel produce specific light effect changes (such as a flashing breathing light or a lit microphone icon).

[0160] This synchronization mechanism effectively solves the problem of users interrupting the conversation due to noise reduction delay, ensuring that the user speaks at the moment when the device's signal-to-noise ratio is at its optimal.

[0161] Step S803: Within a preset time window after the interactive prompt signal is issued, the sound acquisition component of the home appliance is activated to acquire voice commands.

[0162] Here, a preset time window (e.g., 3 to 5 seconds) is opened simultaneously with or immediately after the interactive prompt signal. Within the preset time window, the sound acquisition component is in full-speed recording mode, capturing the user's voice input commands (such as turning on the stir-fry mode).

[0163] The acquired voice commands at this point have a high signal-to-noise ratio (typically ≥5dB), enabling high-precision speech recognition even without relying on complex multi-microphone array algorithms, using only basic noise suppression algorithms. If a valid voice command is acquired within the preset time window, the window is extended or the process proceeds to the next stage. If no valid input is acquired by the end of the window, a timeout mechanism is triggered, and power is prepared to be restored.

[0164] Based on the above embodiments, this application provides a home appliance, referring to... Figure 9 The home appliance provided in this application includes: a home appliance body 1, a controller 2 and a working component 3 respectively disposed inside the home appliance body 1; the working component 3 is communicatively connected to the controller 2; the controller 2 is configured to execute the voice recognition optimization method as described in the foregoing embodiments.

[0165] In one embodiment, the home appliance also includes a sound acquisition component 4 that is communicatively connected to the controller 2.

[0166] Here, home appliances mainly refer to intelligent electronic devices that generate high-decibel environmental noise when performing core functions. Their specific forms include, but are not limited to, range hoods, high-speed blenders, handheld vacuum cleaners, robot vacuum cleaners, or industrial dust removal equipment.

[0167] The home appliance includes the appliance body 1, controller 2, working component 3, and sound acquisition component 4. The controller 2, working component 3, and sound acquisition component 4 are all located inside or on the appliance body 1 and form an electrical or communication connection.

[0168] Working component 3 is the core component of the household appliance that performs physical work, and it is also the main source of environmental noise (i.e., noise source) during the operation of the appliance. The specific form of working component 3 depends on the type of household appliance. For example, when the household appliance is a range hood, working component 3 includes a drive motor and a fan system. When the household appliance is a blender, working component 3 includes a high-speed blending motor. When the household appliance is a vacuum cleaner, working component 3 includes a vacuum suction motor.

[0169] The working component 3 is communicatively connected to the controller 2. The working component 3 is internally or externally connected to a drive circuit (such as a motor driver). The drive circuit can receive control commands from the controller 2 (such as PWM pulse width modulation signals, voltage regulation signals, or frequency conversion signals) and adjust the speed, torque, or power of the working component 3 in real time according to the control commands, thereby changing the output power and the ambient noise sound pressure level generated when the working component 3 is running.

[0170] The sound acquisition component 4 is used to acquire sound signals from the external environment and convert sound waves into electrical signals to be transmitted to the controller 2.

[0171] To reduce the hardware cost of the device and demonstrate the technical advantages of this application, the sound acquisition component 4 preferably adopts a single microphone module. In embodiments where cost constraints are not a concern, the sound acquisition component 4 may also be a dual-microphone or multi-microphone array.

[0172] The controller 2 can be a general-purpose microcontroller 2, a digital signal processor, or a low-cost system chip integrated on the device motherboard.

[0173] The controller 2 is configured to execute the speech recognition optimization logic in the aforementioned method embodiments. In short, when the controller 2 detects a potential voice interaction request through the sound acquisition component 4, it actively and temporarily reduces the output power of the working component 3 by sending an instruction to the working component 3, and then controls the working component 3 to restore its original power after the voice interaction process ends.

[0174] Home appliances may also include the following auxiliary hardware or logic modules: The voice activity detection module can be integrated into controller 2 or be a standalone low-power chip. It is directly connected to the sound acquisition component 4 and is used to monitor the energy value of ambient sound in real time in low-power mode (e.g., power consumption ≤ 0.5W). When the detected sound energy exceeds a preset threshold, the voice activity detection module is responsible for waking up controller 2 or triggering the voice recognition process.

[0175] A noise sensor can also be installed on the main body of the home appliance 1. The noise sensor is used to collect the ambient noise value around the device in real time (accuracy up to ±1dB) and send the collected noise data to the controller 2. The controller 2 uses the real-time noise value to look up a table to determine the amount of power reduction that needs to be achieved.

[0176] The home appliance has an internal memory for storing data such as a preset human voice feature library (including 85-300Hz fundamental frequency and harmonic features), a preset graded adjustment strategy table, and a preset smooth transition strategy, which are called by the controller 2 during operation.

[0177] To ensure operational safety, the home appliance also includes sensors for monitoring the status of the working components 3, such as an NTC temperature sensor located on the motor windings, or a current detection circuit for monitoring the current load. These sensors feed back operational safety status parameters to the controller 2, so that the controller 2 can intercept power reduction operations when the equipment overheats or is overloaded.

[0178] The computer program product provided in this application includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the preceding method embodiments. For specific implementation details, please refer to the method embodiments, which will not be repeated here.

[0179] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system and apparatus described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0180] Furthermore, in the description of the embodiments of this application, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0181] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0182] In the description of this application, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0183] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The scope of protection of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of protection of the claims.

Claims

1. A speech recognition optimization method, characterized in that, Applied to household appliances, the household appliances including working parts that generate environmental noise during operation; the method includes: In response to detecting an audio signal that meets a preset trigger condition, determine whether the audio signal meets a preset speech recognition condition; If the audio signal does not meet the preset voice recognition conditions, obtain the current working status parameters of the home appliance; Based on a preset graded adjustment strategy, the target power adjustment range corresponding to the current operating state parameters is determined; The working component is controlled to reduce its output power according to the target power adjustment range, so as to reduce the environmental noise; When the working component is in a state of reduced output power, voice commands are collected, and voice commands are recognized to obtain voice recognition results.

2. The speech recognition optimization method according to claim 1, characterized in that, After obtaining the speech recognition result, the method further includes: Determine whether the speech recognition result contains multi-turn dialogue intent; If the speech recognition result does not contain the multi-turn dialogue intent, or if no subsequent audio input is detected within a preset waiting time, the working component is controlled to increase the output power at a preset recovery slope until it is restored to the output power before the reduction.

3. The speech recognition optimization method according to claim 1, characterized in that, Prior to the step of responding to the detection of an audio signal that meets a preset trigger condition, the method further includes: The sound acquisition component of the home appliance is controlled to be in a real-time detection state to acquire ambient sound signals; Calculate the real-time energy value of the ambient sound signal; Determine whether the real-time energy value exceeds a preset energy threshold, and whether the duration of exceeding the preset energy threshold reaches a preset time threshold; If the real-time energy value exceeds the preset energy threshold, and the duration of exceeding the preset energy threshold reaches the preset time threshold, the ambient sound signal is determined as the audio signal that meets the preset triggering condition.

4. The speech recognition optimization method according to claim 1, characterized in that, The step of determining whether an audio signal that meets a preset trigger condition meets a preset speech recognition condition in response to detecting such an audio signal includes: Acquire the audio signal and extract its spectral features; Calculate the matching degree between the spectral features and the preset human voice feature library, and determine whether the matching degree is higher than the preset matching threshold; If the matching degree is higher than the preset matching threshold, the audio signal is determined to be a valid human voice signal; Calculate the current signal-to-noise ratio (SNR) of the valid human voice signal, and determine whether the current SNR is lower than a preset SNR threshold. If the current signal-to-noise ratio is lower than the preset signal-to-noise ratio threshold, it is determined that the audio signal does not meet the preset speech recognition conditions.

5. The speech recognition optimization method according to claim 1, characterized in that, The current operating status parameters include at least one of the current operating level of the home appliance and the current ambient noise level. The step of determining the target power adjustment range corresponding to the current operating state parameters according to a preset graded adjustment strategy includes: Identify the noise intensity level corresponding to the current operating status parameters; If the current operating state parameter is identified as corresponding to the first noise intensity level, the target power adjustment range is determined to be the first ratio; If the current operating state parameter is identified as corresponding to the second noise intensity level, the target power adjustment range is determined to be the second ratio; The higher the current working level or the greater the current ambient noise value, the higher the corresponding noise intensity level; the second noise intensity level is higher than the first noise intensity level, and the second ratio is greater than the first ratio.

6. The speech recognition optimization method according to claim 1, characterized in that, The step of controlling the working component to reduce the output power according to the target power adjustment range includes: Based on the target power adjustment range, calculate the target operating parameters of the working component; Based on a preset smooth transition strategy, the actual operating parameters of the working component are controlled to change towards the target operating parameters until the target operating parameters are reached; wherein, the preset smooth transition strategy limits the power change rate of the working component to be less than or equal to a preset change rate threshold.

7. The speech recognition optimization method according to claim 1, characterized in that, Before the step of controlling the working component to reduce the output power according to the target power adjustment range, the method further includes: Obtain the operational safety status parameters of the home appliance; the operational safety status parameters include at least the temperature data and load data of the working components; Determine whether the operational safety status parameters are within the preset safety protection range; If the operating safety status parameter is within the preset safety protection range, the operation of reducing output power is prohibited, and the current output power of the working component remains unchanged.

8. The speech recognition optimization method according to claim 1, characterized in that, The step of collecting voice commands includes: Monitor whether the output power of the working component decreases to the target power value; If the power level is reduced to the target value, the home appliance is controlled to issue an interactive prompt signal; the interactive prompt signal is used to guide the user to start inputting the voice command. Within a preset time window after the interactive prompt signal is issued, the sound acquisition component of the home appliance is activated to acquire the voice command.

9. A household appliance, characterized in that, include: The appliance body comprises a controller and a working component respectively disposed inside the appliance body; the working component is communicatively connected to the controller. The controller is configured to perform the speech recognition optimization method as described in any one of claims 1-8.

10. The household appliance according to claim 9, characterized in that, The home appliance also includes a sound acquisition component that is communicatively connected to the controller.

Images