A method for intelligent dimming control of sound and light in architectural space

The intelligent dimming control method using microphone arrays and acoustic feature extraction solves the problem that traditional building lighting systems cannot dynamically adjust, and achieves robust acoustic event detection and scene classification in variable noise environments, improving lighting quality and energy efficiency, and is applicable to a variety of building spaces.

CN123075779APending Publication Date: 2026-11-10GUANGZHOU LANDE ELECTRONICS TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611561148.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-10
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Traditional building lighting systems cannot dynamically adjust the light environment according to the real-time activity status within the space, resulting in insufficient lighting comfort and energy waste. Existing sound and light mapping technologies suffer from problems such as coarse feature extraction, low scene recognition accuracy, and visual discomfort.

Method used

Sound signals are collected by a microphone array, multi-dimensional acoustic feature extraction and dynamic threshold determination are performed, and intelligent lighting adjustment is achieved by combining an environmental adaptive mechanism and a sound-light mapping model, including dynamic gain factor, glare suppression function and double buffer mechanism, and closed-loop feedback control.

Benefits of technology

It achieves robustness in sound event detection and scene classification under varying noise environments, improves lighting quality and energy efficiency, ensures visual comfort and safety, and is suitable for architectural spaces such as conference rooms, concert halls, classrooms, and open office areas.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123075779A_ABST
    Figure CN123075779A_ABST
Patent Text Reader

Abstract

This invention provides a method for intelligent dimming control of sound and light in architectural spaces, relating to the fields of intelligent building lighting and acoustic environment perception. The method includes sound acquisition and preprocessing, acoustic feature extraction, online scene classification, sound-light linkage mapping, and closed-loop dimming execution. This invention employs the aforementioned method for intelligent dimming control of sound and light in architectural spaces, achieving synergistic improvement in lighting quality and energy-saving optimization by integrating multi-dimensional acoustic feature extraction, dynamic threshold scene classification, and a sound-light mapping model with comfort constraints.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of intelligent building lighting and acoustic environment perception, and in particular to an intelligent dimming control method for sound and light linkage in building spaces. Background Technology

[0002] Traditional building lighting systems largely rely on manual settings or simple timer controls, failing to dynamically adjust the lighting environment based on real-time activity within the space. This results in insufficient lighting comfort and energy waste. Some existing technologies attempt to incorporate acoustic sensors, but these suffer from issues such as coarse feature extraction, low accuracy in scene recognition when background noise changes, and harsh sound-light mapping causing visual discomfort and flicker. Therefore, there is an urgent need for a sound-light linkage dimming method that can adapt to changes in the acoustic environment and achieve smooth, safe, and expressive lighting. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent dimming control method for sound and light linkage in architectural spaces. By integrating multi-dimensional acoustic feature extraction, dynamic threshold scene classification, and a sound and light mapping model with comfort constraints, it achieves synergistic improvement of lighting quality and energy-saving optimization.

[0004] To achieve the above objectives, the present invention provides an intelligent dimming control method for sound and light linkage in architectural spaces, comprising the following steps: Multi-channel sound signals are acquired by a microphone array deployed in the building space at a preset sampling rate, and the sound signals are preprocessed, including bandpass filtering, framing, and windowing. The preprocessed acoustic signal is subjected to a fast Fourier transform to obtain the spectrum. The frequency range is divided into multiple sub-bands using a frequency band division strategy, and the energy of each sub-band is calculated. Acoustic features are extracted, and acoustic events are detected based on a dynamic threshold. The dynamic threshold is adaptively adjusted according to the real-time estimated background noise floor. The extracted acoustic feature vectors are input into the classifier, which outputs the current activity scene label. An environment-adaptive dynamic threshold mechanism is introduced during the classification process to adjust the scene judgment threshold according to changes in background noise. Based on the scene label, the basic light parameters are obtained from the preset mapping library, and a dynamic gain factor, glare suppression function and double buffering mechanism are introduced to generate the final target light parameters; The target light parameters are converted into dimming signals to drive the lamps, and the actual illuminance and color temperature are collected by an ambient light sensor for closed-loop feedback control.

[0005] Preferably, the acoustic features include A-weighted sound pressure level, speech interference level, reverberation time, spectral centroid, and spectral entropy.

[0006] Preferably, the frequency band division strategy adopts one-third octave band division or psychoacoustic critical band division, the language interference level in the acoustic features is obtained by calculating the average sound pressure level of the octave band from 500 Hz to 4 kHz, and the reverberation time is obtained by the frequency domain blind estimation method.

[0007] Preferably, the sound event detection based on dynamic threshold determination specifically includes: real-time estimation of the noise floor of a sub-band, setting a dynamic threshold for the sub-band consisting of the noise floor plus a frequency-related offset, and determining a sound event start point when the energy of a sub-band exceeds its dynamic threshold and the duration exceeds a preset minimum duration, and statistically analyzing the sound event density per unit time and the standard deviation of the event interval as the rhythm intensity.

[0008] Preferably, the environment adaptive dynamic threshold mechanism includes: automatically adjusting the sound pressure level judgment threshold used to distinguish between quiet scenes and low-noise activity scenes when the background noise floor changes significantly; when judging unmanned energy-saving scenes, using a sound event density threshold and a continuous silence duration threshold to form a dual dynamic judgment condition, and only switching to unmanned energy-saving scenes when all sub-band energy is continuously lower than the adaptive silence threshold for more than a preset duration and the sound event density is lower than the preset density threshold.

[0009] Preferably, the dynamic gain factor is calculated based on the short-term sound pressure level change rate, and the gain factor has a logarithmic relationship with the change rate and is limited to a preset minimum and maximum value; different activity scenarios correspond to different gain coefficients, with higher gain coefficients used in music scenarios and lower gain coefficients used in language scenarios.

[0010] Preferably, the glare suppression function works as follows: when the initial target brightness of a lamp calculated based on the acousto-optic mapping exceeds the glare safety threshold determined based on the lamp installation height, beam angle and observer position, the hyperbolic tangent function is used to smooth and compress the brightness, so that the brightness exceeding the safety threshold gradually approaches the preset maximum allowable brightness, thereby limiting glare.

[0011] Preferably, the dual buffering mechanism is as follows: maintaining a current lighting parameter buffer and a target lighting parameter buffer in the control unit, generating intermediate lighting parameters that gradually change from the current parameters to the target parameters within a preset transition time through linear interpolation or S-shaped easing interpolation algorithm, and updating and protecting the target buffer during the transition process to prevent jitter caused by the writing of new calculation results during the transition.

[0012] Preferably, the method further includes a zone dimming step: using the microphone array to determine the location of the main sound source through a direction-of-arrival estimation algorithm, dynamically increasing the illumination brightness of the area corresponding to the sound source location, while maintaining the basic illumination brightness of other areas, and the increased illumination brightness is also output after being limited by the glare suppression function and smoothly transitioned by the double buffering mechanism.

[0013] Therefore, the present invention employs the above-mentioned intelligent dimming control method for sound and light linkage in architectural spaces, and the technical effects are as follows: By extracting rich acoustic features through fast Fourier transform and fine frequency band segmentation, and combining them with dynamic threshold determination, the robustness of sound event detection and scene classification in varying noise environments is greatly enhanced.

[0014] The integration of dynamic gain factor, glare suppression function and double buffering mechanism in the sound-light mapping model not only endows the light with the artistic expressiveness of following the dynamic changes of sound, but also fully guarantees visual comfort and safety, and completely eliminates dimming flicker and abrupt changes.

[0015] Closed-loop feedback and optional zone dimming further enhance lighting accuracy and energy efficiency. This method is widely applicable to various architectural spaces such as conference rooms, concert halls, classrooms, open office areas, and art exhibitions. Attached Figure Description

[0016] Figure 1 Simulation results for energy saving under unmanned conditions: Figure 1 (a) in the figure is a simulation diagram of acoustic feature extraction and dynamic threshold acoustic event detection under unmanned energy-saving conditions; Figure 1 (b) in the figure is a simulation diagram of scene recognition and double-buffered smooth dimming under unmanned energy-saving conditions; Figure 1 (c) in the figure is a simulation diagram of the glare suppression function and the operating point under unmanned energy-saving conditions; Figure 2 Simulation results for a quiet office environment: Figure 2 (a) in the figure is a simulation diagram of acoustic feature extraction and dynamic threshold acoustic event detection under quiet office conditions; Figure 2 (b) in the figure is a simulation diagram of scene recognition and double-buffered smooth dimming under quiet office conditions; Figure 2 (c) in the figure is a simulation diagram of the glare suppression function and the working point under quiet office conditions; Figure 3 The simulation results are shown below: Figure 3 (a) in the figure is a simulation diagram of acoustic feature extraction and dynamic threshold acoustic event detection under music playback conditions; Figure 3 (b) is a simulation diagram of scene recognition and double-buffered smooth dimming under music playback conditions; Figure 3(c) in the figure is a simulation diagram of the glare suppression function and the operating point under music playback conditions. Detailed Implementation

[0017] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0018] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0019] Example 1 The following example uses a multi-functional conference room with an area of ​​50 square meters and a net height of 2.8 meters to further describe the intelligent dimming control method for sound and light linkage in architectural space described in this invention. Four omnidirectional microelectromechanical system (MEMS) microphones are evenly installed on the ceiling of the conference room, forming a microphone array. Simultaneously, 16 LED lamps supporting correlated color temperature adjustment are arranged, with the maximum brightness of the lamps corresponding to an illuminance of 750 lux on the work surface. An illuminance sensor is installed on the work surface for closed-loop feedback. The control system is implemented by an edge computing node, which integrates an audio processing unit and a lighting driver interface.

[0020] Sound Acquisition and Preprocessing A microphone array synchronously picks up acoustic signals within the space at a sampling rate of 48 kHz. Each signal first passes through a bandpass filter from 20 Hz to 20 kHz to filter out infrasound and ultrasonic interference. The continuous signal is then framed with a frame length of 1024 points, with adjacent frames overlapping by 50%. Each frame's data is multiplied by a Hanning window to mitigate leakage effects in subsequent spectral analysis. The preprocessed multi-channel audio frames are then fed into a feature extraction module.

[0021] Acoustic Feature Extraction Based on Fast Fourier Transform and Dynamic Threshold A radix-2 Fast Fourier Transform (FFT) was performed on the 1024-point time-domain sequence of each frame to obtain a linear spectrum, which was then converted into a power spectrum. A one-third octave band division strategy was adopted, dividing the frequency range from 20 Hz to 20 kHz into 24 sub-bands, and the A-weighted sound pressure level (SPL) of each sub-band was calculated. The A-weighted equivalent continuous SPL was obtained by integrating the full-band A-weighted energy; the arithmetic mean of the SPLs of the four octave bands with center frequencies of 500 Hz, 1000 Hz, 2000 Hz, and 4000 Hz was taken as the speech interference level. The reverberation time was obtained by performing a Schroeder inverse integral on the power spectrum, blindly estimating the time required for 60 dB of sound energy attenuation in the frequency domain. Simultaneously, the spectral centroid and spectral entropy of the power spectrum were calculated; the former reflects the energy concentration band, and the latter characterizes the randomness of the signal.

[0022] The sound event detection employs a dynamic thresholding mechanism. A background noise baseline estimator is maintained for each sub-band, using a smoothing filter with a one-second time constant to track the local minimum of the sub-band energy, which serves as the current noise baseline. The dynamic threshold for each sub-band is set to the noise baseline plus a 6-dB sensitivity offset. When the energy of a sub-band exceeds its dynamic threshold for 30 consecutive milliseconds, a sound event start point is recorded. The system uses a one-second statistical window to accumulate the number of sound event start points to obtain the sound event density, and calculates the standard deviation of the time interval between adjacent sound events as the rhythm intensity. This dynamic threshold can adaptively increase or decrease with the start / stop of air conditioning or the intrusion of outdoor traffic noise, ensuring accurate sound event detection in both quiet and noisy periods.

[0023] Introducing environment-adaptive online scene classification A feature vector is extracted from each second of audio frame, containing seven dimensions: equivalent continuous sound pressure level, speech interference level, reverberation time, spectral centroid, spectral entropy, sound event density, and rhythm intensity. This vector is fed into a pre-trained lightweight gradient boosting decision tree classifier, which outputs a label for the current activity scene, including: quiet office, multi-person meeting, music playback, leisure and social interaction, and unattended energy saving.

[0024] An environmentally adaptive dynamic threshold mechanism was introduced during the classification process. Specifically, the background noise level, calculated from the background noise floor, is monitored. When the change exceeds 5 dB, the system automatically adjusts the equivalent sound pressure level threshold for distinguishing between "quiet office" and "multi-person meeting," preventing the noise from air conditioning from misinterpreting a quiet office environment as whispering activity. For the determination of "unmanned energy-saving" scenarios, dual dynamic conditions are used: all sub-band energy must remain below the adaptive silence threshold (i.e., the noise floor plus 3 dB) for at least 5 minutes, and the sound event density within the last 5 minutes must be less than 2 times per minute. Only when both conditions are met simultaneously is the scenario switched to unmanned energy-saving mode, effectively preventing situations where people briefly leave their seats and then turn off the lights.

[0025] Acousto-optic mapping that integrates dynamic gain, glare suppression, and double buffering The edge computing node stores a set of sound and light mapping knowledge base, which presets basic light parameters for various scenarios: quiet office corresponds to a work surface illuminance of 500 lux and a correlated color temperature of 4000 Kelvin; multi-person meetings correspond to an illuminance of 600 lux and a color temperature of 4000 Kelvin; music playback corresponds to a basic illuminance of 300 lux and a color temperature of 3000 Kelvin, and allows the overlay of dynamic effects; leisure and social activities correspond to an illuminance of 400 lux and a color temperature of 3500 Kelvin; and unattended energy-saving illuminance corresponds to an illuminance of 50 lux and a color temperature of 3000 Kelvin.

[0026] Based on this, the final target light parameters are generated through the following three mechanisms in combination: First, the dynamic gain factor. The short-term change in A-weighted sound pressure level over the last 200 milliseconds is taken. Its absolute value is incremented by one, the logarithm is taken, multiplied by a scene-specific gain coefficient, and then incremented again to obtain the dynamic gain factor. The gain coefficient is set to 0.2 for meeting scenes, 0.8 for music scenes, 0.4 for casual social scenes, and 0 for quiet office scenes. The gain factor is limited to between 1.0 and 2.0. Multiplying the base illuminance value by the gain factor yields an initial increase in brightness, allowing the light to subtly fluctuate at musical rhythms or emphasized speech points, enhancing the sense of presence.

[0027] Second, the glare suppression function. For each lamp, based on its installation height of 2.8 meters, beam angle of 30 degrees, and observer's seated eye height of 1.2 meters, a glare safety threshold brightness is pre-calculated, corresponding to 80% of the maximum brightness. If the initial brightness increase exceeds this safety threshold, glare suppression is triggered: a hyperbolic tangent function is used for smooth compression, causing the portion exceeding the safety threshold to grow along a curve gradually approaching the maximum permissible brightness. The softening coefficient during the compression process is set to 0.2 times the difference between the maximum permissible brightness and the safety threshold. Therefore, even during musical climaxes or sudden high sound pressure levels, the lamps will not produce dazzling glare.

[0028] Third, a dual-buffering mechanism is implemented. Two buffers are established in the control node's memory: the current buffer stores the currently executing brightness and color temperature values; the target buffer contains the latest target brightness and color temperature calculated using the mapping, gain, and suppression methods described above. An S-shaped easing function is used for the transition, with a transition duration set to 1.5 seconds. Every 20 milliseconds, an intermediate interpolation parameter is generated based on the values ​​in the current and target buffers and output to the lamp driver. During the 1.5-second transition period, even if the target buffer is updated due to scene changes or new sound events, the transition process will not be interrupted or jittered. The new target value will only begin the next round of smooth gradation after the current transition is completed, completely eliminating flicker and abruptness during dimming.

[0029] Closed-loop dimming execution and optional sound source zoning The intermediate illuminance and color temperature values ​​output by the double buffering mechanism are converted into dimming commands by the digitally addressable lighting interface and sent to the drivers of each LED luminaire. At the same time, the illuminance sensor on the working surface collects the actual illuminance every 500 milliseconds, compares it with the current target illuminance, and inputs the difference into the incremental proportional-integral-derivative controller to fine-tune the luminaire output to compensate for changes in outdoor sunlight and long-term light decay of the luminaire.

[0030] This embodiment also enables regional sound source localization and zoned dimming. Using a microphone array to receive signals, the azimuth angle of the main sound source is estimated through a generalized cross-correlation phase transform algorithm. When continuous speech is detected in a certain direction, the system automatically increases the illuminance of the two lamps corresponding to that direction by 20% above the base value, while maintaining the base illuminance in other areas. The increased brightness in this area is also subject to glare suppression and a smooth transition through a double buffering mechanism, achieving a focused lighting effect where "light follows sound" while ensuring visual comfort.

[0031] Running Example One afternoon, in an unoccupied conference room, the background noise floor remained stable at around 30 dB, with all sub-band energies below 33 dB for more than 5 minutes. The sound event density was zero, and the system was operating in an unoccupied, energy-saving scenario. The lights were in standby mode with low brightness (50 lux, 3000 Kelvin) warm light. Subsequently, four people entered the conference room and began conversing. The sound event density rapidly increased to 5 times per second, with the equivalent continuous sound pressure level reaching 58 dB. The spectral center of gravity showed concentrated mid-frequency energy, and the classifier identified it as a multi-person meeting scenario within one second. The target buffer was assigned 600 lux and 4000 Kelvin. The dual-buffering mechanism smoothly increased the conference room illuminance to the target value within 1.5 seconds, with the light color transitioning evenly from warm white to neutral white. During the conversation, when someone suddenly raised their voice to emphasize their point, the short-term sound pressure level change rate caused the dynamic gain factor to momentarily reach 1.3, resulting in a barely perceptible brief increase in light brightness before returning to normal, enhancing the expressive effect. All brightening processes did not exceed the glare safety threshold. An hour later, the meeting switched to music playback. The system detected an increase in spectral entropy and a stronger rhythm intensity, so the scene changed to music playback. The base illuminance dropped to 300 lux, the color temperature changed to 3000 Kelvin, and the gain coefficient increased to 0.8. The lights began to gently pulsate between 300 and 600 lux in sync with the music rhythm, and the glare suppression function ensured that the peak brightness remained comfortable. The music stopped, and everyone left. After 5 minutes of silence, both conditions were met, and the system smoothly transitioned back to an unattended, energy-saving state, achieving fully automated intelligent lighting control.

[0032] like Figure 1 As shown, under unmanned energy-saving conditions, during acoustic feature extraction, the subband energy is 26.8dB, the dynamic threshold is 19.5+6dB, and the sound event density is 0.5 / s; during scene recognition and double-buffered smooth dimming, the original target is 50lx, the smoothed output is 50lx, and the gain is 1.00; the glare suppression function and operating point are: glare safety threshold of 80% of maximum brightness, current input 6.7%, and output 6.7%.

[0033] like Figure 2As shown, under quiet office conditions, during acoustic feature extraction, the subband energy is 49.9dB, the dynamic threshold is 28.0+6dB, and the sound event density is 1.2 / s; during scene recognition and double-buffered smooth dimming, the original target is 513lx, the smoothed output is 519lx, and the gain is 1.03; the glare suppression function and operating point are: glare safety threshold 80% of maximum brightness, current input 69.9%, and output 69.2%.

[0034] like Figure 3 As shown, during music playback, the subband energy is 76.5dB, the dynamic threshold is 38.6+6dB, and the sound event density is 7.9 / s during acoustic feature extraction; during scene recognition and double-buffered smooth dimming, the original target is 539lx, the smoothed output is 458lx, and the gain is 1.80; the glare suppression function and operating point are: glare safety threshold 80% of maximum brightness, current input 51.4%, and output 58.8%.

[0035] Figure 1 (a) Figure 2 (a) and Figure 3 Image (a) shows both the sound pressure waveform and an adaptively changing dynamic threshold line. Figure 2 (a) and Figure 3 The trigger point for the sound event is also marked in (a). This indicates that: The dynamic threshold is not a fixed value, but automatically fluctuates based on the real-time estimated background noise floor. The threshold rises during noisy periods and falls during quiet periods, thus avoiding false positives or false negatives. Only when the subband energy exceeds the current dynamic threshold and maintains the minimum duration is it considered a valid sound event (dots in the diagram), thereby reliably calculating sound event density and rhythm intensity. During scene transitions, both waveform amplitude and threshold change accordingly. For example, in a music scene, the waveform is dense and the amplitude is large, so the threshold rises accordingly, demonstrating the robustness of environmental adaptation.

[0036] Figure 1 (b) Figure 2 (b) and Figure 3 In (b), three curves represent the original target illuminance, the smoothed illuminance output after double buffering, and the dynamic gain factor, respectively. This illustrates that: The original target illuminance fluctuates frequently due to changes in sound pressure level and dynamic gain factor, but the actual output illuminance is extremely smooth due to double-buffered interpolation, without abrupt changes or flicker. When switching scenes (e.g., from unattended energy-saving mode to multi-person conference mode), the smooth illuminance gradually transitions to the new target value in an S-shaped motion, with adjustable transition time, completely eliminating the abrupt transition visually. The dynamic gain factor is calculated in real time with the rate of change of sound pressure. In music scenes, the gain coefficient is high, and the light produces expressive changes in brightness and darkness according to the rhythm, while in conference scenes, the gain coefficient is low, and the lighting remains stable without being distracting.

[0037] Figure 1 (c) Figure 2 (c) and Figure 3 Figure (c) shows the linear mapping line without suppression and the actual output curve after introducing glare suppression, with the current state marked by the operating point. It illustrates that: When the brightness does not exceed the safety threshold (80% of maximum brightness in this case), the output remains consistent with the input. Once the input exceeds the safety threshold, the output is smoothly compressed by a hyperbolic tangent function, gradually approaching the maximum allowable brightness without abruptly truncating. The current operating point directly reflects whether the system is triggering glare suppression: in scenarios such as during musical climaxes, the operating point may exceed the threshold, but the actual luminaire brightness is safely constrained to ensure eye comfort. The curve is smooth with no inflection point, indicating that glare protection is not a simple limiting mechanism, but rather a gradual compression that follows the characteristics of human visual perception.

[0038] Therefore, the present invention adopts the above-mentioned intelligent dimming control method for sound and light linkage in architectural space, which integrates multi-dimensional acoustic feature extraction, dynamic threshold scene classification and sound and light mapping model with comfort constraints to achieve synergistic improvement of lighting quality and energy-saving optimization.

[0039] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for intelligent dimming control of sound and light linkage in architectural space, characterized in that, Includes the following steps: Multi-channel sound signals are acquired by a microphone array deployed in the building space at a preset sampling rate, and the sound signals are preprocessed, including bandpass filtering, framing, and windowing. The preprocessed acoustic signal is subjected to a fast Fourier transform to obtain the spectrum. The frequency range is divided into multiple sub-bands using a frequency band division strategy, and the energy of each sub-band is calculated. Acoustic features are extracted, and acoustic events are detected based on a dynamic threshold. The dynamic threshold is adaptively adjusted according to the real-time estimated background noise floor. The extracted acoustic feature vectors are input into the classifier, which outputs the current activity scene label. An environment-adaptive dynamic threshold mechanism is introduced during the classification process to adjust the scene judgment threshold according to changes in background noise. Based on the scene label, the basic light parameters are obtained from the preset mapping library, and a dynamic gain factor, glare suppression function and double buffering mechanism are introduced to generate the final target light parameters; The target light parameters are converted into dimming signals to drive the lamps, and the actual illuminance and color temperature are collected by an ambient light sensor for closed-loop feedback control.

2. The intelligent dimming control method for sound and light linkage in architectural space according to claim 1, characterized in that, Acoustic characteristics include A-weighted sound pressure level, speech interference level, reverberation time, spectral centroid, and spectral entropy.

3. The intelligent dimming control method for sound and light linkage in architectural space according to claim 1, characterized in that, The frequency band division strategy adopts one-third octave band division or psychoacoustic critical band division. The speech interference level in the acoustic features is obtained by calculating the average sound pressure level of the octave band from 500 Hz to 4 kHz, and the reverberation time is obtained by the frequency domain blind estimation method.

4. The intelligent dimming control method for sound and light linkage in architectural space according to claim 1, characterized in that, The sound event detection based on dynamic threshold determination specifically includes: real-time estimation of the noise floor of a sub-band, setting a dynamic threshold for the sub-band consisting of the noise floor plus a frequency-related offset, and determining a sound event start point when the energy of a sub-band exceeds its dynamic threshold and the duration exceeds the preset minimum duration, and statistically analyzing the sound event density per unit time and the standard deviation of the event interval as the rhythm intensity.

5. The intelligent dimming control method for sound and light linkage in architectural space according to claim 1, characterized in that, The environmental adaptive dynamic threshold mechanism includes: when the background noise floor changes significantly, automatically adjusting the sound pressure level judgment threshold used to distinguish between quiet scenes and low-noise activity scenes; when judging unmanned energy-saving scenes, using a sound event density threshold and a continuous silence duration threshold to form a dual dynamic judgment condition, only switching to unmanned energy-saving scenes when all sub-band energy is continuously lower than the adaptive silence threshold for more than a preset duration and the sound event density is lower than the preset density threshold.

6. The intelligent dimming control method for sound and light linkage in architectural space according to claim 1, characterized in that, The dynamic gain factor is calculated based on the short-term sound pressure level change rate. The gain factor has a logarithmic relationship with the change rate and is limited to a preset minimum and maximum value. Different activity scenarios correspond to different gain coefficients. Music scenarios use a higher gain coefficient, while language scenarios use a lower gain coefficient.

7. The intelligent dimming control method for sound and light linkage in architectural space according to claim 1, characterized in that, The glare suppression function works as follows: when the initial target brightness of a lamp calculated based on the acousto-optic mapping exceeds the glare safety threshold determined based on the lamp installation height, beam angle, and observer position, the hyperbolic tangent function is used to smooth and compress the brightness, so that the brightness exceeding the safety threshold gradually approaches the preset maximum allowable brightness, thereby limiting glare.

8. The intelligent dimming control method for sound and light linkage in architectural space according to claim 1, characterized in that, The dual buffering mechanism is as follows: the current lighting parameter buffer and the target lighting parameter buffer are maintained in the control unit. Intermediate lighting parameters that gradually change from the current parameters to the target parameters are generated through linear interpolation or S-shaped easing interpolation algorithm within a preset transition time. During the transition, the target buffer is updated and protected to prevent jitter caused by the writing of new calculation results during the transition.

9. A method for intelligent dimming control of sound and light linkage in architectural space according to any one of claims 1 to 8, characterized in that, It also includes a zone dimming step: using the microphone array to determine the location of the main sound source through a direction-of-arrival estimation algorithm, dynamically increasing the illumination brightness of the area corresponding to the sound source location, while maintaining the basic illumination brightness of other areas, and outputting the increased illumination brightness after being limited by the glare suppression function and smoothly transitioned by the double buffering mechanism.