Room calibration triggering method, apparatus, device, and storage medium

Abstract

This invention relates to the field of electronic digital data processing technology, and discloses a room calibration triggering method, apparatus, device, and storage medium to improve the accuracy of room calibration triggering. The room calibration triggering method includes: acquiring the current acoustic signal and auxiliary decision-making information related to the current acoustic environment, and analyzing the audio content features based on the current acoustic signal; the auxiliary decision-making information includes the current usage scenario, user identity, and environmental context information; extracting the current acoustic fingerprint based on the current acoustic signal; matching the corresponding benchmark acoustic fingerprint from a dynamic benchmark fingerprint database based on the current usage scenario and user identity; calculating the target fingerprint difference between the current acoustic fingerprint and the benchmark acoustic fingerprint by combining the environmental context information and audio content feature information; determining whether the calibration triggering conditions are met based on the target fingerprint difference; and performing either a lightweight calibration or a full calibration if the calibration triggering conditions are met.

Description

Technical Field

[0001] This invention relates to the field of electronic digital data processing technology, and in particular to a room calibration triggering method, apparatus, device, and storage medium. Background Technology

[0002] Current room acoustic calibration systems typically perform a fixed calibration procedure when manually triggered by the user or during initial device installation. These solutions often rely on pre-set test signals to acquire the room's impulse response, extracting limited acoustic parameters, and then comparing and compensating based on static benchmark data or ideal curves. However, this calibration method has significant limitations: firstly, the calibration process lacks scene awareness; secondly, the calibration triggering mechanism is rigid, unable to incorporate dynamic environmental changes and audio content characteristics for real-time judgment, leading to inaccurate calibration timing. Furthermore, traditional methods often use a single frequency response parameter as the calibration basis, resulting in incomplete calibration decision-making. Therefore, existing technologies are prone to frequent and unnecessary calibrations or failure to trigger calibration in a timely manner, impacting user experience and hindering accurate and efficient calibration for different acoustic environments and usage needs, thus limiting the adaptive optimization capabilities of audio equipment in complex room acoustic environments. Summary of the Invention

[0003] This invention provides a room calibration triggering method, apparatus, device, and storage medium to solve the problem of low accuracy of room calibration triggering in the prior art due to rigid calibration triggering mechanisms and reliance on fixed acoustic parameters and static references.

[0004] The first aspect of this invention provides a room calibration triggering method, comprising: acquiring a current acoustic signal and auxiliary decision information related to the current acoustic environment, and parsing audio content features based on the current acoustic signal, wherein the auxiliary decision information includes the current usage scenario, user identity, and environmental context information; extracting a current acoustic fingerprint based on the current acoustic signal, wherein the acoustic fingerprint includes at least one of frequency response curve, reverberation time, and reflection energy distribution; matching a corresponding reference acoustic fingerprint from a dynamic reference fingerprint database based on the current usage scenario and the user identity; calculating a target fingerprint difference degree between the current acoustic fingerprint and the reference acoustic fingerprint by combining the environmental context information and the audio content feature information; determining whether a calibration triggering condition is met based on the target fingerprint difference degree; and performing a lightweight calibration or a full calibration if the calibration triggering condition is met.

[0005] In one feasible implementation, the step of acquiring the current acoustic signal and auxiliary decision-making information related to the current acoustic environment, and obtaining audio content features based on the current acoustic signal, includes: acquiring the original acoustic signal in the room using a microphone array, and preprocessing the original acoustic signal to obtain the current acoustic signal; acquiring the current usage scenario, user identity, and environmental context information in real time by linking device sensors, user interaction interfaces, and cloud account data, and analyzing device operating status and user behavior patterns; performing time-frequency analysis on the current acoustic signal to extract audio content features, wherein the audio content features include at least signal category, main frequency components, loudness dynamic range, and harmonic structure.

[0006] In one feasible implementation, the step of extracting the current acoustic fingerprint based on the current acoustic signal includes: performing framing and windowing processing on the current acoustic signal, converting the time-domain signal into a multi-frame signal and performing frequency domain transformation to generate a time spectrum; based on the time spectrum, obtaining the reverberation time by calculating the energy attenuation curve of a specified frequency band, obtaining the reflected energy distribution by analyzing the energy distribution relationship between direct sound and reflected sound, and generating a frequency response curve by calculating the relative sound pressure level at each frequency point; and combining at least one of the reverberation time, the reflected energy distribution, and the frequency response curve to construct acoustic fingerprint data characterizing the current room acoustic characteristics.

[0007] In one feasible implementation, the step of matching the corresponding baseline acoustic fingerprint from the dynamic baseline fingerprint database based on the current usage scenario and the user identity includes: performing multi-level queries in the dynamic baseline fingerprint database using the current usage scenario as the primary index and the user identity as the secondary index. The dynamic baseline fingerprint database stores sets of baseline acoustic fingerprints measured at different historical time points under different scenarios and user combinations. If a completely matching scenario and user combination is found, the most recently recorded baseline acoustic fingerprint is selected first. If there is no complete match, an approximate match is performed based on scenario similarity and default user configuration, and the most similar baseline acoustic fingerprint is selected.

[0008] In one feasible implementation, calculating the target fingerprint difference between the current acoustic fingerprint and the reference acoustic fingerprint by combining the environmental context information and the audio content features includes: calculating an initial fingerprint difference based on the current acoustic fingerprint and the reference acoustic fingerprint; calculating an environmental consistency compensation coefficient based on the environmental context information; calculating a signal suitability compensation coefficient based on the audio content features; and weightedly fusing the initial fingerprint difference, the environmental consistency compensation coefficient, and the signal suitability compensation coefficient to obtain the target fingerprint difference.

[0009] In one feasible implementation, the step of calculating the initial fingerprint difference based on the current acoustic fingerprint and the reference acoustic fingerprint includes: extracting the frequency response curve vectors from the current acoustic fingerprint and the reference acoustic fingerprint respectively, calculating the sound pressure level difference between the two in each preset frequency band, and weighted summing the sound pressure level difference values ​​of all frequency bands to obtain the frequency response difference component; extracting the reverberation time values ​​from the current acoustic fingerprint and the reference acoustic fingerprint respectively, calculating the relative percentage deviation of the reverberation time between the two in a specified frequency band to obtain the reverberation time difference component; extracting the reflection energy distribution feature vectors from the current acoustic fingerprint and the reference acoustic fingerprint respectively, calculating the difference between the two in the ratio of early reflected sound to late reflected sound energy to obtain the reflection energy distribution difference component; and linearly combining the frequency response difference component, the reverberation time difference component, and the reflection energy distribution difference component according to preset weights, and normalizing them to obtain the initial fingerprint difference.

[0010] In one feasible implementation, the step of calculating the environmental consistency compensation coefficient based on the environmental context information includes: obtaining the real-time background noise spectrum from the environmental context information and querying the historical background noise spectrum associated with the current benchmark acoustic fingerprint in the dynamic benchmark fingerprint database; calculating the difference in average sound pressure level between the current noise spectrum and the historical noise spectrum across all frequency bands as the noise environment mismatch; simultaneously, obtaining the current temperature and humidity data and calculating its deviation from the standard calibration environmental conditions; and inputting the noise environment mismatch and the temperature and humidity deviation into a preset environmental impact assessment model to obtain the environmental consistency compensation coefficient.

[0011] In one feasible implementation, calculating the signal suitability compensation coefficient based on the audio content features includes: extracting the signal category from the audio content features; if the signal is speech, a single instrument, or narrowband noise, it is determined to have low suitability; if the signal is full-band music, pink noise, or a pulse sequence, it is determined to have high suitability; extracting the distribution breadth of the main frequency components and the richness of the harmonic structure from the audio content features; if the signal energy is concentrated in a few narrowbands and the harmonics are severely lacking, it is determined to have low excitation; if the signal energy is evenly distributed and the harmonics are rich, it is determined to have high excitation; and obtaining the corresponding signal suitability compensation coefficient by using a lookup table method or fuzzy logic rules based on the determination results of the signal suitability and the signal excitation.

[0012] In one feasible implementation, the step of weightedly fusing the initial fingerprint difference, the environmental consistency compensation coefficient, and the signal suitability compensation coefficient to obtain the target fingerprint difference includes: multiplying the environmental consistency compensation coefficient and the signal suitability compensation coefficient to obtain a comprehensive confidence correction factor; multiplying the initial fingerprint difference and the comprehensive confidence correction factor to obtain a preliminary corrected difference; and inputting the preliminary corrected difference into a nonlinear saturation function to map and normalize it to a fixed range, thereby obtaining the target fingerprint difference.

[0013] In one feasible implementation, determining whether the calibration trigger condition has been met based on the target fingerprint difference includes: dynamically obtaining a first trigger threshold from a preset trigger threshold mapping table according to the device operating status of the current usage scenario; calculating a dynamic sensitivity adjustment factor based on the noise level in the environmental context information and the signal type in the audio content features, and fine-tuning the first trigger threshold according to the factor to obtain a second trigger threshold; comparing the target fingerprint difference with the second trigger threshold, and if the target fingerprint difference is not less than the second trigger threshold, determining that the calibration trigger condition has been met.

[0014] In one feasible implementation, performing lightweight calibration or full calibration includes: determining whether to perform lightweight calibration or full calibration based on the target fingerprint difference, the difference distribution characteristics of the acoustic fingerprint, and the current usage scenario; if lightweight calibration is performed, a simplified pulse test signal is sent to the speaker system, the test signal being optimized for the difference frequency band, and filtering and gain parameter adjustment are performed on the difference frequency band based on the response signal collected by the microphone; if full calibration is performed, a complete sweep frequency test sequence is sent to the speaker system, and global acoustic parameters, including frequency response compensation, delay correction, and reverberation control, are recalculated and applied based on the multi-channel response signal collected by the microphone array.

[0015] In one feasible implementation, after performing the light calibration or full calibration, the method further includes: entering a silent verification phase, playing a verification signal and acquiring a calibrated acoustic fingerprint, calculating the matching degree between the calibrated acoustic fingerprint and the target fingerprint, and updating the dynamic benchmark fingerprint database based on the matching degree result.

[0016] A second aspect of the present invention provides a room calibration triggering device, comprising: an acquisition module, configured to acquire a current acoustic signal and auxiliary decision information related to the current acoustic environment, and to analyze audio content features based on the current acoustic signal, wherein the auxiliary decision information includes the current usage scenario, user identity, and environmental context information; an extraction module, configured to extract a current acoustic fingerprint based on the current acoustic signal, wherein the acoustic fingerprint includes at least one of a frequency response curve, reverberation time, and reflection energy distribution; a matching module, configured to match a corresponding reference acoustic fingerprint from a dynamic reference fingerprint database based on the current usage scenario and the user identity; a calculation module, configured to calculate a target fingerprint difference degree between the current acoustic fingerprint and the reference acoustic fingerprint by combining the environmental context information and the audio content feature information; a judgment module, configured to determine whether a calibration triggering condition is met based on the target fingerprint difference degree; and an execution module, configured to perform a lightweight calibration or a full calibration if the calibration triggering condition is met.

[0017] In one feasible implementation, the acquisition module is specifically used to: acquire the original acoustic signal in the room using a microphone array, and preprocess the original acoustic signal to obtain the current acoustic signal; acquire the current usage scenario, user identity, and environmental context information in real time by linking device sensors, user interaction interface, and cloud account data, and analyzing device operating status and user behavior patterns; perform time-frequency analysis on the current acoustic signal to extract audio content features, wherein the audio content features include at least signal category, main frequency components, loudness dynamic range, and harmonic structure.

[0018] In one feasible implementation, the extraction module is specifically used to: perform framing and windowing processing on the current acoustic signal, convert the time-domain signal into a multi-frame signal and perform frequency domain transformation to generate a time spectrum; based on the time spectrum, obtain the reverberation time by calculating the energy attenuation curve of a specified frequency band, obtain the reflected energy distribution by analyzing the energy distribution relationship between direct sound and reflected sound, and generate a frequency response curve by calculating the relative sound pressure level at each frequency point; combine at least one of the reverberation time, the reflected energy distribution and the frequency response curve to construct acoustic fingerprint data characterizing the current room acoustic characteristics.

[0019] In one feasible implementation, the matching module is specifically used to: perform multi-level queries in a dynamic benchmark fingerprint database, using the current usage scenario as the primary index and the user identity as the secondary index. The dynamic benchmark fingerprint database stores sets of benchmark acoustic fingerprints measured at different historical time points under different scenarios and user combinations. If a completely matching scenario and user combination is found, the most recently recorded benchmark acoustic fingerprint is selected first. If there is no complete match, an approximate match is performed based on scenario similarity and default user configuration, and the most similar benchmark acoustic fingerprint is selected.

[0020] In one feasible implementation, the calculation module includes: a first calculation unit for calculating an initial fingerprint difference based on the current acoustic fingerprint and the reference acoustic fingerprint; a second calculation unit for calculating an environmental consistency compensation coefficient based on the environmental context information; a third calculation unit for calculating a signal suitability compensation coefficient based on the audio content features; and a weighting unit for weighted fusion of the initial fingerprint difference, the environmental consistency compensation coefficient, and the signal suitability compensation coefficient to obtain a target fingerprint difference.

[0021] In one feasible implementation, the first calculation unit is specifically used to: extract the frequency response curve vectors from the current acoustic fingerprint and the reference acoustic fingerprint respectively, calculate the sound pressure level difference between the two in each preset frequency band, and perform a weighted summation of the sound pressure level difference values ​​of all frequency bands to obtain the frequency response difference component; extract the reverberation time values ​​from the current acoustic fingerprint and the reference acoustic fingerprint respectively, calculate the relative deviation percentage of the reverberation time between the two in a specified frequency band, and obtain the reverberation time difference component; The reflection energy distribution feature vectors of the current acoustic fingerprint and the reference acoustic fingerprint are extracted respectively, and the difference between the two in the ratio of early reflected sound energy to late reflected sound energy is calculated to obtain the reflection energy distribution difference component; the frequency response difference component, reverberation time difference component and reflection energy distribution difference component are linearly combined according to preset weights, and the initial fingerprint difference degree is obtained after normalization.

[0022] In one feasible implementation, the second calculation unit is specifically used to: obtain the real-time background noise spectrum from the environmental context information, and query the historical background noise spectrum associated with the current reference acoustic fingerprint in the dynamic reference fingerprint database; calculate the difference in average sound pressure level between the current noise spectrum and the historical noise spectrum across all frequency bands as the noise environment mismatch; simultaneously, obtain the current temperature and humidity data, and calculate its deviation from the standard calibration environmental conditions; input the noise environment mismatch and the temperature and humidity deviation into a preset environmental impact assessment model to obtain an environmental consistency compensation coefficient.

[0023] In one feasible implementation, the third calculation unit is specifically used to: extract signal categories from the audio content features; if the signal is speech, a single instrument, or narrowband noise, it is determined to have low applicability; if the signal is full-band music, pink noise, or a pulse sequence, it is determined to have high applicability; extract the distribution breadth of the main frequency components and the richness of the harmonic structure from the audio content features; if the signal energy is concentrated in a few narrowbands and the harmonics are severely lacking, it is determined to have low excitation; if the signal energy is evenly distributed and the harmonics are rich, it is determined to have high excitation; and obtain the corresponding signal applicability compensation coefficient by using a lookup table method or fuzzy logic rules based on the determination results of the signal applicability and the determination results of the signal excitation.

[0024] In one feasible implementation, the weighting unit is specifically used to: multiply the environmental consistency compensation coefficient by the signal applicability compensation coefficient to obtain a comprehensive confidence correction factor; multiply the initial fingerprint difference by the comprehensive confidence correction factor to obtain a preliminary corrected difference; and input the preliminary corrected difference into a nonlinear saturation function to map and normalize it to a fixed range, thereby obtaining the target fingerprint difference.

[0025] In one feasible implementation, the judgment module is specifically used to: dynamically obtain a first trigger threshold from a preset trigger threshold mapping table based on the device operating status of the current usage scenario; calculate a dynamic sensitivity adjustment factor based on the noise level in the environmental context information and the signal type in the audio content features, and fine-tune the first trigger threshold according to the factor to obtain a second trigger threshold; compare the target fingerprint difference with the second trigger threshold, and if the target fingerprint difference is not less than the second trigger threshold, determine that the calibration trigger condition has been met.

[0026] In one feasible implementation, the execution module is specifically used to: determine whether to perform a light calibration or a full calibration based on the target fingerprint difference, the difference distribution characteristics of the acoustic fingerprint, and the current usage scenario; if a light calibration is performed, a simplified pulse test signal is sent to the speaker system, the test signal being optimized for the difference frequency band, and the difference frequency band is filtered and the gain parameters are adjusted according to the response signal collected by the microphone; if a full calibration is performed, a complete sweep frequency test sequence is sent to the speaker system, and global acoustic parameters, including frequency response compensation, delay correction, and reverberation control, are recalculated and applied based on the multi-channel response signal collected by the microphone array.

[0027] In one feasible implementation, the device further includes: an update module, configured to enter a silent verification phase, play a verification signal and acquire a calibrated acoustic fingerprint, calculate the matching degree between the calibrated acoustic fingerprint and the target fingerprint, and update the dynamic benchmark fingerprint database based on the matching degree result.

[0028] A third aspect of the present invention provides an electronic device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor invokes the instructions in the memory to cause the electronic device to perform the room calibration triggering method described above.

[0029] A fourth aspect of the present invention provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the room calibration triggering method described above.

[0030] The technical solution provided by this invention involves acquiring the current acoustic signal and auxiliary decision-making information related to the current acoustic environment, and analyzing the current acoustic signal to obtain audio content features. The auxiliary decision-making information includes the current usage scenario, user identity, and environmental context information. A current acoustic fingerprint is extracted based on the current acoustic signal, and the acoustic fingerprint includes at least one of the following: frequency response curve, reverberation time, and reflection energy distribution. Based on the current usage scenario and the user identity, a corresponding reference acoustic fingerprint is matched from a dynamic reference fingerprint database. Combining the environmental context information and the audio content feature information, a target fingerprint difference degree is calculated between the current acoustic fingerprint and the reference acoustic fingerprint. Based on the target fingerprint difference degree, it is determined whether a calibration trigger condition is met. If the calibration trigger condition is met, a lightweight calibration or a full calibration is performed. In this embodiment of the invention, multi-dimensional information including audio content features, usage scenarios, user identity, and environmental context is dynamically collected and integrated to extract a comprehensive acoustic fingerprint containing frequency response curves, reverberation time, and reflection energy distribution. Based on the scenario and user identity, the corresponding benchmark fingerprint is matched from a dynamic benchmark library. Then, the difference degree of the target fingerprint is accurately calculated by combining the environmental context and audio content features, so as to realize intelligent judgment of calibration triggering time. This can effectively avoid the problems of inaccurate calibration timing, frequent invalid calibration, or calibration lag caused by traditional methods that rely on fixed test signals, single frequency response parameters, and static benchmarks. It significantly improves the adaptive calibration accuracy and response efficiency of audio devices in different usage scenarios, users, and dynamic acoustic environments, and ultimately improves the listening experience and optimizes system resource allocation. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of one embodiment of the room calibration triggering method in this invention; Figure 2This is a schematic diagram of another embodiment of the room calibration triggering method in this invention; Figure 3 This is a schematic diagram of one embodiment of the room calibration triggering device in this invention; Figure 4 This is a schematic diagram of another embodiment of the room calibration triggering device in this invention; Figure 5 This is a schematic diagram of one embodiment of the electronic device in this invention. Detailed Implementation

[0032] This invention provides a room calibration triggering method, apparatus, device, and storage medium, which improves the accuracy of room calibration triggering through scenario-based adaptive judgment and closed-loop dynamic optimization mechanism.

[0033] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0034] It is understood that the executing entity of this invention can be a room calibration trigger device, a terminal, or a server; no specific limitation is made here. This embodiment of the invention will be described using a server as the executing entity as an example.

[0035] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 One embodiment of the room calibration triggering method in this invention includes: 101. Obtain the current acoustic signal and auxiliary decision-making information related to the current acoustic environment, and analyze the audio content features based on the current acoustic signal. The auxiliary decision-making information includes the current usage scenario, user identity, and environmental context information. The system captures a mixed audio stream within the room in real time using a built-in microphone array as the current acoustic signal. Simultaneously, it integrates data from multiple sensors to generate auxiliary decision-making information: it automatically determines the current usage scenario (such as music listening, video conferencing, or gaming) by utilizing the camera's captured image content and combining it with the device's operating mode; it confirms the current primary user's identity through a facial recognition module or the device's logged-in account information; furthermore, environmental temperature and humidity sensors, ambient light sensors, and a network status detection module jointly provide environmental context information such as ambient noise levels, indoor lighting conditions, and device connection stability. Based on this, the current acoustic signal is fed into a pre-trained neural network model for real-time analysis. This model can identify and separate different audio elements in the signal, such as speech, music, and special effects sounds, thereby resolving the main type, rhythmic features, and dynamic range characteristics of the audio content, thus completing the extraction of audio content features.

[0036] 102. Extract the current acoustic fingerprint based on the current acoustic signal. The acoustic fingerprint includes at least one of the following: frequency response curve, reverberation time, and reflection energy distribution. High-resolution spectral analysis techniques were used to calculate and plot frequency response curves reflecting changes in sound pressure level at various frequencies. Simultaneously, the reverberation decay process of the room was obtained using impulse response measurement, and the reverberation time was calculated. Furthermore, a sound field spatial analysis algorithm was employed to separate direct and reflected sound components, and the energy distribution of reflected sound in different directions was statistically analyzed. Finally, the generated frequency response curves, reverberation time, and reflected energy distribution were combined to construct a current acoustic fingerprint characterizing the acoustic environment.

[0037] 103. Based on the current usage scenario and user identity, match the corresponding baseline acoustic fingerprint from the dynamic baseline fingerprint database; The current usage scenario is combined with the user's identity information verified through authentication to form a key index for the query. Then, a search is performed in a dynamically updated benchmark fingerprint database, which pre-stores standard acoustic fingerprint data of different users in different typical scenarios. Using a retrieval logic that combines scenario matching and user identity matching, a set of benchmark data that is completely consistent with or closest to the current key index for the query is found. By calculating the similarity of the currently extracted acoustic fingerprint with this set of benchmark fingerprints in multidimensional features, the benchmark acoustic fingerprint that best matches the current usage scenario and user identity is finally determined and retrieved.

[0038] 104. Combining environmental context information and audio content feature information, calculate the target fingerprint difference between the current acoustic fingerprint and the reference acoustic fingerprint; Taking into account both environmental context information and audio content characteristics, the difference analysis process between the current acoustic fingerprint and the benchmark acoustic fingerprint is dynamically weighted. For example, the confidence weight of the frequency response curve comparison results is adjusted based on the ambient noise level and network connection stability, while the sensitivity to differences in reverberation time and reflection energy distribution is calibrated based on the main type and rhythmic characteristics of the audio content. Through a multivariate fusion algorithm, the calibrated differences in various acoustic features, including the degree of frequency response curve shift, the amount of change in reverberation time, and the spatial differences in reflection energy distribution, are integrated and calculated into a comprehensive target fingerprint difference value that reflects the adaptation requirements of the current environment and audio content.

[0039] 105. Determine whether the calibration triggering conditions are met based on the target fingerprint difference. The calculated target fingerprint difference value is compared with preset multi-level dynamic thresholds. These thresholds are not fixed but are fine-tuned in real time based on the dynamic range characteristics of the current audio content, ambient lighting conditions, and device operating status. If the target fingerprint difference value exceeds the fine-tuned primary threshold for the current scene, it is directly determined that the calibration trigger condition is met. If the difference value is between the primary and secondary thresholds, a second evaluation is performed by combining the user's historical preferences and the short-term fluctuation trend of environmental sensors to ultimately determine whether the trigger condition is met.

[0040] A multi-level dynamic threshold matrix is ​​maintained in memory, using audio content type, ambient light level, and device load as three-dimensional index coordinates. For example, when the audio content is identified as a symphony with a wide dynamic range, the corresponding threshold subset for "high dynamic range music" is invoked. The initial value of the main threshold is set relatively loosely to accommodate the natural fluctuations of the music, but this initial value is affected in real time by ambient light sensor data. In scenarios with stable lighting, the main threshold is appropriately tightened to improve calibration sensitivity, while in scenarios with frequently changing lighting, such as when curtains are opened or closed, the main threshold is temporarily relaxed to avoid unnecessary calibration caused by instantaneous environmental changes. Simultaneously, the device load status is obtained by monitoring processor utilization and temperature. During high-load game operation, the secondary threshold is proactively increased to suppress false triggers caused by temporary factors such as device cooling fan noise. The specific comparison and judgment process is executed in a pipeline manner: First, the target fingerprint difference value is compared with the main threshold, which has been fine-tuned in real time by the aforementioned multi-dimensional factors. If it exceeds the threshold, a calibration trigger command is immediately generated. If the target difference value falls into the "fuzzy interval" jointly defined by the main threshold and the secondary threshold, an auxiliary decision-making module called the scene stability evaluator is triggered. This module performs two analyses in parallel: First, it queries the historical preference database of the currently verified user, extracts the user's past behavioral pattern data of manually intervening in calibration or expressing satisfaction (through voice feedback or application rating) under similar audio content and lighting conditions, and calculates a preference weight factor. Second, it accesses the real-time data stream of the environmental sensor array, performs short-time Fourier transform analysis on the fluctuation trends of noise level, temperature and humidity, and network latency in the recent period, quantifies its fluctuation frequency and amplitude, and generates an environmental stability index. Finally, the scene stability evaluator uses a pre-trained lightweight neural network model to perform fusion inference using the preference weight factor, the environmental stability index, and the current target fingerprint difference value itself as input features. The neural network outputs a confidence score between zero and one. This score is compared with a fixed decision threshold. If the confidence score exceeds the decision threshold, it is determined that the triggering condition is met. Otherwise, it is considered that the current difference is within the user's acceptable range and the environmental fluctuation is controllable, thereby suppressing the triggering of this calibration.

[0041] 106. If the calibration trigger conditions are met, perform a light calibration or a full calibration.

[0042] When the trigger condition is met, the system first determines the calibration mode based on the severity of the target fingerprint difference and the main type of audio content. For scenarios with minor differences or continuous speech content, the system initiates a lightweight calibration, quickly fine-tuning only the equalizer parameters of the speaker system to correct local deviations in the frequency response curve. For scenarios with significant differences or high-fidelity music and game sound effects, a full calibration is performed. This process remeasures the room impulse response and comprehensively adjusts a full set of audio processing parameters, including multi-channel delay, phase alignment, global equalization, and dynamic compression, to achieve deep adaptation and reconstruction of the sound field environment.

[0043] In this embodiment of the invention, multi-dimensional information including audio content features, usage scenarios, user identity, and environmental context is dynamically collected and integrated to extract a comprehensive acoustic fingerprint containing frequency response curves, reverberation time, and reflection energy distribution. Based on the scenario and user identity, the corresponding benchmark fingerprint is matched from a dynamic benchmark library. Then, the difference degree of the target fingerprint is accurately calculated by combining the environmental context and audio content features, so as to realize intelligent judgment of calibration triggering time. This can effectively avoid the problems of inaccurate calibration timing, frequent invalid calibration, or calibration lag caused by traditional methods that rely on fixed test signals, single frequency response parameters, and static benchmarks. It significantly improves the adaptive calibration accuracy and response efficiency of audio devices in different usage scenarios, users, and dynamic acoustic environments, and ultimately improves the listening experience and optimizes system resource allocation.

[0044] Please see Figure 2 Another embodiment of the room calibration triggering method in this invention includes: 201. Obtain the current acoustic signal and auxiliary decision-making information related to the current acoustic environment, and analyze the audio content features based on the current acoustic signal. The auxiliary decision-making information includes the current usage scenario, user identity and environmental context information. The system uses a microphone array to collect raw acoustic signals from the room and preprocesses them to obtain the current acoustic signal. By linking device sensors, user interaction interfaces, and cloud account data, and analyzing device operating status and user behavior patterns, it obtains real-time information on the current usage scenario, user identity, and environmental context. It then performs time-frequency analysis on the current acoustic signal to extract audio content features, which include at least signal category, main frequency components, loudness dynamic range, and harmonic structure.

[0045] 202. Extract the current acoustic fingerprint based on the current acoustic signal. The acoustic fingerprint includes at least one of the following: frequency response curve, reverberation time, and reflection energy distribution. The current acoustic signal is framed and windowed to convert the time-domain signal into a multi-frame signal and perform frequency domain transformation to generate a time spectrum. Based on the time spectrum, the reverberation time is obtained by calculating the energy attenuation curve of a specified frequency band, the reflected energy distribution is obtained by analyzing the energy distribution relationship between direct sound and reflected sound, and the frequency response curve is generated by calculating the relative sound pressure level at each frequency point. At least one of the reverberation time, reflected energy distribution, and frequency response curve is combined to construct acoustic fingerprint data characterizing the current room acoustic characteristics.

[0046] First, the multi-channel raw audio stream captured by the microphone array is preprocessed for synchronization and noise reduction. Then, an adaptive window length algorithm is used to analyze the preprocessed time-domain signal. A longer time window is used for steady-state signal segments to improve frequency resolution, while a shorter window is used for transient signal segments to maintain time accuracy. Each signal frame is weighted by a variable conical window function to minimize spectral leakage. Subsequently, a high-precision Fast Fourier Transform is performed on each windowed frame of signal, and the complex spectra of all frames are arranged in chronological order to form a three-dimensional joint expression matrix of time-frequency amplitude spectrum and phase spectrum.

[0047] Based on the generated time-frequency spectrum, three core feature extraction tasks are performed. First, to calculate reverberation time, the algorithm automatically detects significant audio events with energy exceeding the background noise threshold in the time-frequency spectrum as sound source excitation points. From these points, it tracks the energy attenuation trajectories within multiple predefined sub-bands covering key listening frequencies. Inverse integration is applied to these attenuation curves, and the logarithmic energy attenuation slope is fitted using the least squares method. The slope is then converted into the time required for the sound to attenuate to a specific decibel value in the corresponding frequency band. Finally, the median of the results for each frequency band is taken as the global reverberation time estimate for the current room. Second, extracting the reflected energy distribution is more complex. Beamforming is applied to the multi-channel signal to enhance the direct sound component from the assumed speaker direction. Then, adaptive eigenvalue decomposition or a sparse recovery algorithm is used to separate the main early reflected sound components from the mixed impulse response. The algorithm calculates the arrival direction of these early reflected sounds on the time axis, their delay time relative to the direct sound, and their respective energy intensity, constructing a three-dimensional spatial energy distribution map of the reflected sound. Simultaneously, the ratio of the total energy of later diffuse reflections to the total energy of early reflections is used as a supplementary description of the reverberation characteristics. Third, the frequency response curve is not generated based on external test signals, but rather by identifying and extracting stable and continuous tonal or broadband noise segments from the current audio signal. The algorithm performs one-third octave band smoothing on the long-time average power spectrum of these segments and, with reference to a pre-calibrated reference frequency response curve of the device itself and microphone combination in an anechoic environment, calculates the sound pressure level offset of each center frequency point relative to the reference curve in the current environment, thus obtaining the normalized room frequency response curve. Finally, the reverberation time value, reflection energy distribution vector, and normalized frequency response curve array are encapsulated into a structured data packet with a timestamp and signal-to-noise ratio quality tag, which together constitute the current acoustic fingerprint characterizing the room's acoustic properties.

[0048] 203. Based on the current usage scenario and user identity, match the corresponding baseline acoustic fingerprint from the dynamic baseline fingerprint database; Using the current usage scenario as the primary index and user identity as the secondary index, multi-level queries are performed in the dynamic baseline fingerprint database. The dynamic baseline fingerprint database stores the set of baseline acoustic fingerprints measured at different historical time points under different scenarios and user combinations. If a scenario and user combination that matches exactly is found, the most recently recorded baseline acoustic fingerprint is selected first. If there is no perfect match, an approximate match is performed based on scenario similarity and default user configuration, and the most similar baseline acoustic fingerprint is selected.

[0049] A multi-level hybrid index structure is constructed for efficient retrieval of a dynamic benchmark fingerprint database. This database stores data using a time-series versioning method. Each record contains a scene tag, a unique user identifier, a timestamp, a complete acoustic fingerprint feature vector, and associated environmental metadata. When the matching process begins, the algorithm uses the real-time identified normalized scene name string as the first-level hash key, and the verified user biometric hash value or account unique code as the second-level hash key, performing a precise search within the index structure. If a record with a completely matching key is found in the database, the system further filters all historical records under that combination and sorts them according to the record's timestamp and additional data quality score, prioritizing the most recent record with a quality score higher than a threshold, extracting its acoustic fingerprint as the benchmark. If no perfect match is found, a multi-stage approximate matching process is initiated. The first stage performs scene similarity matching, where the algorithm converts the current scene name and all first-level keys in the database into semantic vectors using a natural language processing model, calculates cosine similarity, and selects all scene groups with similarity exceeding a preset threshold. The second stage performs user context matching within these candidate scene groups, where the system matches the current scene name with the user context. The system queries the user's identity to determine their predefined user group and retrieves records marked as members of that group from the database. If no specific group is defined, the system uses the default user configuration as the filtering condition. The third stage involves a comprehensive evaluation of the candidate record set selected in the first two stages. The evaluation factors include not only scene semantic similarity and user relevance, but also the degree of matching between the current environmental context and the candidate record's historical environmental metadata, as well as the freshness of the candidate record. The system assigns dynamic weights to each factor and calculates the final fit score for each candidate record using a weighted scoring model. Finally, the acoustic fingerprint of the record with the highest fit score is selected as the baseline acoustic fingerprint for this matching.

[0050] 204. Based on the current acoustic fingerprint and the reference acoustic fingerprint, calculate the initial fingerprint difference. The frequency response curve vectors of the current acoustic fingerprint and the reference acoustic fingerprint are extracted separately. The sound pressure level difference between the two in each preset frequency band is calculated, and the sound pressure level difference values ​​of all frequency bands are weighted and summed to obtain the frequency response difference component. The reverberation time values ​​of the current acoustic fingerprint and the reference acoustic fingerprint are extracted separately. The relative percentage deviation of the reverberation time between the two in the specified frequency band is calculated to obtain the reverberation time difference component. The reflection energy distribution feature vectors of the current acoustic fingerprint and the reference acoustic fingerprint are extracted separately. The difference between the two in the ratio of early reflection sound energy to late reflection sound energy is calculated to obtain the reflection energy distribution difference component. The frequency response difference component, the reverberation time difference component, and the reflection energy distribution difference component are linearly combined according to preset weights and normalized to obtain the initial fingerprint difference degree.

[0051] The algorithm performs structured analysis on the current acoustic fingerprint and the matched benchmark acoustic fingerprint to extract a comparable set of features. For the frequency response curves, the algorithm resamples and aligns the two normalized curves according to the international standard one-third octave band. It calculates the absolute difference of the sound pressure level at the center frequency of each band, and then performs a weighted sum of the differences of each band according to a preset frequency domain weight matrix adjusted based on the equal loudness curves and the importance of room modes to obtain the frequency response difference component. The low-frequency band and key human ear-sensitive bands are usually given higher weights. For the reverberation time, the system selects the geometric mean of the reverberation time in multiple sub-bands in the mid-frequency range as a representative value, calculates the absolute difference between the current value and the benchmark value, divides the difference by the benchmark value to obtain a relative deviation ratio, and then maps this ratio to the reverberation time difference component through a nonlinear function. This function amplifies deviations that exceed the acceptable range. For the reflected energy distribution, the algorithm simplifies the complex distribution vector into several key feature indicators, including the ratio of early reflected sound energy to total reflected sound energy, the standard deviation of early reflected sound on the time axis, and the concentration index of the main reflection directions. The Euclidean distance between the current and baseline fingerprints on these indicators is calculated, and then a weighted average is performed based on prior knowledge of the influence of each indicator on spatial perception to obtain the reflected energy distribution difference components. Finally, the system inputs these three components into a fusion model pre-trained with a large amount of data. This model is not a simple linear combination, but rather learns the nonlinear interaction between the three components through a shallow neural network and outputs a scalar value between zero and one. This value is the normalized initial fingerprint difference degree. This process ensures that the fusion of acoustic characteristic differences in different dimensions better conforms to the subjective perception of the human ear and the objective requirements of system performance.

[0052] 205. Calculate the environmental consistency compensation coefficient based on environmental context information; The system obtains the real-time background noise spectrum from the environmental context information and queries the historical background noise spectrum associated with the current benchmark acoustic fingerprint in the dynamic benchmark fingerprint database. It calculates the difference in average sound pressure level between the current noise spectrum and the historical noise spectrum across all frequency bands as the noise environment mismatch. At the same time, it obtains the current temperature and humidity data and calculates its deviation from the standard calibration environmental conditions. It inputs the noise environment mismatch and temperature and humidity deviation into the preset environmental impact assessment model to obtain the environmental consistency compensation coefficient.

[0053] The background noise one-third octave band spectrum, collected and analyzed by a digital microphone array during periods of no sound source activation, is read in real time from the environmental context information stream. Simultaneously, the historical background noise spectrum of the same specification recorded when the currently successfully matched benchmark acoustic fingerprint was created is extracted from the metadata field of the benchmark acoustic fingerprint as a comparison benchmark. The calculation of noise environment mismatch is not a simple difference in average sound pressure level. Instead, the two noise spectra are first perceptually weighted and filtered to simulate the difference in human ear sensitivity to noise at different frequencies. Then, the root mean square error of the sound pressure level after weighting each frequency band is calculated. Furthermore, the time stability index of the current noise spectrum is combined. That is, the variance of the recent noise spectrum sequence is analyzed to determine whether the noise is steady-state or transient. If it is highly transient noise, an attenuation factor is applied to the mismatch calculation result to reduce its impact. Simultaneously, the system acquires the current Celsius temperature and relative humidity readings from temperature and humidity sensors and compares them with standard environmental condition reference values ​​stored in the device's factory calibration parameters. After calculating the absolute difference, it maps these values ​​using two independent S-curve functions to simulate the nonlinear relationship between temperature and humidity and the effects of sound speed and air absorption. The two mapping results are then geometrically averaged to obtain the temperature and humidity deviation. Finally, the processed noise environment mismatch and temperature and humidity deviation, along with the day / night time identifier of the current time, are used as input feature vectors and fed into a gradient boosting decision tree model pre-trained with extensive environmental disturbance experimental data. This model analyzes the complex correlation between these features and acoustic fingerprint measurement errors, outputting an environmental consistency compensation coefficient ranging from zero to two. This coefficient directly quantifies the comprehensive impact of current environmental conditions on the reliability of acoustic fingerprint matching. The closer the coefficient value is to one, the higher the environmental consistency; the greater the deviation from one, the greater the environmental impact, thus requiring corresponding compensation for subsequent difference calculations.

[0054] 206. Calculate the signal suitability compensation coefficient based on the characteristics of the audio content; The signal category is extracted from the audio content features. If the signal is speech, a single instrument, or narrowband noise, it is judged as low applicability; if the signal is full-band music, pink noise, or a pulse sequence, it is judged as high applicability. The distribution breadth of the main frequency components and the richness of the harmonic structure are extracted from the audio content features. If the signal energy is concentrated in a few narrowbands and the harmonics are severely lacking, it is judged as low excitation; if the signal energy is evenly distributed and the harmonics are rich, it is judged as high excitation. Based on the judgment results of signal applicability and signal excitation, the corresponding signal applicability compensation coefficient is obtained by using a lookup table method or fuzzy logic rules.

[0055] The algorithm decodes audio content feature data packets and extracts the signal category probability distribution vector output by the neural network model. This vector describes the probability that the current signal belongs to each category, such as speech, music, or noise. Instead of using a simple threshold decision, the algorithm calculates a probability-based applicability score, where the probability of categories such as full-band music, white noise, or swept-frequency signals contributes a positive weight, while the probability of categories such as speech and fixed monotones contributes a negative weight. Simultaneously, the algorithm analyzes the distribution breadth of major frequency components, quantifying the uniformity of frequency distribution by calculating the entropy value of the signal power spectrum. A higher entropy value indicates a more uniform energy distribution and a wider excitation bandwidth. Furthermore, the algorithm assesses the richness of harmonic structure by analyzing the significance and number of fundamental harmonic peaks in the cepstral domain. Finally, the system inputs the frequency distribution entropy and harmonic richness index into a pre-trained support vector machine classifier, which classifies the signal into high, medium, or low excitation levels. Ultimately, the determination of the signal suitability compensation coefficient was not achieved through a simple table lookup method, but rather through a lightweight neural network regression model. The model's inputs were the aforementioned basic suitability score, excitation level, and the instantaneous dynamic range rate of change of the audio signal. Through internal nonlinear mapping, the model outputs a continuous coefficient value between zero and one. This coefficient comprehensively reflects the reliability and completeness of the current audio content in assessing differences in the acoustic environment. A high coefficient value indicates that the signal can fully excite the room's acoustic characteristics and provide a reliable assessment, while a low coefficient value suggests that the current audio content may not be able to effectively reveal the true changes in the acoustic environment.

[0056] 207. The initial fingerprint difference, environmental consistency compensation coefficient and signal suitability compensation coefficient are weighted and fused to obtain the target fingerprint difference.

[0057] Multiply the environmental consistency compensation coefficient by the signal applicability compensation coefficient to obtain a comprehensive confidence correction factor; multiply the initial fingerprint difference by the comprehensive confidence correction factor to obtain a preliminary corrected difference; input the preliminary corrected difference into a nonlinear saturation function, map it and normalize it to a fixed range, thereby obtaining the target fingerprint difference.

[0058] Multiplying the environmental consistency compensation coefficient by the signal suitability compensation coefficient yields an initial comprehensive confidence correction factor between zero and two. This factor reflects the impact of current environmental and signal conditions on the overall confidence of the difference assessment. However, to avoid the factor being too large and excessively suppressing differences under optimal conditions, or too small and excessively amplifying differences under extremely poor conditions, this initial factor needs to be processed by a constraint function based on a bisigma curve. This ensures that its output value is smoothly constrained within a narrower, more reasonable range, thus generating the final comprehensive confidence correction factor used for correction. Next, the initial fingerprint difference is multiplied by this comprehensive confidence correction factor for preliminary correction. This step essentially performs a dynamic scaling of the initial difference based on data confidence. The initially corrected difference is then fed into a parameter-adjustable nonlinear saturation function, employing a modified sigmoid form. The steepness parameter is dynamically adjusted based on the current usage scenario; for example, a steeper curve is used in high-fidelity music scenarios to quickly distinguish subtle differences, while a smoother curve is used in voice communication scenarios to improve robustness. This function nonlinearly maps and compresses the input values ​​into a fixed output range of zero to one, achieving normalization. Finally, the system applies a short-time smoothing filter based on the recent historical sequence of target fingerprint difference values ​​to eliminate jumps caused by random fluctuations, outputting a final, stable, and comparable target fingerprint difference value.

[0059] In one implementation, the adaptive weighted fusion algorithm based on historical feedback achieves intelligent correction of the initial fingerprint difference by dynamically adjusting the weights of three factors: environmental consistency, signal suitability, and historical calibration feedback. The algorithm first collects the environmental stability index. and signal quality score Combined with preset benchmark weights , and sensitivity coefficient , Dynamically calculate environment weights and signal weights Residual weight Assign historical correction factors. Then, based on recent calibration success rates. and the time interval since the last successful calibration Through historical influence intensity coefficient and decay time constant Calculate the historical correction factor Then, the environmental consistency compensation coefficient is... Signal suitability compensation coefficient Historical correction coefficient By dynamic weight , , The fusion weight is obtained by weighted summation. Meanwhile, considering the time interval since the last arbitrary calibration... Introducing a time-domain decay factor This is to suppress repeated triggering immediately after calibration. Ultimately, the target fingerprint difference... From the initial degree of difference fusion weight Time-domain decay factor By multiplying these components, more accurate, stable, and efficient calibration triggering decisions can be achieved in a variety of complex scenarios.

[0060] The formula for calculating the difference in target fingerprints is:

[0061] Among them, the fusion weight It is obtained by weighting three factors:

[0062] Among them, dynamic weights , , The calculation is as follows:

[0063]

[0064]

[0065] Historical correction coefficient for:

[0066] Time decay factor for:

[0067] For example, in a scenario with high-quality music playback and a stable environment, the input condition received by the algorithm is: initial difference. =0.42, Environmental Consistency Compensation Coefficient =0.95, signal suitability compensation coefficient =0.85, Environmental Stability Index =0.9, signal quality score =0.8, historical calibration success rate =0.85, the time interval since the last successful calibration. =7200 seconds, the time interval since the last arbitrary calibration. =600 seconds; the algorithm first dynamically calculates the environmental weight based on the baseline weight and the sensitivity coefficient. =0.472 and signal weight =0.360, residual weight =0.168 is allocated to the historical correction term; then the historical correction coefficient is calculated. =1.014, fusion weight =0.924, time-domain decay factor =0.893; finally passed =0.42×0.924×0.893, the target difference was calculated to be 0.346. The results show that although the initial difference is high, the time domain suppression factor significantly reduces the final output value because the calibration has just been completed, thus effectively avoiding unnecessary repeated calibration triggers in the steady state.

[0068] By dynamically allocating the weights of various influencing factors based on real-time environmental stability and signal quality, and combining historical calibration records for intelligent feedback adjustment, the system significantly improves its adaptability to complex and ever-changing scenarios and the accuracy of calibration trigger decisions. By introducing a time-domain suppression mechanism, the system effectively avoids repetitive or invalid calibration operations, reducing resource consumption and improving user experience while ensuring timely system response. The algorithm has a clear structure, high computational efficiency, and good configurability and scalability, making it easy to customize deployments for different hardware platforms and application scenarios, thus comprehensively enhancing the overall robustness, reliability, and practicality of the system.

[0069] 208. Determine whether the calibration triggering conditions are met based on the target fingerprint difference. Based on the device's operating status in the current usage scenario, the first trigger threshold is dynamically obtained from a preset trigger threshold mapping table; based on the noise level in the environmental context information and the signal type in the audio content features, a dynamic sensitivity adjustment factor is calculated, and the first trigger threshold is fine-tuned according to this factor to obtain the second trigger threshold; the target fingerprint difference is compared with the second trigger threshold, and if the target fingerprint difference is not less than the second trigger threshold, it is determined that the calibration trigger condition has been met.

[0070] Based on the current usage scenario and the real-time load status of the device's internal processor and audio codec, a basic trigger threshold is obtained from a multidimensional lookup table. This threshold mapping table is pre-established before the device leaves the factory through extensive subjective listening tests and objective parameter analysis under different scenarios and loads. Subsequently, the dynamic sensitivity adjustment module starts working. This module comprehensively analyzes the A-weighted equivalent continuous sound pressure level continuously monitored and calculated by the digital microphone in the environmental context information, as well as the probability distribution of signal type and the instantaneous dynamic range of the signal in the audio content characteristics. Based on a set of fuzzy logic rules, for example, when the ambient noise level is consistently high and the signal type is dynamically smooth speech, the system generates an adjustment factor greater than one to appropriately relax the trigger threshold. Conversely, when the environment is very quiet and the signal is classical music with a very large dynamic range, an adjustment factor less than one is generated to tighten the threshold. This adjustment factor is used to multiply and scale the basic trigger threshold to obtain the final second trigger threshold used for decision-making. Finally, the decision maker compares the target fingerprint difference calculated through all the aforementioned steps with the dynamically generated second trigger threshold in real time. If the target fingerprint difference is greater than or equal to the threshold, a high-level calibration trigger signal is immediately generated. At the same time, the system also adds an anti-jitter mechanism, which requires that the trigger conditions must be met continuously for more than a short time window to avoid false triggering due to instantaneous changes in the signal or single interference from the environment.

[0071] 209. If the calibration trigger conditions are met, perform a light calibration or a full calibration. Based on the target fingerprint difference, the difference distribution characteristics of the acoustic fingerprint, and the current usage scenario, it is determined whether to perform a light calibration or a full calibration. If a light calibration is performed, a simplified pulse test signal is sent to the speaker system. The test signal is optimized for the difference frequency band, and the difference frequency band is filtered and the gain parameters are adjusted according to the response signal collected by the microphone. If a full calibration is performed, a complete sweep frequency test sequence is sent to the speaker system, and the global acoustic parameters, including frequency response compensation, delay correction, and reverberation control, are recalculated and applied based on the multi-channel response signal collected by the microphone array.

[0072] A pattern decision engine is activated, which comprehensively analyzes the specific numerical value of the target fingerprint difference, the contribution ratio of each component in the initial fingerprint difference, and the stringency of the current usage scenario. For example, if the difference is mainly concentrated in the frequency response shift of a few frequency bands and the target difference does not exceed the full calibration threshold, and the current scenario is a voice call, then a lightweight calibration is decided. After lightweight calibration is initiated, the digital signal processor generates a set of sparse multi-frequency sine wave clusters or a specially designed narrowband maximum length sequence as test signals. The frequency positions of these signals are precisely targeted at the problem frequency bands identified in the fingerprint difference analysis. After being played through the speaker, the response is collected by the microphone. The system then uses an adaptive filtering algorithm to iteratively update only the digital filter coefficients corresponding to the problem frequency bands, quickly completing the fine-tuning of local gain and equalization. The entire process is completed in a very short time to avoid being noticed by the user. If the decision engine determines that the target has a high degree of difference, the difference is widely distributed across multiple dimensions such as frequency response and reverberation, or the current environment is high-fidelity music or immersive cinema, then a full calibration is initiated. The full calibration first controls all speaker units to sequentially play an exponentially swept frequency signal or pseudo-random noise sequence covering the entire audio range, while simultaneously acquiring the multi-channel impulse response of the microphone array at a high sampling rate. Based on these responses, the system uses advanced acoustic algorithms, including calculating the precision equalization filter for each speaker channel using the minimum mean square error method in the frequency domain, accurately measuring and compensating for the propagation delay between each channel using the cross-correlation method, and generating a secondary sound field to suppress unfavorable room reflections through a multi-channel adaptive reverberation cancellation algorithm. Finally, this entire set of complex acoustic parameters is safely loaded into the audio processing pipeline to complete the deep recalibration of the entire playback system.

[0073] 210. Update the dynamic baseline fingerprint database; Entering the silent verification phase, the verification signal is played and the calibrated acoustic fingerprint is acquired. The matching degree between the calibrated acoustic fingerprint and the target fingerprint is calculated, and the dynamic benchmark fingerprint database is updated based on the matching degree result.

[0074] During this phase, the device plays a specially designed broadband verification signal sequence at an extremely low volume, typically below the ambient noise masking threshold, during a short period when the user does not actively play the signal or the system deems it idle. This sequence simultaneously includes a sweep component for evaluating frequency response and a pulse component for evaluating transient response. A high-sensitivity microphone array synchronously acquires the room's response to this verification signal, and a new set of calibrated acoustic fingerprints is calculated using the exact same algorithm as for extracting the current acoustic fingerprint. Subsequently, the system performs a feature-by-feature comparison between this calibrated acoustic fingerprint and the target acoustic fingerprint expected to be achieved in this calibration. The target fingerprint is an ideal state derived from a baseline fingerprint, historical preferences, and a general acoustic target model. The calculation of the fit includes not only the mean square error of the frequency response curve and the relative deviation of the reverberation time, but also a similarity measure of the spatial reflection distribution. Finally, a weighted fusion model outputs an overall fit score. The system will only perform an update operation when the score exceeds a high-confidence threshold that is dynamically adjusted based on the calibration mode and historical success rate. The update operation adds the acoustic fingerprint after the current calibration, the precise scene label when the calibration was triggered, the user identity, the environmental context snapshot, and the calibration parameter version number as a brand new entry to the dynamic benchmark fingerprint database, and adds a timestamp and quality label to this entry. At the same time, the system will use a least recently used strategy or an evaluation based on the validity of the entry to clean up the most old or lowest quality redundant entries in the database to maintain the timeliness, representativeness, and storage efficiency of the fingerprint database.

[0075] In this embodiment of the invention, by fusing and analyzing audio content features with usage scenarios, user identities, and environmental context information, and extracting a comprehensive acoustic fingerprint including frequency response curves, reverberation time, and reflection energy distribution, the initial fingerprint difference can be further optimized by calculating environmental consistency compensation coefficients and signal suitability compensation coefficients, based on multi-dimensional scenario-based matching of dynamic benchmark fingerprints. This makes the calibration trigger condition judgment more accurate and adaptive, reducing false triggers caused by subtle environmental changes or differences in audio content characteristics, and avoiding the systematic bias of traditional single-parameter matching. At the same time, by introducing a silent verification stage and a benchmark library dynamic update mechanism based on consistency, closed-loop verification of calibration effects and self-learning optimization of the benchmark library are achieved, ensuring that the audio system maintains optimal acoustic state over a long period of time under different user, scenario, and environmental changes, significantly improving the intelligence, accuracy, and continuous adaptive capability of the calibration.

[0076] The room calibration triggering method in the embodiments of the present invention has been described above. The room calibration triggering device in the embodiments of the present invention will be described below. Please refer to [link / reference]. Figure 3 One embodiment of the room calibration triggering device in this invention includes: The acquisition module 301 is used to acquire the current acoustic signal and auxiliary decision-making information related to the current acoustic environment, and to obtain audio content features based on the current acoustic signal. The auxiliary decision-making information includes the current usage scenario, user identity and environmental context information. Extraction module 302 is used to extract the current acoustic fingerprint based on the current acoustic signal. The acoustic fingerprint includes at least one of the following: frequency response curve, reverberation time, and reflection energy distribution. Matching module 303 is used to match the corresponding reference acoustic fingerprint from the dynamic reference fingerprint database based on the current usage scenario and user identity; The calculation module 304 is used to combine environmental context information and audio content feature information to calculate the target fingerprint difference between the current acoustic fingerprint and the reference acoustic fingerprint; The judgment module 305 is used to determine whether the calibration triggering condition is met based on the difference of the target fingerprint. The execution module 306 is used to perform a light calibration or a full calibration if the calibration triggering conditions are met.

[0077] In this embodiment of the invention, multi-dimensional information including audio content features, usage scenarios, user identity, and environmental context is dynamically collected and integrated to extract a comprehensive acoustic fingerprint containing frequency response curves, reverberation time, and reflection energy distribution. Based on the scenario and user identity, the corresponding benchmark fingerprint is matched from a dynamic benchmark library. Then, the difference degree of the target fingerprint is accurately calculated by combining the environmental context and audio content features, so as to realize intelligent judgment of calibration triggering time. This can effectively avoid the problems of inaccurate calibration timing, frequent invalid calibration, or calibration lag caused by traditional methods that rely on fixed test signals, single frequency response parameters, and static benchmarks. It significantly improves the adaptive calibration accuracy and response efficiency of audio devices in different usage scenarios, users, and dynamic acoustic environments, and ultimately improves the listening experience and optimizes system resource allocation.

[0078] Please see Figure 4 Another embodiment of the room calibration triggering device in this invention includes: The acquisition module 301 is used to acquire the current acoustic signal and auxiliary decision-making information related to the current acoustic environment, and to obtain audio content features based on the current acoustic signal. The auxiliary decision-making information includes the current usage scenario, user identity and environmental context information. Extraction module 302 is used to extract the current acoustic fingerprint based on the current acoustic signal. The acoustic fingerprint includes at least one of the following: frequency response curve, reverberation time, and reflection energy distribution. Matching module 303 is used to match the corresponding reference acoustic fingerprint from the dynamic reference fingerprint database based on the current usage scenario and user identity; The calculation module 304 is used to combine environmental context information and audio content feature information to calculate the target fingerprint difference between the current acoustic fingerprint and the reference acoustic fingerprint; The judgment module 305 is used to determine whether the calibration triggering condition is met based on the difference of the target fingerprint. The execution module 306 is used to perform a light calibration or a full calibration if the calibration triggering conditions are met.

[0079] Optionally, the acquisition module 301 can be specifically used for: The system uses a microphone array to collect raw acoustic signals from the room and preprocesses them to obtain the current acoustic signal. By linking device sensors, user interaction interfaces, and cloud account data, and analyzing device operating status and user behavior patterns, it obtains real-time information on the current usage scenario, user identity, and environmental context. It then performs time-frequency analysis on the current acoustic signal to extract audio content features, which include at least signal category, main frequency components, loudness dynamic range, and harmonic structure.

[0080] Optionally, the extraction module 302 can be specifically used for: The current acoustic signal is framed and windowed to convert the time-domain signal into a multi-frame signal and perform frequency domain transformation to generate a time spectrum. Based on the time spectrum, the reverberation time is obtained by calculating the energy attenuation curve of a specified frequency band, the reflected energy distribution is obtained by analyzing the energy distribution relationship between direct sound and reflected sound, and the frequency response curve is generated by calculating the relative sound pressure level at each frequency point. At least one of the reverberation time, reflected energy distribution, and frequency response curve is combined to construct acoustic fingerprint data characterizing the current room acoustic characteristics.

[0081] Optionally, the matching module 303 can be specifically used for: Using the current usage scenario as the primary index and user identity as the secondary index, multi-level queries are performed in the dynamic baseline fingerprint database. The dynamic baseline fingerprint database stores the set of baseline acoustic fingerprints measured at different historical time points under different scenarios and user combinations. If a scenario and user combination that matches exactly is found, the most recently recorded baseline acoustic fingerprint is selected first. If there is no perfect match, an approximate match is performed based on scenario similarity and default user configuration, and the most similar baseline acoustic fingerprint is selected.

[0082] Optionally, the calculation module 304 includes: The first calculation unit 3041 is used to calculate the initial fingerprint difference based on the current acoustic fingerprint and the reference acoustic fingerprint; The second calculation unit 3042 is used to calculate the environmental consistency compensation coefficient based on environmental context information; The third calculation unit 3043 is used to calculate the signal suitability compensation coefficient based on the characteristics of the audio content; The weighting unit 3044 is used to weight and fuse the initial fingerprint difference, the environmental consistency compensation coefficient and the signal suitability compensation coefficient to obtain the target fingerprint difference.

[0083] Optionally, the first computing unit 3041 can be specifically used for: The frequency response curve vectors of the current acoustic fingerprint and the reference acoustic fingerprint are extracted separately. The sound pressure level difference between the two in each preset frequency band is calculated, and the sound pressure level difference values ​​of all frequency bands are weighted and summed to obtain the frequency response difference component. The reverberation time values ​​of the current acoustic fingerprint and the reference acoustic fingerprint are extracted separately. The relative percentage deviation of the reverberation time between the two in the specified frequency band is calculated to obtain the reverberation time difference component. The reflection energy distribution feature vectors of the current acoustic fingerprint and the reference acoustic fingerprint are extracted separately. The difference between the two in the ratio of early reflection sound energy to late reflection sound energy is calculated to obtain the reflection energy distribution difference component. The frequency response difference component, the reverberation time difference component, and the reflection energy distribution difference component are linearly combined according to preset weights and normalized to obtain the initial fingerprint difference degree.

[0084] Optionally, the second computing unit 3042 can be specifically used for: The system obtains the real-time background noise spectrum from the environmental context information and queries the historical background noise spectrum associated with the current benchmark acoustic fingerprint in the dynamic benchmark fingerprint database. It calculates the difference in average sound pressure level between the current noise spectrum and the historical noise spectrum across all frequency bands as the noise environment mismatch. At the same time, it obtains the current temperature and humidity data and calculates its deviation from the standard calibration environmental conditions. It inputs the noise environment mismatch and temperature and humidity deviation into the preset environmental impact assessment model to obtain the environmental consistency compensation coefficient.

[0085] Optionally, the third computing unit 3043 can be specifically used for: The signal category is extracted from the audio content features. If the signal is speech, a single instrument, or narrowband noise, it is judged as low applicability; if the signal is full-band music, pink noise, or a pulse sequence, it is judged as high applicability. The distribution breadth of the main frequency components and the richness of the harmonic structure are extracted from the audio content features. If the signal energy is concentrated in a few narrowbands and the harmonics are severely lacking, it is judged as low excitation; if the signal energy is evenly distributed and the harmonics are rich, it is judged as high excitation. Based on the judgment results of signal applicability and signal excitation, the corresponding signal applicability compensation coefficient is obtained by using a lookup table method or fuzzy logic rules.

[0086] Optionally, the weighting unit 3044 can be specifically used for: Multiply the environmental consistency compensation coefficient by the signal applicability compensation coefficient to obtain a comprehensive confidence correction factor; multiply the initial fingerprint difference by the comprehensive confidence correction factor to obtain a preliminary corrected difference; input the preliminary corrected difference into a nonlinear saturation function, map it and normalize it to a fixed range, thereby obtaining the target fingerprint difference.

[0087] Optionally, the judgment module 305 can be specifically used for: Based on the device's operating status in the current usage scenario, the first trigger threshold is dynamically obtained from a preset trigger threshold mapping table; based on the noise level in the environmental context information and the signal type in the audio content features, a dynamic sensitivity adjustment factor is calculated, and the first trigger threshold is fine-tuned according to this factor to obtain the second trigger threshold; the target fingerprint difference is compared with the second trigger threshold, and if the target fingerprint difference is not less than the second trigger threshold, it is determined that the calibration trigger condition has been met.

[0088] Optionally, execution module 306 can be specifically used for: Based on the target fingerprint difference, the difference distribution characteristics of the acoustic fingerprint, and the current usage scenario, it is determined whether to perform a light calibration or a full calibration. If a light calibration is performed, a simplified pulse test signal is sent to the speaker system. The test signal is optimized for the difference frequency band, and the difference frequency band is filtered and the gain parameters are adjusted according to the response signal collected by the microphone. If a full calibration is performed, a complete sweep frequency test sequence is sent to the speaker system, and the global acoustic parameters, including frequency response compensation, delay correction, and reverberation control, are recalculated and applied based on the multi-channel response signal collected by the microphone array.

[0089] Optionally, the room calibration trigger also includes: The update module 307 is used to enter the silent verification stage, play the verification signal and collect the calibrated acoustic fingerprint, calculate the matching degree between the calibrated acoustic fingerprint and the target fingerprint, and update the dynamic benchmark fingerprint library based on the matching degree result.

[0090] In this embodiment of the invention, by fusing and analyzing audio content features with usage scenarios, user identities, and environmental context information, and extracting a comprehensive acoustic fingerprint including frequency response curves, reverberation time, and reflection energy distribution, the initial fingerprint difference can be further optimized by calculating environmental consistency compensation coefficients and signal suitability compensation coefficients, based on multi-dimensional scenario-based matching of dynamic benchmark fingerprints. This makes the calibration trigger condition judgment more accurate and adaptive, reducing false triggers caused by subtle environmental changes or differences in audio content characteristics, and avoiding the systematic bias of traditional single-parameter matching. At the same time, by introducing a silent verification stage and a benchmark library dynamic update mechanism based on consistency, closed-loop verification of calibration effects and self-learning optimization of the benchmark library are achieved, ensuring that the audio system maintains optimal acoustic state over a long period of time under different user, scenario, and environmental changes, significantly improving the intelligence, accuracy, and continuous adaptive capability of the calibration.

[0091] above Figure 3 and Figure 4 The room calibration triggering device in this embodiment of the invention will be described in detail from the perspective of modular functional entities. The electronic device in this embodiment of the invention will be described in detail from the perspective of hardware processing.

[0092] See Figure 5 As shown, the electronic device includes a processor 500 and a memory 501. The memory 501 stores machine-executable instructions that can be executed by the processor 500. The processor 500 executes the machine-executable instructions to implement the room calibration triggering method described above.

[0093] Furthermore, Figure 5 The electronic device shown also includes a bus 502 and a communication interface 503. The processor 500, the communication interface 503 and the memory 501 are connected via the bus 502.

[0094] The memory 501 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 503 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc. The bus 502 may be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 5 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0095] The processor 500 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 500 or by instructions in software form. The processor 500 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this disclosure. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this disclosure can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules may reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 501. The processor 500 reads the information in memory 501 and, in conjunction with its hardware, completes the method steps of the aforementioned embodiment.

[0096] The present invention also provides an electronic device, the computer device including a memory and a processor, the memory storing computer-readable instructions, which, when executed by the processor, cause the processor to perform the steps of the room calibration triggering method described in the above embodiments.

[0097] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of the room calibration triggering method.

[0098] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0099] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. 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 the present invention. 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.

[0100] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications 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 the present invention.

Claims

1. A room calibration triggering method, characterized in that, The method includes: The system acquires the current acoustic signal and auxiliary decision-making information related to the current acoustic environment, and analyzes the audio content features based on the current acoustic signal. The auxiliary decision-making information includes the current usage scenario, user identity, and environmental context information. The current acoustic fingerprint is extracted based on the current acoustic signal, and the acoustic fingerprint includes at least one of the following: frequency response curve, reverberation time, and reflection energy distribution; Based on the current usage scenario and the user identity, a corresponding reference acoustic fingerprint is matched from the dynamic reference fingerprint database; By combining the environmental context information and the audio content feature information, the target fingerprint difference between the current acoustic fingerprint and the reference acoustic fingerprint is calculated; Determine whether the calibration trigger condition is met based on the target fingerprint difference. If the calibration trigger conditions are met, perform a light calibration or a full calibration.

2. The room calibration triggering method according to claim 1, characterized in that, The process of acquiring the current acoustic signal and auxiliary decision-making information related to the current acoustic environment, and obtaining audio content features based on the current acoustic signal, includes: The original acoustic signals in the room are collected using a microphone array, and the original acoustic signals are preprocessed to obtain the current acoustic signal; By linking device sensors, user interaction interfaces and cloud account data, and analyzing device operating status and user behavior patterns, the system can obtain real-time information on the current usage scenario, user identity and environmental context. The current acoustic signal is subjected to time-frequency analysis to extract audio content features, which include at least signal category, main frequency components, loudness dynamic range and harmonic structure.

3. The room calibration triggering method according to claim 1, characterized in that, The step of extracting the current acoustic fingerprint based on the current acoustic signal includes: The current acoustic signal is framed and windowed to convert the time-domain signal into a multi-frame signal and perform frequency domain transformation to generate a time-spectrum. Based on the time spectrum, the reverberation time is obtained by calculating the energy attenuation curve of a specified frequency band, the reflected energy distribution is obtained by analyzing the energy distribution relationship between direct sound and reflected sound, and the frequency response curve is generated by calculating the relative sound pressure level at each frequency point. The reverberation time, the reflected energy distribution, and the frequency response curve are combined to construct acoustic fingerprint data characterizing the current acoustic properties of the room.

4. The room calibration triggering method according to claim 1, characterized in that, The step of matching the corresponding baseline acoustic fingerprint from the dynamic baseline fingerprint database based on the current usage scenario and the user identity includes: Using the current usage scenario as the primary index and the user identity as the secondary index, multi-level queries are performed in the dynamic benchmark fingerprint database, which stores the set of benchmark acoustic fingerprints measured at different historical time points under different scenarios and user combinations. If a matching scenario and user combination is found, the most recently recorded baseline acoustic fingerprint will be selected first. If there is no perfect match, an approximate match will be made based on the scene similarity and the default user configuration, and the most similar baseline acoustic fingerprint will be selected.

5. The room calibration triggering method according to claim 1, characterized in that, The step of calculating the target fingerprint difference between the current acoustic fingerprint and the reference acoustic fingerprint by combining the environmental context information and the audio content features includes: Calculate the initial fingerprint difference based on the current acoustic fingerprint and the reference acoustic fingerprint; Based on the aforementioned environmental context information, calculate the environmental consistency compensation coefficient; Based on the aforementioned audio content characteristics, a signal suitability compensation coefficient is calculated; The initial fingerprint difference, the environmental consistency compensation coefficient, and the signal suitability compensation coefficient are weighted and fused to obtain the target fingerprint difference.

6. The room calibration triggering method according to claim 5, characterized in that, The calculation of the initial fingerprint difference based on the current acoustic fingerprint and the reference acoustic fingerprint includes: Extract the frequency response curve vectors from the current acoustic fingerprint and the reference acoustic fingerprint respectively, calculate the sound pressure level difference between the two in each preset frequency band, and perform a weighted summation of the sound pressure level difference values ​​in all frequency bands to obtain the frequency response difference component; Extract the reverberation time values ​​from the current acoustic fingerprint and the reference acoustic fingerprint respectively, calculate the percentage of relative deviation between the two reverberation times in a specified frequency band, and obtain the reverberation time difference component. Extract the reflection energy distribution feature vectors from the current acoustic fingerprint and the reference acoustic fingerprint respectively, calculate the difference between the two in the ratio of early reflected sound energy to late reflected sound energy, and obtain the reflection energy distribution difference component; The frequency response difference component, reverberation time difference component, and reflection energy distribution difference component are linearly combined according to preset weights, and then normalized to obtain the initial fingerprint difference degree.

7. The room calibration triggering method according to claim 5, characterized in that, The calculation of the environmental consistency compensation coefficient based on the environmental context information is included in the following block: The real-time background noise spectrum is obtained from the environmental context information, and the historical background noise spectrum associated with the current benchmark acoustic fingerprint is queried from the dynamic benchmark fingerprint database. The difference in average sound pressure level across all frequency bands between the current noise spectrum and the historical noise spectrum is calculated as the noise environment mismatch; at the same time, the current temperature and humidity data are acquired, and their deviation from the standard calibration environmental conditions is calculated. The noise environment mismatch and the temperature and humidity deviation are input into a preset environmental impact assessment model to obtain the environmental consistency compensation coefficient.

8. The room calibration triggering method according to claim 5, characterized in that, The calculation of the signal suitability compensation coefficient based on the audio content features includes: The signal category is extracted from the audio content features. If the signal is speech, a single instrument, or narrowband noise, it is determined to be of low applicability; if the signal is full-band music, pink noise, or a pulse sequence, it is determined to be of high applicability. The distribution breadth of the main frequency components and the richness of the harmonic structure are extracted from the audio content features. If the signal energy is concentrated in a few narrow bands and the harmonics are severely lacking, it is judged as low excitation degree; if the signal energy is evenly distributed and the harmonics are rich, it is judged as high excitation degree. Based on the determination results of the signal applicability and the signal excitation degree, the corresponding signal applicability compensation coefficient is obtained by using a lookup table method or fuzzy logic rules.

9. The room calibration triggering method according to claim 5, characterized in that, The step of weightedly fusing the initial fingerprint difference, the environmental consistency compensation coefficient, and the signal suitability compensation coefficient to obtain the target fingerprint difference includes: Multiply the environmental consistency compensation coefficient by the signal suitability compensation coefficient to obtain a comprehensive reliability correction factor; Multiply the initial fingerprint difference by the comprehensive confidence correction factor to obtain a preliminary corrected difference. The initially corrected difference is input into a nonlinear saturation function, mapped and normalized to a fixed range, thereby obtaining the target fingerprint difference.

10. The room calibration triggering method according to claim 1, characterized in that, The step of determining whether the calibration trigger condition is met based on the target fingerprint difference includes: Based on the device operating status of the current usage scenario, the first trigger threshold is dynamically obtained from the preset trigger threshold mapping table; Based on the noise level in the environmental context information and the signal type in the audio content features, a dynamic sensitivity adjustment factor is calculated, and the first trigger threshold is fine-tuned according to the factor to obtain a second trigger threshold. The target fingerprint difference is compared with the second trigger threshold. If the target fingerprint difference is not less than the second trigger threshold, it is determined that the calibration trigger condition has been met.

11. The room calibration triggering method according to claim 1, characterized in that, The process of performing a lightweight calibration or a full calibration includes: Based on the target fingerprint difference, the difference distribution characteristics of the acoustic fingerprint, and the current usage scenario, determine whether to perform a lightweight calibration or a full calibration. If a lightweight calibration is performed, a simplified pulse test signal is sent to the speaker system. This test signal is optimized for the difference frequency band, and the difference frequency band is filtered and the gain parameters are adjusted based on the response signal acquired by the microphone. If a full calibration is performed, a complete sweep test sequence is sent to the loudspeaker system, and global acoustic parameters, including frequency response compensation, delay correction, and reverberation control, are recalculated and applied based on the multi-channel response signals acquired by the microphone array.

12. The room calibration triggering method according to claim 1, characterized in that, After performing the light or full calibration, the following is also included: Entering the silent verification phase, the verification signal is played and the calibrated acoustic fingerprint is acquired. The matching degree between the calibrated acoustic fingerprint and the target fingerprint is calculated, and the dynamic benchmark fingerprint database is updated based on the matching degree result.

13. A room calibration triggering device, characterized in that, The room calibration triggering device includes: The acquisition module is used to acquire the current acoustic signal and auxiliary decision-making information related to the current acoustic environment, and to parse the audio content features based on the current acoustic signal. The auxiliary decision-making information includes the current usage scenario, user identity and environmental context information. The extraction module is used to extract the current acoustic fingerprint based on the current acoustic signal, wherein the acoustic fingerprint includes at least one of the following: frequency response curve, reverberation time, and reflection energy distribution. The matching module is used to match the corresponding reference acoustic fingerprint from the dynamic reference fingerprint database based on the current usage scenario and the user identity. The calculation module is used to combine the environmental context information and the audio content feature information to calculate the target fingerprint difference between the current acoustic fingerprint and the reference acoustic fingerprint; The judgment module is used to determine whether the calibration triggering condition is met based on the difference of the target fingerprint; The execution module is used to perform either a light calibration or a full calibration if the calibration trigger conditions are met.

14. An electronic device, characterized in that, The electronic device includes: a memory and at least one processor, wherein the memory stores instructions; The at least one processor invokes the instructions in the memory to cause the electronic device to perform the room calibration triggering method as described in any one of claims 1-12.

15. A computer-readable storage medium storing instructions thereon, characterized in that, When the instruction is executed by the processor, it implements the room calibration triggering method as described in any one of claims 1-12.

Images