A conference room idle recognition and resource reservation release method and system

Abstract

The application discloses a conference room idle recognition and resource reservation release method and system, non-semantic acoustic features are extracted from environmental sound signals in the conference room, the non-semantic acoustic features contain sound pressure statistics and quiet duration, and do not contain semantic content; an unattended environmental noise baseline corresponding to the conference room is established according to the non-semantic acoustic features, and is updated according to time periods and equipment operating states; a suspected idle event is generated according to a comparison result of the non-semantic acoustic features and the unattended environmental noise baseline, target reservation records are obtained by associating reservation records, confirmation notifications are sent to corresponding reservation objects and response information is received, or recheck instructions are generated when the response information is not returned, recheck results are obtained by controlling an acoustic monitoring device to perform recheck collection, and then conference room resource release instructions are generated according to the response information or the recheck results, and the reservation state is updated, so that accurate idle recognition and reliable resource release are realized.

Description

Technical Field

[0001] This invention relates to the field of office resource management technology, and in particular to a method and system for identifying meeting room vacancy and releasing reserved resources. Background Technology

[0002] Current meeting room management systems generally rely on reservation systems to mark occupancy status. Users pre-select meeting rooms and time slots, and the system registers occupancy accordingly, automatically resetting availability after the reservation ends. However, reservation records only reflect planned usage and can deviate from actual usage. When the person who made the reservation doesn't arrive, the meeting ends early, or the reservation is forgotten, the meeting room may be actually vacant, but the system still shows it as occupied, resulting in wasted resources that cannot be reserved by others. To address this, the industry has proposed installing human infrared, temperature, door magnetic, or chair pressure sensors in meeting rooms, or using cameras and facial recognition, access control and terminal check-in, and monitoring the operating status of electrical equipment to determine whether a meeting room is occupied. While these solutions can reflect actual occupancy to some extent, they all rely on the visible presence of people or equipment, requiring dedicated hardware deployment, which presents problems such as high investment, complex installation and modification, and high maintenance costs. Furthermore, when images and personal information are involved, privacy concerns are more likely to arise. Moreover, when participants sit quietly for extended periods without obvious movement or equipment operation, these methods struggle to distinguish between true vacancy and silent usage, easily leading to misjudgments. Summary of the Invention

[0003] In view of this, the purpose of this invention is to propose a method and system for identifying meeting room vacancy and releasing reserved resources. By comparing non-semantic acoustic features with an unmanned noise baseline and confirming and re-checking, the system can accurately identify vacancy and reliably release reserved resources.

[0004] To achieve the aforementioned technical objectives, in a first aspect, the technical solution adopted by the present invention is: a method for identifying and releasing reserved meeting room vacancy resources, comprising: Acoustic monitoring devices are installed in the conference room to acquire ambient sound signals. Non-semantic acoustic features are extracted from environmental sound signals. These features include sound pressure statistics and quiet duration, but do not contain semantic content. A baseline of unmanned ambient noise for the conference room is established based on non-semantic acoustic features, and the baseline is updated according to time period and equipment operating status. Based on the comparison results of non-semantic acoustic features and the noise baseline of the unoccupied environment, suspected idle events of the conference room are generated; Based on the reservation records of meeting rooms associated with suspected vacancy events, the target reservation record is obtained; Send a confirmation notification to the corresponding reservation recipient based on the target reservation record, and receive the response information returned by the reservation recipient based on the confirmation notification; or, if the response information is not returned, generate a re-inspection instruction based on the suspected idle event. The acoustic monitoring device is controlled to perform re-inspection data collection according to the re-inspection command, and the re-inspection results are obtained. Generate a meeting room resource release instruction based on the response information or re-inspection results; Update the meeting room reservation status according to the meeting room resource release instruction, and mark the meeting room as unconfirmable and terminate the generation of the meeting room resource release instruction when the acoustic monitoring device is offline or the non-semantic acoustic features are not updated, and output the status update result.

[0005] In some embodiments, non-semantic acoustic features are extracted based on ambient sound signals, including: The ambient sound signal is windowed and framed according to the set frame length and frame shift, and adjacent acoustic frames in time are overlapped and connected to obtain an acoustic frame sequence. Perform a short-time Fourier transform on each acoustic frame in the acoustic frame sequence to transform the acoustic frame from the time domain to the frequency domain, and obtain the short-time energy spectrum and spectral distribution characteristics of the corresponding acoustic frame; The logarithmic mean of the sound pressure level is obtained by taking the logarithm of the short-time energy spectrum of each acoustic frame, summing them in time, and taking the mean. The dispersion statistics of the spectral distribution characteristics in the time dimension are obtained to obtain the sound fluctuation amplitude, and the sound pressure statistics are composed of the logarithmic mean of the sound pressure and the sound fluctuation amplitude. Autocorrelation analysis is performed on the acoustic frame sequence to obtain the fundamental frequency trajectory, and the spectral flatness is calculated based on the spectral distribution characteristics. Based on the continuity of the fundamental frequency trajectory and the comparison between the spectral flatness and the set activity decision threshold, continuous human voice activity is determined, and human voice activity identifier is obtained. Based on the human voice activity identifier, acoustic frames without continuous human voice activity are identified as quiet frames, and consecutively adjacent quiet frames in time are accumulated to obtain the continuous quiet time. The sound pressure statistics, human activity indicators, and continuous quiet time are aligned and combined using the same temporal index to obtain non-semantic acoustic features, while discarding the original waveform and acoustic frame sequence of the environmental sound signal.

[0006] In some embodiments, an unoccupied environmental noise baseline for the conference room is established based on non-semantic acoustic features, and the unoccupied environmental noise baseline is updated according to time period and equipment operating status, including: The meeting room is confirmed to be unoccupied based on the duration of quiet in the non-semantic acoustic features. During the collection period corresponding to the unoccupied status, the non-semantic acoustic features are determined as the baseline sample features, and abnormal fluctuation samples that deviate from the overall baseline sample features are removed. Based on the usage time of the meeting room, the characteristics of the removed baseline samples are divided into time periods to obtain a time period sample set; The time-period average value and fluctuation range of the sound pressure statistics of non-semantic acoustic features in the time-period sample set are obtained to obtain the time-period baseline component. By associating the baseline components of each time period with the corresponding time periods, the unmanned environmental noise baseline for the conference room is obtained. Obtain the equipment operating status and current time period in the conference room; When the device operating status changes, or when the current time period is inconsistent with the time period associated with the time period baseline component, the non-semantic acoustic features under the current time period are determined as updated samples; The time period baseline components are updated using a weighted fusion method based on the updated samples, and the update of the unmanned environmental noise baseline is triggered by the cumulative deviation between the updated samples and the time period baseline components, so that different meeting rooms have their own corresponding unmanned environmental noise baselines.

[0007] In some embodiments, based on a comparison of non-semantic acoustic features with an unoccupied ambient noise baseline, a suspected idle event for the conference room is generated, including: Extract the baseline sound pressure statistics and baseline fluctuation range corresponding to the current time period from the unmanned environmental noise baseline; The sound pressure offset is obtained by subtracting the sound pressure statistics in the non-semantic acoustic features from the reference sound pressure statistics. The discreteness of the sound pressure statistics in the non-semantic acoustic features is calculated in the time domain to obtain the temporal fluctuation quantity. The temporal fluctuation quantity is then compared with the baseline fluctuation range to obtain the fluctuation comparison result. Based on the comparison results of the sound pressure offset with the set sound pressure offset threshold, the comparison results of the quiet duration in non-semantic acoustic features with the set quiet duration threshold, and the fluctuation comparison results, an idle judgment condition set is generated according to the logical AND relationship. If all conditions in the idle determination condition set remain true within a set time period, a suspected idle indicator for the meeting room is generated. Based on the suspected idle identifier, a suspected idle event for the meeting room is generated. The suspected idle event includes the meeting room identifier, detection time, sound pressure offset, quiet duration, and equipment operating status of the meeting room. Among them, the sound pressure offset threshold and the quiet duration threshold are dynamically generated based on the historical data of the conference room and are adjusted as the noise baseline of the unmanned environment is updated.

[0008] In some embodiments, the target reservation record is obtained based on the reservation records of the meeting room associated with the suspected vacancy event, including: Search for the meeting room reservation records based on the meeting room icons in suspected vacancy events; The detection time in the suspected idle event is matched with the reservation time period of each reservation record in the reservation record set to determine whether the detection time falls within the reservation time period and obtain the matching reservation record. The validity of the matched reservation records is filtered based on their reservation status. If the reservation status is canceled or ended, or if the meeting room identifier corresponding to the matched reservation record is inconsistent with the meeting room identifier in the suspected available event, the corresponding matched reservation record is excluded, and the valid reservation records are obtained. When there are multiple valid appointment records, calculate the degree of overlap between the appointment time and the detection time in each valid appointment record, and determine the valid appointment record with the largest degree of overlap as the target appointment record; When there is only one valid reservation record, the valid reservation record will be identified as the target reservation record. When there are no valid reservation records, update the actual status of the meeting room to be available for reservation based on suspected idle events; The target appointment record includes the appointment person's identifier, the appointment end time, and the remaining appointment time slot, which is determined by the appointment end time and the testing time.

[0009] In some embodiments, a confirmation notification is sent to the corresponding reservation recipient based on the target reservation record, and response information is received from the reservation recipient based on the confirmation notification, including: Extract the reservation person's identifier and reservation end time from the target reservation record, and query the communication session identifier of the reservation person in the instant messaging platform based on the reservation person's identifier. If the communication session identifier is missing, determine the communication session identifier based on the backup notification method corresponding to the reservation person's identifier. A confirmation notification is generated based on the detection time and reservation end time in the suspected idle event. The confirmation notification includes the meeting room name, reservation time, suspected idle time, operation prompts and operation options. The operation options include the option to continue using, the option to release immediately and the option to release later. The confirmation notification does not carry ambient sound signals. A confirmation notification is sent to the reservation recipient through the instant messaging channel corresponding to the communication session identifier. The delivery status of the confirmation notification is bound to the response waiting timer, and the confirmation notification is resent when the delivery status is not delivered. Receive the response information returned by the reservation recipient based on the confirmation notification within the response waiting time; The operation options selected by the reservation object and the time of receiving the response information are parsed from the response information, and the parsed operation options and receiving times are associated with the target reservation record to obtain the response parsing result.

[0010] In some embodiments, when no response information is returned, a re-inspection instruction is generated based on a suspected idle event, including: If the response timer for the confirmation notification expires and no response is returned, and the delivery status of the confirmation notification is "delivered", a non-response determination result is generated. Based on the non-response determination result, query the online status of the acoustic monitoring device and the update status of non-semantic acoustic features; When the online status is online and the update status is normal, extract the meeting room identifier and detection time from the suspected idle events, and extract the comparison results of non-semantic acoustic features and unmanned environmental noise baseline; A re-inspection instruction is generated based on the meeting room identifier, detection time, and comparison results. The re-inspection instruction includes the re-inspection trigger time, re-inspection duration, re-inspection collection items, and re-inspection threshold. The re-inspection threshold is consistent with the threshold corresponding to the comparison results, and the re-inspection collection items correspond to the constituent items of the non-semantic acoustic features. The re-inspection command is associated with the suspected idle event to obtain the associated re-inspection command. The associated re-inspection command is used to instruct the acoustic monitoring device to extract non-semantic acoustic features again according to the re-inspection collection items. When the online status is offline or the update status is interrupted, mark the meeting room as unconfirmable, terminate the generation of re-inspection instructions, and output the unconfirmable status.

[0011] In some embodiments, the acoustic monitoring device is controlled to perform re-inspection data acquisition according to the re-inspection command to obtain the re-inspection results, including: Based on the re-inspection instruction, acquisition control parameters are sent to the acoustic monitoring device. The acquisition control parameters include acquisition time, acquisition window, acquisition feature items and acquisition number. The acoustic monitoring device collects the sound signal of the conference room within the acquisition window at the acquisition time, and performs feature statistics on the sound signal according to the acquired feature items to extract the acoustic features for re-examination. Based on the comparison results of suspected idle events associated with the re-inspection instruction, the acoustic characteristics of the re-inspection are compared with the noise baseline of the unmanned environment using the same aperture to obtain the re-inspection comparison results. When the re-inspection comparison results indicate that the acoustic characteristics of the re-inspection are still within the range defined by the unmanned environmental noise baseline, a re-inspection result representing the idle period is generated. When the sound pressure statistics in the re-examined acoustic features exceed the range defined by the unmanned environmental noise baseline, the re-examination results used to characterize the process are generated. When the acoustic monitoring device interrupts acquisition within the acquisition window or the update of the re-examination acoustic features is interrupted, a re-examination result characterized as unconfirmable is generated. The re-inspection results include the acoustic characteristics of the re-inspection, the re-inspection comparison results, and the re-inspection mark of the conference room. The re-inspection mark is determined by the conference room mark in the suspected idle event and the detection time.

[0012] In some embodiments, the reservation status of the meeting room is updated according to the meeting room resource release instruction, and the meeting room is marked as unconfirmable and the generation of the meeting room resource release instruction is terminated when the acoustic monitoring device is offline or the non-semantic acoustic features are not updated. The status update result is output, including: Based on the release intention indicated by the response information or the idle conclusion represented by the re-inspection results, a release determination result is generated, and a meeting room resource release instruction is generated when the release determination result indicates release; The resource release command for the meeting room is validated based on the meeting room's attributes. If the meeting room's attributes are in reminder mode only, the resource release command is stopped and only the reminder result is output. When the meeting room is in automatic release mode, the meeting room reservation interface is called according to the meeting room resource release command to cancel the remaining reservation time slots of the meeting room; Based on the result returned by the reservation interface, update the reservation status of the meeting room from occupied to available, and obtain the reservation status update result; When the acoustic monitoring device is offline, the network is interrupted, the non-semantic acoustic features are not updated, or the re-inspection results are unconfirmable, the meeting room will be marked as unconfirmable, and the generation of meeting room resource release instructions and the updating of reservation status will be terminated. Based on the reservation status update results and the unconfirmed status, generate and output status update results. The status update results include the meeting room identifier, reservation status, release duration, and number of suspected idle times.

[0013] In a second aspect, the present invention also provides a meeting room vacancy identification and reservation resource release system for executing the method described in the first aspect. The system includes an acoustic monitoring device, a baseline management unit, a vacancy identification unit, a reservation association unit, a notification interaction unit, a re-inspection control unit, and a release execution unit. The acoustic monitoring device is located in the meeting room and includes a sound acquisition module, a sound pressure detection module, a processing module, a communication module, and a power supply module. The sound acquisition module is used to acquire ambient sound signals within the meeting room. The sound pressure detection module and the processing module are used to extract non-semantic acoustic features based on the ambient sound signals. The communication module is used to upload the non-semantic acoustic features. The baseline management unit is used to establish an unoccupied ambient noise baseline corresponding to the meeting room based on the non-semantic acoustic features and to update the unoccupied ambient noise baseline according to the time period and equipment operating status. The vacancy identification unit is used to identify non-semantic acoustic features and... The comparison results of the unmanned environmental noise baseline generate suspected idle events for the meeting room; the reservation association unit is used to associate the reservation records of the meeting room with the suspected idle events to obtain the target reservation record; the notification interaction unit is used to send a confirmation notification to the corresponding reservation object according to the target reservation record and receive the response information returned by the reservation object based on the confirmation notification; the re-inspection control unit is used to generate a re-inspection command based on the suspected idle events when the response information is not returned, and controls the acoustic monitoring device to perform re-inspection acquisition according to the re-inspection command to obtain the re-inspection result; the release execution unit is used to generate a meeting room resource release command according to the response information or re-inspection result, update the reservation status of the meeting room according to the meeting room resource release command, and mark the meeting room as unconfirmable and terminate the generation of the meeting room resource release command when the acoustic monitoring device is offline or the non-semantic acoustic features are not updated, and output the status update result.

[0014] By adopting the above technical solution, the present invention has the following beneficial effects compared with the prior art: The present invention provides a method and system for identifying vacancy and releasing reserved resources in a conference room. It extracts non-semantic acoustic features from the environmental sound signals in the conference room. These non-semantic acoustic features include sound pressure statistics and quiet duration, but do not contain semantic content. A baseline of unoccupied environmental noise corresponding to the conference room is established based on the non-semantic acoustic features and updated according to the time period and equipment operating status. A suspected vacancy event is generated based on the comparison result between the non-semantic acoustic features and the unoccupied environmental noise baseline. The target reservation record is obtained by associating the reservation record with the corresponding reservation record. A confirmation notification is sent to the corresponding reservation object and a response information is received. Alternatively, a re-inspection command is generated if the response information is not returned. The acoustic monitoring device is controlled to perform a re-inspection and collect the re-inspection results. Then, a conference room resource release command is generated based on the response information or the re-inspection results, and the reservation status is updated, thus achieving accurate vacancy identification and reliable resource release. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of steps S101 to S109 of the method described in the specific implementation embodiment; Figure 2 This is a schematic diagram of steps S201 to S207 of the method described in the specific implementation. Detailed Implementation

[0017] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be particularly noted that the following embodiments are for illustrative purposes only and do not limit the scope of the invention. Similarly, the following embodiments are only some, not all, embodiments of the present invention, and all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1 In a first aspect, this embodiment provides a method for identifying meeting room vacancy and releasing reserved resources, including: S101. Acquire ambient sound signals in the conference room; the conference room is equipped with an acoustic monitoring device. S102. Extract non-semantic acoustic features based on environmental sound signals. Non-semantic acoustic features include sound pressure statistics and quiet duration, but do not contain semantic content. S103. Establish an unmanned environmental noise baseline for the conference room based on non-semantic acoustic features, and update the unmanned environmental noise baseline according to time period and equipment operating status. S104. Based on the comparison results between non-semantic acoustic features and the noise baseline of the unoccupied environment, generate suspected idle events in the conference room; S105. Based on the reservation records of the meeting rooms associated with the suspected vacancy events, obtain the target reservation record; S106. Send a confirmation notification to the corresponding reservation object according to the target reservation record, and receive the response information returned by the reservation object based on the confirmation notification, or generate a re-inspection instruction based on the suspected idle event if the response information is not returned. S107. Control the acoustic monitoring device to perform re-inspection data collection according to the re-inspection instruction, and obtain the re-inspection results; S108. Generate a meeting room resource release instruction based on the response information or re-inspection results; S109. Update the reservation status of the meeting room according to the meeting room resource release instruction, and mark the meeting room as unconfirmable if the acoustic monitoring device is offline or the non-semantic acoustic features are not updated, and terminate the generation of the meeting room resource release instruction, and output the status update result.

[0019] In step S101, the acoustic monitoring device is responsible for the acquisition and preliminary processing of ambient sound signals. It can be positioned in the middle of the ceiling or above the conference table, where it can uniformly pick up the indoor sound field, avoiding the introduction of localized airflow noise due to proximity to air conditioning vents or doors and windows. The acquisition of ambient sound signals can be triggered based on the occupancy status of the conference room reservation system; that is, acquisition only starts when the conference room is registered as occupied, and the acquisition frequency is reduced or paused when it is not registered as occupied. The acquired sound signals are organized into time windows of a set duration, with each time window corresponding to a collection cycle, forming the sole data source for all subsequent feature extraction and status determination.

[0020] In step S102, the extraction target of non-semantic acoustic features is the environmental sound signal itself, not its speech content. The extraction process does not require speech recognition or speaker identification, nor does it generate any text transcription results. Sound pressure level statistics characterize the sound energy level within the acquisition window and its fluctuation over time, distinguishing between steady-state background noise and intermittent sound. Quiet duration is counted based on acquisition windows that continuously meet the quiet condition. Non-semantic acoustic features are extracted locally on the acoustic monitoring device; the original waveform and intermediate data are discarded after feature generation.

[0021] In step S103, the unoccupied environmental noise baseline characterizes the inherent level of background noise in a specific meeting room when it is confirmed to be unoccupied. It is established using collected data from areas confirmed to be unoccupied as the sample source. The baseline can be organized by time period, allowing the same meeting room to have its own reference value in different time intervals, to accommodate background noise differences caused by the start and stop of equipment such as air conditioning and ventilation during different time periods. Equipment operating status characterizes whether equipment such as air conditioning, fresh air systems, and projectors are in operation. When the equipment operating status changes, the acoustic conditions corresponding to the original baseline have changed, triggering an update. The update can be performed by progressively correcting the original baseline with newly collected samples, allowing the baseline to evolve slowly with environmental changes and preventing abrupt changes due to occasional noise.

[0022] In step S104, the non-semantic acoustic features obtained in the current acquisition window are compared with the baseline values ​​of the corresponding time period in the unmanned environmental noise baseline. The comparison includes the level of acoustic energy and the magnitude of fluctuation. The result is characterized by the degree of deviation, representing the closeness between the current acoustic state and the unmanned state. The generation of suspected idle events does not rely on instantaneous comparison at a single moment, but requires that the comparison result continuously meets the idle condition within a set duration to avoid short pauses in the meeting being misjudged as idle. Suspected idle events serve as event-based records for state determination, containing determination information that can be called upon by subsequent steps. This step transforms the continuously acquired feature sequence into discrete events that can be identified and processed by downstream steps.

[0023] In step S105, the reservation records are sourced from the meeting room reservation system, recording the occupancy arrangements of each meeting room on a timeline. The target reservation record is determined based on information carried by suspected idle events as the search criteria. The system searches the reservation records for occupancy arrangements corresponding to that time point and filters out records that are still valid. When the search result is empty, it indicates that there are no planned occupancy arrangements for the meeting room. In this case, there is no need to proceed with the subsequent confirmation process; the actual status of the meeting room can be directly updated to be available for reservation. This step correlates the acoustic-level idleness determination with the business-level reservation arrangements, enabling subsequent processing to be tailored to specific reservation recipients.

[0024] In step S106, the recipient of the confirmation notification is determined by the reservation holder information in the target reservation record. The notification content is organized around elements such as the meeting room name, reservation time, and suspected available time, and includes operation options for the reservation holder to choose from. The response information is generated by the reservation holder based on the confirmation notification, and its parsing result is used to determine whether the reservation holder agrees to release. If no response information is received within the set waiting time, it indicates that the reservation holder may have left or not paid attention to the notification. At this time, a re-check instruction is generated to initiate a second verification, rather than directly executing the release. This step retains a channel for manual intervention in the automated process by setting up a reservation holder confirmation step.

[0025] In step S107, the acoustic monitoring device maintains the same data acquisition target and judgment caliber as the initial acquisition to ensure comparability between the two acquisition results. The re-inspection result is obtained by comparing the characteristics acquired during the re-inspection with the baseline noise of the unmanned environment. The result can be categorized into three types: indicating idleness, indicating use, and indicating unconfirmation, each corresponding to different subsequent processing branches. This step, through secondary verification of suspected idle states, compensates for the shortcomings of a single judgment that may be affected by occasional noise or brief silence.

[0026] In step S108, the generation of the meeting room resource release instruction is based on the release intention represented in the response information returned by the reservation object and the idle conclusion represented by the re-inspection result. The instruction is triggered if either of these conditions is met. This instruction, as an operation request to the reservation system, targets the remaining reservation time slot corresponding to the target reservation record.

[0027] In step S109, the reservation status is updated by calling the interface provided by the meeting room reservation system. The updated status makes the meeting room available for reservation again. The "unconfirmable" status is set to address situations where the data foundation upon which the determination relies is missing or invalid, including offline acoustic monitoring devices and non-semantic acoustic features not being updated on time. In such cases, the system lacks the conditions to make a reliable determination, and the generation of the release command is terminated to avoid performing irreversible resource operations when the data is incomplete. The status update result, as the final output data, includes the meeting room identifier and its current status, which can be displayed and statistically analyzed by the management platform.

[0028] This embodiment collects ambient sound signals from the conference room and extracts non-semantic acoustic features unrelated to semantics. It establishes and dynamically maintains an individualized unoccupied ambient noise baseline for the conference room, thereby generating suspected vacancy events and associating them with reservation records. Upon confirmation by the reservation recipient or in the absence of a response, a re-inspection is performed. Based on the confirmation or re-inspection results, a conference room resource release command is generated and the reservation status is updated. Simultaneously, if the device is offline or the features are not updated, an unconfirmable status is marked and the generation of the release command is terminated. This achieves the identification of the conference room's vacancy status and the reliable release of reserved resources without collecting meeting content or relying on cameras and human sensing devices.

[0029] Please see Figure 2 In some embodiments, non-semantic acoustic features are extracted based on ambient sound signals, including: S201. The ambient sound signal is windowed and framed according to the set frame length and frame shift, and adjacent acoustic frames in time are overlapped and connected to obtain an acoustic frame sequence. S202. Perform a short-time Fourier transform on each acoustic frame in the acoustic frame sequence to transform the acoustic frame from the time domain to the frequency domain, and obtain the short-time energy spectrum and spectral distribution characteristics corresponding to the acoustic frame. S203. Take the logarithm of the short-time energy spectrum of each acoustic frame, sum them in time sequence, and calculate the mean to obtain the logarithmic mean of the sound pressure. S204. Calculate the dispersion statistics of the spectral distribution characteristics in the time dimension to obtain the sound fluctuation amplitude, and use the logarithmic mean of the sound pressure and the sound fluctuation amplitude together to form the sound pressure statistics. S205. Perform autocorrelation analysis on the acoustic frame sequence to obtain the fundamental frequency trajectory, and calculate the spectral flatness based on the spectral distribution characteristics. Determine continuous human voice activity based on the continuity of the fundamental frequency trajectory and the comparison between the spectral flatness and the set activity decision threshold, and obtain the human voice activity identifier. S206. Based on the human voice activity identifier, acoustic frames without continuous human voice activity are identified as quiet frames, and sequentially adjacent quiet frames are cumulatively counted to obtain the continuous quiet time. S207. Align and combine the sound pressure statistics, human activity identifiers, and continuous quiet time with the same temporal index to obtain non-semantic acoustic features, and discard the original waveform and acoustic frame sequence of the environmental sound signal.

[0030] In step S201, the frame length is selected based on being able to cover more than one cycle of the fundamental frequency of human voices in the conference room. The frame shift is taken as one-half to one-quarter of the frame length, so that adjacent frames partially overlap in time. Windowed framing is achieved by multiplying each frame by a window function, which can be a Hamming window or a Hanning window. The overlapping connection of adjacent acoustic frames means that adjacent frames partially overlap in time according to the interval determined by the frame shift. The data of the overlapping part participates in the calculation of the preceding and following frames at the same time, thereby obtaining an acoustic frame sequence arranged along the time axis.

[0031] In step S202, the number of transformation points for the short-time Fourier transform is selected to be a value not less than the frame length, and the frequency resolution is determined by the ratio of the sampling rate to the number of transformation points. The short-time energy spectrum characterizes the energy distribution at each frequency point, and the spectral distribution characteristics characterize the amplitude variation with frequency. Both are derived from the same transformation result, providing a data basis for subsequent extraction of statistics in the time and frequency dimensions.

[0032] In step S203, the base used for taking the logarithm can be either base 10 or the natural logarithm. The object of the summation and averaging can be the energy value of each acoustic frame after frequency band aggregation. That is, first sum by frequency points within the frame or by grouping by a set frequency band, then take the logarithm of the energy of the frame and accumulate it along the time sequence, and divide it by the number of frames. The result reflects the overall sound energy level within the acquisition window.

[0033] In step S204, the dispersion statistic can be expressed as standard deviation, variance, or range. It is calculated from the sequence of acoustic frames arranged along the time dimension. Specifically, the spectral distribution characteristics of each frame are first obtained, and then the dispersion of these characteristics along time is calculated. The result characterizes the degree of sound energy fluctuation over time. The logarithmic mean of sound pressure and the amplitude of sound fluctuation together constitute the sound pressure statistic.

[0034] In step S205, autocorrelation analysis determines the fundamental frequency by calculating the autocorrelation function of the acoustic frame and finding its peak position. The reciprocal of the fundamental frequency period is the fundamental frequency of that frame. The fundamental frequencies of each frame are arranged in time sequence to form a fundamental frequency trajectory. The continuity of the fundamental frequency trajectory is determined based on whether the relative change in the fundamental frequency of adjacent frames exceeds a set range. Spectral flatness is obtained by the ratio of the geometric mean to the arithmetic mean of the spectrum. The activity decision threshold is determined based on the statistical distribution of spectral flatness in the conference room when it is unoccupied. When both spectral flatness and fundamental frequency trajectory continuity meet the decision requirements, it is determined that there is continuous human voice activity in that frame and a corresponding human voice activity identifier is assigned.

[0035] In step S206, the human voice activity identifier is a marker indicating whether there is continuous human voice activity in each acoustic frame. The cumulative count of consecutive quiet frames is interrupted when an acoustic frame with continuous human voice activity is encountered. The count is reset to zero and restarted from the subsequent quiet frame. The resulting continuous quiet time is expressed as the number of frames or the converted duration.

[0036] In step S207, temporal index alignment refers to using the temporal sequence number of the acoustic frame as a common index to map the sound pressure statistics, human activity indicators, and continuous quiet time frame by frame, ensuring that the features of the same frame are at the same index position in the combined result. The combined non-semantic acoustic features are based on frames as basic units, with each unit containing the feature values ​​corresponding to that frame. The original waveform and acoustic frame sequence are discarded after the feature is generated and are no longer stored in the storage medium of the acoustic monitoring device.

[0037] This embodiment converts the ambient sound signal to the frequency domain by windowing and short-time Fourier transform. The logarithmic mean of sound pressure and the amplitude of sound fluctuation are extracted along the time and frequency dimensions, respectively. Continuous human voice activity is determined by autocorrelation analysis and spectral flatness. Based on this, quiet frames and continuous quiet times are determined. Finally, the data is combined into non-semantic acoustic features according to a unified temporal index and the original data is discarded. This allows subsequent idle time determination to be performed based solely on acoustic physical quantities without involving speech content.

[0038] In some embodiments, an unoccupied environmental noise baseline for the conference room is established based on non-semantic acoustic features, and the unoccupied environmental noise baseline is updated according to time period and equipment operating status, including: The meeting room is confirmed to be unoccupied based on the duration of quiet in the non-semantic acoustic features. During the collection period corresponding to the unoccupied status, the non-semantic acoustic features are determined as the baseline sample features, and abnormal fluctuation samples that deviate from the overall baseline sample features are removed. Based on the usage time of the meeting room, the characteristics of the removed baseline samples are divided into time periods to obtain a time period sample set; The time-period average value and fluctuation range of the sound pressure statistics of non-semantic acoustic features in the time-period sample set are obtained to obtain the time-period baseline component. By associating the baseline components of each time period with the corresponding time periods, the unmanned environmental noise baseline for the conference room is obtained. Obtain the equipment operating status and current time period in the conference room; When the device operating status changes, or when the current time period is inconsistent with the time period associated with the time period baseline component, the non-semantic acoustic features under the current time period are determined as updated samples; The time period baseline components are updated using a weighted fusion method based on the updated samples, and the update of the unmanned environmental noise baseline is triggered by the cumulative deviation between the updated samples and the time period baseline components, so that different meeting rooms have their own corresponding unmanned environmental noise baselines.

[0039] In this embodiment, the status of the meeting room being unoccupied is determined based on the duration of quiet. Specifically, this can be based on whether the duration of quiet reaches a preset confirmation duration threshold. Only when this threshold is reached are the non-semantic acoustic features within the corresponding collection period determined as baseline sample features. The confirmation duration threshold can be determined based on the statistical distribution of the duration of quiet in the meeting room during historical confirmed unoccupied periods. For example, the lower limit of the distribution or a set quantile value can be used as the threshold. Alternatively, an initial value can be set by the management terminal when the baseline is first established and adjusted as the unoccupied environmental noise baseline is updated. The removal of abnormal fluctuation samples can be achieved through statistical outlier determination. That is, the mean and standard deviation of the overall baseline sample features are calculated, and samples that deviate from the mean by more than a set multiple of the standard deviation are removed. Alternatively, a quantile truncation method can be used to remove samples that are outside the set proportions at both ends of the distribution. The deviation multiple and truncation proportion are determined based on the degree of concentration of the baseline sample features; the more concentrated the samples, the smaller the deviation multiple.

[0040] The time slots are divided based on the meeting room usage periods. These periods can be determined by dividing a day into several time intervals based on historical occupancy patterns in the reservation system or building operating hours. The granularity of the division is determined by the actual usage patterns of the meeting rooms, and the interval boundaries can be obtained by clustering the historical start and end times of meeting room occupancy in the reservation system. The time slot sample set is the sample group obtained by grouping the baseline sample features after removal according to their respective time intervals.

[0041] The time-period average is obtained by calculating the mean of each component of the sound pressure statistics contained in the non-semantic acoustic features of the time-period sample set along the sample. The fluctuation range can be characterized by the range, standard deviation, or quantile interval of the sound pressure statistics within the time-period sample set. The time-period baseline component is the benchmark representation of the sound pressure statistics within that time period, and is composed of the time-period average and the fluctuation range.

[0042] The mapping relationship between the time period baseline components and the divided time periods is established by using the time period identifier as an index. By querying any current time period, the corresponding time period baseline component can be obtained. The time period baseline components are organized in an orderly manner according to the time period identifier to form the unmanned environmental noise baseline corresponding to the conference room.

[0043] The operating status of the equipment can be obtained by reading the start / stop status of equipment such as air conditioners, fresh air systems, and projectors through the conference room IoT gateway or equipment management interface. The current time period is determined by the timestamp of the acoustic monitoring device or the platform side.

[0044] When the operating status of the equipment changes, the acoustic conditions corresponding to the original baseline change accordingly. In this case, the non-semantic acoustic features collected in the current time period are determined as updated samples. When the current time period is inconsistent with the time period associated with the time period baseline component, it indicates that there is no corresponding benchmark available for that time period. Similarly, the non-semantic acoustic features in the current time period are determined as updated samples.

[0045] The weighted fusion method can be implemented by weighting the time-period baseline component and the updated sample with weighted coefficients. The weighting coefficients are determined based on the interval between the collection time of the updated sample and the current time; the longer the interval, the lower the weight. The cumulative deviation is obtained by accumulating the deviations between the updated sample and the time-period baseline component from multiple consecutive updates. When the accumulated value reaches a set amount, the overall update of the unmanned environmental noise baseline is triggered, resulting in the updated unmanned environmental noise baseline. Each conference room thus has an unmanned environmental noise baseline adapted to its acoustic environment.

[0046] This embodiment confirms the unoccupied state by the duration of quiet and collects baseline sample characteristics accordingly. After removing abnormal samples and dividing the time period, the sound pressure statistics benchmark for each time period is obtained to form an individualized unoccupied environmental noise baseline for the conference room. The baseline is then updated by weighted fusion combined with the cumulative deviation triggering condition, so that the baseline continuously matches the actual acoustic environment of the conference room as the equipment operating status and time period change, providing a reliable reference for subsequent vacancy determination.

[0047] In some embodiments, based on a comparison of non-semantic acoustic features with an unoccupied ambient noise baseline, a suspected idle event for the conference room is generated, including: Extract the baseline sound pressure statistics and baseline fluctuation range corresponding to the current time period from the unmanned environmental noise baseline; The sound pressure offset is obtained by subtracting the sound pressure statistics in the non-semantic acoustic features from the reference sound pressure statistics. The discreteness of the sound pressure statistics in the non-semantic acoustic features is calculated in the time domain to obtain the temporal fluctuation quantity. The temporal fluctuation quantity is then compared with the baseline fluctuation range to obtain the fluctuation comparison result. Based on the comparison results of the sound pressure offset with the set sound pressure offset threshold, the comparison results of the quiet duration in non-semantic acoustic features with the set quiet duration threshold, and the fluctuation comparison results, an idle judgment condition set is generated according to the logical AND relationship. If all conditions in the idle determination condition set remain true within a set time period, a suspected idle indicator for the meeting room is generated. Based on the suspected idle identifier, a suspected idle event for the meeting room is generated. The suspected idle event includes the meeting room identifier, detection time, sound pressure offset, quiet duration, and equipment operating status of the meeting room. Among them, the sound pressure offset threshold and the quiet duration threshold are dynamically generated based on the historical data of the conference room and are adjusted as the noise baseline of the unmanned environment is updated.

[0048] In this embodiment, the reference sound pressure statistics and reference fluctuation range are obtained by querying the unmanned environmental noise baseline using the current time period as the index. That is, the corresponding time period baseline component is located in the time period mapping relationship established by the baseline based on the current time period, and the reference sound pressure statistics and reference fluctuation range are extracted from it. The sound pressure offset is obtained by subtracting the sound pressure statistics in the non-semantic acoustic features from the reference sound pressure statistics according to the same components, and the resulting differences can be combined to form the sound pressure offset. The temporal fluctuation is obtained by arranging the sound pressure statistics along the time sequence within a set time window. The dispersion can be measured by standard deviation, variance, or range. The fluctuation comparison result is obtained by comparing the temporal fluctuation with the upper and lower bounds of the reference fluctuation range. Exceeding the range indicates that the fluctuation deviates from the unmanned baseline.

[0049] The idle condition set is organized logically, meaning that the comparison results of sound pressure offset and sound pressure offset threshold, quiet duration and quiet duration threshold, and fluctuation comparison results must all be true simultaneously. The sound pressure offset threshold and quiet duration threshold can be determined based on the distribution of condition values ​​between idle and used periods in the conference room's historical data; for example, a boundary value that can distinguish between the two states can be used as the threshold. The set duration can be determined based on the distribution of durations during which conditions are consistently true in the conference room's historical idle events, or it can be configured by the management system. A suspected idle flag is generated when all conditions in the idle condition set remain true within the set duration, and a suspected idle event is generated accordingly. This suspected idle event is organized using the conference room identifier and detection time as indexes to contain all the information.

[0050] The dynamic generation of the sound pressure offset threshold and quiet duration threshold is based on the distribution of condition values ​​in historical data. The boundary values ​​can be determined by using quantile values ​​or maximizing the inter-class distance. After the unmanned environmental noise baseline is updated, the thresholds are recalculated in the same way according to the updated condition value distribution, so that the thresholds are adjusted synchronously with the baseline.

[0051] This embodiment extracts a baseline value from the unoccupied environmental noise baseline according to the current time period, calculates the sound pressure offset by subtracting the sound pressure statistics from the baseline sound pressure statistics, compares the temporal fluctuation with the baseline fluctuation range, and generates an idle judgment condition set by combining the sound pressure offset threshold and the quiet duration threshold according to a logical AND relationship. When all conditions are met within the set duration, a suspected idle event is generated, and the threshold is dynamically adjusted as the baseline is updated, so that the idle judgment has the reliability of multiple condition constraints while conforming to the current acoustic environment of the conference room.

[0052] In some embodiments, the target reservation record is obtained based on the reservation records of the meeting room associated with the suspected vacancy event, including: Search for the meeting room reservation records based on the meeting room icons in suspected vacancy events; The detection time in the suspected idle event is matched with the reservation time period of each reservation record in the reservation record set to determine whether the detection time falls within the reservation time period and obtain the matching reservation record. The validity of the matched reservation records is filtered based on their reservation status. If the reservation status is canceled or ended, or if the meeting room identifier corresponding to the matched reservation record is inconsistent with the meeting room identifier in the suspected available event, the corresponding matched reservation record is excluded, and the valid reservation records are obtained. When there are multiple valid appointment records, calculate the degree of overlap between the appointment time and the detection time in each valid appointment record, and determine the valid appointment record with the largest degree of overlap as the target appointment record; When there is only one valid reservation record, the valid reservation record will be identified as the target reservation record. When there are no valid reservation records, update the actual status of the meeting room to be available for reservation based on suspected idle events; The target appointment record includes the appointment person's identifier, the appointment end time, and the remaining appointment time slot, which is determined by the appointment end time and the testing time.

[0053] In this embodiment, the set of reservation records is provided by the database maintained by the meeting room reservation system. The query uses the meeting room identifier in suspected available events as the search field, and the query range can be limited to the date range where the detection time falls. For interval matching, the start and end times of the reservation period are taken as the interval boundaries; if the detection time falls within this interval, it is considered to have fallen within the range. In the validity screening, a "cancelled" reservation status indicates that the reservation has been actively withdrawn, and a "completed" status indicates that the reservation period has expired. Inconsistencies in meeting room identifiers may stem from cross-meeting room queries or historical record attribution discrepancies.

[0054] When there are multiple valid reservation records, the degree of overlap can be characterized by the overlap duration between the reservation time period and the detection time, or by the proportion of this overlap duration to the total reservation time period. The record with the largest value is determined as the reservation record most closely corresponding to the current detection time. When there are no valid reservation records, the actual status of the meeting room represents the availability of the meeting room at the business level, which is distinct from the reservation status. The remaining reservation time period in the target reservation record is determined by the difference between the reservation end time and the detection time, representing the length of time period that can be recovered if release is executed.

[0055] This embodiment queries the set of reservation records by meeting room identifier and performs interval matching by detection time. After filtering the validity of reservation status and meeting room identifier, the target reservation record is determined by the one with the greatest overlap among multiple valid reservation records. When there is no valid reservation record, the actual status of the meeting room is directly updated, so that the acoustic level of idleness determination and the business level of reservation arrangement can be accurately matched, providing a basis for subsequent confirmation and release processing for specific reservation objects.

[0056] In some embodiments, a confirmation notification is sent to the corresponding reservation recipient based on the target reservation record, and response information is received from the reservation recipient based on the confirmation notification, including: Extract the reservation person's identifier and reservation end time from the target reservation record, and query the communication session identifier of the reservation person in the instant messaging platform based on the reservation person's identifier. If the communication session identifier is missing, determine the communication session identifier based on the backup notification method corresponding to the reservation person's identifier. A confirmation notification is generated based on the detection time and reservation end time in the suspected idle event. The confirmation notification includes the meeting room name, reservation time, suspected idle time, operation prompts and operation options. The operation options include the option to continue using, the option to release immediately and the option to release later. The confirmation notification does not carry ambient sound signals. A confirmation notification is sent to the reservation recipient through the instant messaging channel corresponding to the communication session identifier. The delivery status of the confirmation notification is bound to the response waiting timer, and the confirmation notification is resent when the delivery status is not delivered. Receive the response information returned by the reservation recipient based on the confirmation notification within the response waiting time; The operation options selected by the reservation object and the time of receiving the response information are parsed from the response information, and the parsed operation options and receiving times are associated with the target reservation record to obtain the response parsing result.

[0057] In this embodiment, the communication session identifier is provided by the user directory maintained by the instant messaging platform and is obtained by querying using the reservation user identifier as the associated field. Alternative notification methods may include the SMS channel corresponding to the enterprise address book email address or employee ID associated with the reservation user identifier. When the direct mapping relationship is missing, the communication session identifier in the instant messaging platform is obtained through reverse mapping using the alternative method. The meeting room name in the confirmation notification is taken from the meeting room basic information table, the reservation time is taken from the reservation time period in the target reservation record, the suspected idle time is taken from the detection time in the suspected idle event, and the set of operation options is presented in the form of interactive controls, including options to continue using, release immediately, and delay release. The notification content is assembled from feature data and reservation data, and the assembly process does not reference ambient sound signals.

[0058] The delivery status can be determined by the message receipt returned by the instant messaging channel. The response waiting timer starts from when the delivery status is set to delivered. If delivery is not achieved, the confirmation notification can be sent repeatedly according to the maximum number of retries. The operation options in the response information can be parsed from the option identifier or button callback value carried in the response information. The receiving time can be determined by the message timestamp returned by the instant messaging platform or the platform's receiving time. The operation options and the receiving time are written into the processing context corresponding to the record using the record identifier of the target reservation record as a foreign key to obtain the response parsing result.

[0059] This embodiment queries the communication session identifier in the instant messaging platform using the reservation user's identifier and determines it using a backup notification method if the session is missing. It assembles the meeting room name, reservation time, suspected available time, and operation options into a confirmation notification without carrying ambient sound signals. This notification is sent through the instant messaging channel, and the delivery status is bound to the response waiting timer. After receiving and parsing the response information within the timer, it is associated with the target reservation record. This allows the reservation recipient to independently choose their processing intention based on the suspected available time information, providing a basis for manual confirmation for subsequent resource release decisions.

[0060] In some embodiments, when no response information is returned, a re-inspection instruction is generated based on a suspected idle event, including: If the response timer for the confirmation notification expires and no response is returned, and the delivery status of the confirmation notification is "delivered", a non-response determination result is generated. Based on the non-response determination result, query the online status of the acoustic monitoring device and the update status of non-semantic acoustic features; When the online status is online and the update status is normal, extract the meeting room identifier and detection time from the suspected idle events, and extract the comparison results of non-semantic acoustic features and unmanned environmental noise baseline; A re-inspection instruction is generated based on the meeting room identifier, detection time, and comparison results. The re-inspection instruction includes the re-inspection trigger time, re-inspection duration, re-inspection collection items, and re-inspection threshold. The re-inspection threshold is consistent with the threshold corresponding to the comparison results, and the re-inspection collection items correspond to the constituent items of the non-semantic acoustic features. The re-inspection command is associated with the suspected idle event to obtain the associated re-inspection command. The associated re-inspection command is used to instruct the acoustic monitoring device to extract non-semantic acoustic features again according to the re-inspection collection items. When the online status is offline or the update status is interrupted, mark the meeting room as unconfirmable, terminate the generation of re-inspection instructions, and output the unconfirmable status.

[0061] In this embodiment, whether the response waiting timer has expired is determined by the expiration flag of the timer instance maintained by the platform; whether the response information has not been returned is determined by whether the response parsing result has been written into the processing context corresponding to the target reservation record, and if it has not been written, it is considered that it has not been returned; whether the delivery status is delivered is determined by the message receipt returned by the instant messaging channel for the confirmation notification.

[0062] Optionally, the online status of the acoustic monitoring device is determined by the heartbeat messages exchanged between the acoustic monitoring device and the platform at a set period. The platform records the receiving time of the most recent heartbeat; if the difference between this time and the current time exceeds the set period, it is considered offline. The update status of non-semantic acoustic features is determined by the timestamp carried by the acoustic monitoring device when reporting non-semantic acoustic features. This timestamp is compared with the set reporting period; if no new timestamp appears within the set reporting period, it is considered interrupted.

[0063] Extract the meeting room identifier and detection time from suspected idle events, i.e., read them from the information items recorded for that event. The extraction of the comparison results is not a recalculation in this step, but rather the reading of the judgment data that was already recorded along with the event when the suspected idle event was generated.

[0064] The parameters included in the re-inspection command are determined as follows: The re-inspection trigger time is obtained by adding the current time to the set delay. The re-inspection duration is determined based on the set duration used in the initial judgment, ensuring that the re-inspection covers an observation length equivalent to that of the initial judgment. The re-inspection acquisition items correspond to the constituent items of the non-semantic acoustic features, ensuring that the features extracted in the re-inspection are consistent in type with those used in the initial judgment. The re-inspection threshold uses the threshold value corresponding to the comparison result, ensuring that both judgments are performed under the same judgment caliber.

[0065] The method for generating the associated re-inspection instruction is to use the event identifier of the suspected idle event as a foreign key, and write the re-inspection instruction and the suspected idle event into the same processing record, so that the acoustic monitoring device can trace back to the corresponding judgment context based on the event identifier when performing the re-inspection.

[0066] When the online status is offline or the update status is interrupted, the meeting room is marked as unconfirmable. This status is reported to the meeting room management platform along with the status update result, and is displayed as a prompt that it cannot be automatically released in the reservation interface. At the same time, the generation of re-inspection instructions is terminated.

[0067] This embodiment generates a non-response determination result when the response waiting time expires, the response information is not returned, and the delivery status is "delivered". Based on this, it queries the online status of the acoustic monitoring device and the update status of the non-semantic acoustic features. When both are normal, it generates a re-inspection instruction based on the conference room identifier, detection time, and comparison results, and uses the original judgment criteria. When any abnormality occurs, it marks an unconfirmable state and terminates the generation of the re-inspection instruction. This ensures that the re-inspection process is only initiated when the data is reliable, avoiding unreliable release judgments when the device is offline or the features are interrupted.

[0068] In some embodiments, the acoustic monitoring device is controlled to perform re-inspection data acquisition according to the re-inspection command to obtain the re-inspection results, including: Based on the re-inspection instruction, acquisition control parameters are sent to the acoustic monitoring device. The acquisition control parameters include acquisition time, acquisition window, acquisition feature items and acquisition number. The acoustic monitoring device collects the sound signal of the conference room within the acquisition window at the acquisition time, and performs feature statistics on the sound signal according to the acquired feature items to extract the acoustic features for re-examination. Based on the comparison results of suspected idle events associated with the re-inspection instruction, the acoustic characteristics of the re-inspection are compared with the noise baseline of the unmanned environment using the same aperture to obtain the re-inspection comparison results. When the re-inspection comparison results indicate that the acoustic characteristics of the re-inspection are still within the range defined by the unmanned environmental noise baseline, a re-inspection result representing the idle period is generated. When the sound pressure statistics in the re-examined acoustic features exceed the range defined by the unmanned environmental noise baseline, the re-examination results used to characterize the process are generated. When the acoustic monitoring device interrupts acquisition within the acquisition window or the update of the re-examination acoustic features is interrupted, a re-examination result characterized as unconfirmable is generated. The re-inspection results include the acoustic characteristics of the re-inspection, the re-inspection comparison results, and the re-inspection mark of the conference room. The re-inspection mark is determined by the conference room mark in the suspected idle event and the detection time.

[0069] In this embodiment, the acquisition time in the acquisition control parameters is determined by the re-inspection trigger time included in the re-inspection command, the duration of the acquisition window is determined by the re-inspection duration included in the re-inspection command, the acquisition feature items correspond to the re-inspection acquisition items included in the re-inspection command, and the number of acquisitions is set according to the required credibility of the re-inspection results. The above parameters are sent by the platform to the acoustic monitoring device through control messages.

[0070] The acoustic monitoring device collects sound signals within the acquisition window at the acquisition time and performs feature statistics according to the acquired feature items. The obtained re-examined acoustic features are consistent with the initially extracted non-semantic acoustic features in terms of type composition, making the re-examined acoustic features comparable to the unmanned environmental noise baseline. The re-examined comparison results are obtained by using the time period baseline component corresponding to the comparison results recorded in the suspected idle event as the benchmark, and comparing the components in the re-examined acoustic features that are the same as the benchmark one by one, with the comparison caliber being the same as the initial judgment. The judgment of being within or outside the range is based on the upper and lower boundaries of the benchmark fluctuation range in the unmanned environmental noise baseline as the comparison basis.

[0071] The re-inspection identifier is determined by the meeting room identifier in the suspected idle event and the detection time. The re-inspection result carries the re-inspection acoustic features and re-inspection comparison results with the re-inspection identifier as an index, so that the re-inspection result can be uniquely traced back to the corresponding suspected idle event.

[0072] In this embodiment, the acoustic monitoring device receives acquisition control parameters based on the re-inspection command. The acoustic monitoring device then extracts the re-inspection acoustic features according to the acquisition feature items. After comparing the re-inspection results with the unmanned environmental noise baseline using the same caliber, it generates re-inspection results indicating whether the environment is idle, in use, or unconfirmable. The re-inspection results are indexed by a re-inspection identifier, ensuring that the re-inspection process is consistent with the initial judgment in terms of both acquisition parameters and judgment caliber. This provides a comparable verification basis for subsequent release decisions.

[0073] In some embodiments, the reservation status of the meeting room is updated according to the meeting room resource release instruction, and the meeting room is marked as unconfirmable and the generation of the meeting room resource release instruction is terminated when the acoustic monitoring device is offline or the non-semantic acoustic features are not updated. The status update result is output, including: Based on the release intention indicated by the response information or the idle conclusion represented by the re-inspection results, a release determination result is generated, and a meeting room resource release instruction is generated when the release determination result indicates release; The resource release command for the meeting room is validated based on the meeting room's attributes. If the meeting room's attributes are in reminder mode only, the resource release command is stopped and only the reminder result is output. When the meeting room is in automatic release mode, the meeting room reservation interface is called according to the meeting room resource release command to cancel the remaining reservation time slots of the meeting room; Based on the result returned by the reservation interface, update the reservation status of the meeting room from occupied to available, and obtain the reservation status update result; When the acoustic monitoring device is offline, the network is interrupted, the non-semantic acoustic features are not updated, or the re-inspection results are unconfirmable, the meeting room will be marked as unconfirmable, and the generation of meeting room resource release instructions and the updating of reservation status will be terminated. Based on the reservation status update results and the unconfirmed status, generate and output status update results. The status update results include the meeting room identifier, reservation status, release duration, and number of suspected idle times.

[0074] In this embodiment, the release determination result is based on the release intention indicated by the response information and the idle conclusion represented by the re-inspection result. The release intention can be determined by the operation options parsed from the aforementioned response parsing result, and the idle conclusion can be determined by the re-inspection comparison result in the re-inspection result. A release is determined if either of the two conditions is met. The meeting room resource release instruction targets the remaining reservation time slot corresponding to the target reservation record.

[0075] Meeting room attributes can be read from the meeting room configuration table maintained by the meeting room management platform. This configuration table records the release policies of each meeting room using the meeting room identifier as an index. When the meeting room attribute is in reminder-only mode, the meeting room resource release command is intercepted before being sent and converted into a reminder message. When the meeting room attribute is in automatic release mode, the meeting room resource release command is executed by the platform calling the interface provided by the meeting room reservation system, with the meeting room identifier and remaining reservation time slots as input parameters.

[0076] The result returned by the reservation interface indicates whether the cancellation operation was successful. If successful, the meeting room's reservation status is changed from occupied to available, resulting in a reservation status update. If the acoustic monitoring device is offline, the network is interrupted, non-semantic acoustic features are not updated, or the re-inspection result is unconfirmable, the generation of the meeting room resource release command is terminated, and the reservation status update is simultaneously terminated, maintaining the meeting room's original status. The release duration in the status update result can be determined by the remaining reservation time slots, and the number of suspected idle periods can be obtained from the cumulative count of suspected idle events.

[0077] This embodiment generates a release judgment result based on the release intention indicated by the response information or the idle conclusion represented by the re-inspection result, and generates a meeting room resource release instruction accordingly. After the meeting room attribute is verified, the reservation interface is called in the automatic release mode to cancel the remaining reservation time and update the reservation status to the reservation status. In the reminder-only mode, the instruction is stopped and the reminder result is output. In abnormal situations such as device offline or feature not updated, the status is marked as unconfirmable and the instruction generation and status update are terminated. This ensures that the release action is executed under the premise of having a release basis and conforming to the meeting room configuration strategy, taking into account both resource recovery efficiency and management controllability.

[0078] In a second aspect, this embodiment also provides a meeting room vacancy identification and reservation resource release system for executing the method described in the first aspect. The system includes an acoustic monitoring device, a baseline management unit, a vacancy identification unit, a reservation association unit, a notification interaction unit, a re-inspection control unit, and a release execution unit. The acoustic monitoring device is located in the meeting room and includes a sound acquisition module, a sound pressure detection module, a processing module, a communication module, and a power supply module. The sound acquisition module is used to acquire ambient sound signals in the meeting room. The sound pressure detection module and the processing module are used to extract non-semantic acoustic features based on the ambient sound signals. The communication module is used to upload the non-semantic acoustic features. The baseline management unit is used to establish an unoccupied ambient noise baseline corresponding to the meeting room based on the non-semantic acoustic features, and to update the unoccupied ambient noise baseline according to the time period and equipment operating status. The vacancy identification unit is used to identify non-semantic acoustic features based on the non-semantic acoustic features. The comparison results with the ambient noise baseline generate suspected vacancy events for the meeting room; the reservation association unit associates the meeting room reservation records with the suspected vacancy events to obtain the target reservation record; the notification interaction unit sends a confirmation notification to the corresponding reservation object based on the target reservation record and receives the response information returned by the reservation object based on the confirmation notification; the re-inspection control unit generates a re-inspection command based on the suspected vacancy events when the response information is not returned, and controls the acoustic monitoring device to perform re-inspection acquisition based on the re-inspection command to obtain the re-inspection result; the release execution unit generates a meeting room resource release command based on the response information or the re-inspection result, updates the meeting room reservation status based on the meeting room resource release command, and marks the meeting room as unconfirmable and terminates the generation of the meeting room resource release command when the acoustic monitoring device is offline or the non-semantic acoustic features are not updated, and outputs the status update result.

[0079] This system places the acoustic monitoring device at the conference room side to handle the local extraction and uploading of non-semantic acoustic features. The baseline management unit, idleness identification unit, appointment association unit, notification interaction unit, re-inspection control unit, and release execution unit are deployed on the platform side to handle baseline maintenance, event generation, appointment association, notification confirmation, re-inspection control, and release execution. These units are sequentially connected via feature data and event records. The acoustic monitoring device only uploads non-semantic acoustic features, not the original sound signal, allowing the platform to obtain the necessary data for idleness determination without accessing the meeting content. The baseline management unit and idleness identification unit work together to adapt the judgment criteria to the conference room environment and time period. The output of the idleness identification unit drives subsequent units in the form of events. The appointment association unit maps the acoustic idleness status to the business-level appointment records. The notification interaction unit and re-inspection control unit provide dual protection of manual confirmation and automatic verification before release. Based on this, the release execution unit decides whether to execute the release according to the conference room attributes and blocks the release in an unconfirmed state if there are data anomalies. Through the division of labor and coordination among its various units, the system automates the identification of meeting room vacancy status and the release of reserved resources, while ensuring accuracy in identification, reliability in release, and controllability in management.

[0080] By adopting the above technical solution, this invention differs from existing technologies and has the following beneficial effects: It acquires ambient sound signals from the conference room using an acoustic monitoring device and extracts non-semantic acoustic features containing sound pressure statistics and quiet duration. These features do not contain semantic content and are discarded after local extraction. A baseline of unoccupied ambient noise corresponding to the conference room is established based on the non-semantic acoustic features, and this baseline is updated according to the time period and equipment operating status, so that different conference rooms have judgment criteria adapted to their acoustic environment. A suspected idle event is generated based on the comparison result between the non-semantic acoustic features and the unoccupied ambient noise baseline, and the target reservation record is obtained by associating it with the reservation record. A confirmation notification is sent to the reservation recipient, and a response information is received. Alternatively, if the response information is not returned, a re-inspection command is generated based on the suspected idle event, and the acoustic monitoring device is controlled to perform a re-inspection to obtain the re-inspection result. A conference room resource release command is generated based on the response information or the re-inspection result, and the reservation status is updated. If the acoustic monitoring device is offline or the non-semantic acoustic features are not updated, the conference room is marked as unconfirmable, and the generation of the release command is terminated. The status update result is output. The above technical solution achieves accurate identification of meeting room vacancy status and reliable release of reserved resources without deploying cameras and human body sensing devices, or involving meeting content and identity information, thereby improving the utilization rate and management efficiency of meeting room resources.

[0081] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0082] 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 this 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.) or processor to execute all or part of the steps of the methods of various embodiments of this 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.

[0083] The above description is only a part of the embodiments of the present invention and does not limit the scope of protection of the present invention. Any equivalent device or equivalent process transformation made based on the content of the present invention specification and drawings, or direct or indirect application in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for identifying meeting room vacancy and releasing reserved resources, characterized in that, include: Acquire ambient sound signals from the conference room, which is equipped with an acoustic monitoring device; Non-semantic acoustic features are extracted based on the environmental sound signals. These non-semantic acoustic features include sound pressure statistics and quiet duration, but do not contain semantic content. Based on the non-semantic acoustic features, an unmanned environmental noise baseline corresponding to the conference room is established, and the unmanned environmental noise baseline is updated according to the time period and equipment operating status. Based on the comparison results between the non-semantic acoustic features and the unmanned environmental noise baseline, a suspected idle event of the conference room is generated; Based on the suspected vacancy events and the meeting room reservation records, the target reservation record is obtained; A confirmation notification is sent to the corresponding reservation object according to the target reservation record, and a response information is received from the reservation object based on the confirmation notification; or, if the response information is not returned, a re-inspection instruction is generated based on the suspected idle event. The acoustic monitoring device is controlled to perform re-inspection data acquisition according to the re-inspection command, and the re-inspection results are obtained; A meeting room resource release command is generated based on the response information or the re-inspection result; The reservation status of the meeting room is updated according to the meeting room resource release instruction. If the acoustic monitoring device is offline or the non-semantic acoustic features are not updated, the meeting room is marked as unconfirmable and the generation of the meeting room resource release instruction is terminated. The status update result is then output.

2. The method for identifying and releasing reserved meeting room vacancy resources according to claim 1, characterized in that, Extracting non-semantic acoustic features based on the environmental sound signals includes: The ambient sound signal is windowed and framed according to the set frame length and frame shift, and temporally adjacent acoustic frames are overlapped and connected to obtain an acoustic frame sequence. Perform a short-time Fourier transform on each acoustic frame in the acoustic frame sequence to transform the acoustic frame from the time domain to the frequency domain, and obtain the short-time energy spectrum and spectral distribution characteristics corresponding to the acoustic frame; The logarithmic mean of the sound pressure level is obtained by taking the logarithm of the short-time energy spectrum of each acoustic frame, summing them in time, and calculating the mean. The dispersion statistic of the spectral distribution characteristics in the time dimension is obtained to obtain the sound fluctuation amplitude, and the logarithmic mean of the sound pressure and the sound fluctuation amplitude together constitute the sound pressure statistic. Autocorrelation analysis is performed on the acoustic frame sequence to obtain the fundamental frequency trajectory, and the spectral flatness is calculated based on the spectral distribution characteristics. Based on the continuity of the fundamental frequency trajectory and the comparison result of the spectral flatness with the set activity decision threshold, continuous human voice activity is determined to obtain a human voice activity identifier. Based on the human voice activity identifier, the acoustic frame in which there is no continuous human voice activity is identified as a quiet frame, and the consecutive quiet frames in time are accumulated and counted to obtain the continuous quiet time. The sound pressure statistics, the human voice activity identifier, and the continuous quiet time are aligned and combined according to the same temporal index to obtain the non-semantic acoustic features, and the original waveform of the environmental sound signal and the acoustic frame sequence are discarded.

3. The method for identifying and releasing meeting room vacancy resources according to claim 1, characterized in that, Based on the aforementioned non-semantic acoustic features, an unoccupied environmental noise baseline corresponding to the conference room is established, and the unoccupied environmental noise baseline is updated according to the time period and equipment operating status, including: The meeting room is confirmed to be unoccupied based on the quiet duration in the non-semantic acoustic features. During the collection period corresponding to the unoccupied status, the non-semantic acoustic features are determined as baseline sample features, and abnormal fluctuation samples that deviate from the overall baseline sample features are removed. Based on the usage time of the meeting room, the characteristics of the removed baseline samples are divided into time periods to obtain a time period sample set; The time-period average value and fluctuation range of the sound pressure statistics are obtained from the non-semantic acoustic features in the time-period sample set to obtain the time-period baseline component; By associating each of the time period baseline components with the corresponding time periods, the unmanned environmental noise baseline corresponding to the conference room is obtained; Obtain the equipment operating status and current time period of the conference room; When the device operating status changes, or when the current time period is inconsistent with the time period associated with the time period baseline component, the non-semantic acoustic features under the current time period are determined as updated samples; The time period baseline component is updated by weighted fusion based on the updated sample, and the update of the unmanned environmental noise baseline is triggered by the cumulative deviation between the updated sample and the time period baseline component, so that different meeting rooms have their own corresponding unmanned environmental noise baseline.

4. The method for identifying and releasing reserved meeting room vacancy resources according to claim 1, characterized in that, Based on the comparison results between the non-semantic acoustic features and the unoccupied environmental noise baseline, a suspected idle event for the conference room is generated, including: Extract the baseline sound pressure statistics and baseline fluctuation range corresponding to the current time period from the unmanned environmental noise baseline; The sound pressure offset is obtained by subtracting the sound pressure statistic from the reference sound pressure statistic in the non-semantic acoustic features. The discreteness of the sound pressure statistics in the non-semantic acoustic features is calculated in the time domain to obtain the temporal fluctuation quantity, and the temporal fluctuation quantity is compared with the benchmark fluctuation range to obtain the fluctuation comparison result. Based on the comparison results of the sound pressure offset with the set sound pressure offset threshold, the comparison results of the quiet duration in the non-semantic acoustic features with the set quiet duration threshold, and the fluctuation comparison results, an idle judgment condition set is generated according to the logical AND relationship. When all conditions in the set of idle determination conditions remain true within a set time period, a suspected idle indicator for the meeting room is generated. Based on the suspected idle identifier, a suspected idle event for the conference room is generated. The suspected idle event includes the conference room identifier, detection time, sound pressure offset, quiet duration, and equipment operating status of the conference room. The sound pressure offset threshold and the quiet duration threshold are dynamically generated based on the historical data of the conference room and are adjusted as the unmanned environmental noise baseline is updated.

5. The method for identifying and releasing reserved meeting room vacancy resources according to claim 1, characterized in that, Based on the suspected vacancy events and the associated meeting room reservation records, the target reservation records are obtained, including: Based on the meeting room identifier in the suspected idle event, query the set of reservation records for the meeting room; The detection time in the suspected idle event is matched with the reservation time period of each reservation record in the reservation record set to determine whether the detection time falls within the reservation time period, and thus obtain the matched reservation record. The validity of the hit reservation records is filtered based on their reservation status. If the reservation status is canceled or ended, or if the meeting room identifier corresponding to the hit reservation record is inconsistent with the meeting room identifier in the suspected idle event, the corresponding hit reservation record is excluded to obtain valid reservation records. When there are multiple valid reservation records, the degree of overlap between the reservation time period and the detection time in each valid reservation record is calculated, and the valid reservation record with the largest degree of overlap is determined as the target reservation record; When there is only one valid reservation record, that valid reservation record is identified as the target reservation record. When the valid reservation record is empty, the actual status of the meeting room is updated to a reservationable status based on the suspected idle event; The target reservation record includes the reservation holder's identifier, the reservation end time, and the remaining reservation time slot, which is determined by the reservation end time and the detection time.

6. The method for identifying and releasing reserved meeting room vacancy resources according to claim 1, characterized in that, Based on the target reservation record, a confirmation notification is sent to the corresponding reservation recipient, and response information is received from the reservation recipient based on the confirmation notification, including: Extract the reservation person's identifier and reservation end time from the target reservation record, and query the communication session identifier of the reservation person in the instant messaging platform based on the reservation person's identifier. If the communication session identifier is missing, determine the communication session identifier based on the backup notification method corresponding to the reservation person's identifier. A confirmation notification is generated based on the detection time in the suspected idle event and the reservation end time. The confirmation notification includes the meeting room name, reservation time, suspected idle time, operation prompt information and operation option set. The operation option set includes the option to continue using, the option to release immediately and the option to release later. The confirmation notification does not carry the ambient sound signal. The confirmation notification is sent to the reservation recipient through the instant messaging channel corresponding to the communication session identifier. The delivery status of the confirmation notification is bound to the response waiting timer, and the confirmation notification is resent when the delivery status is not delivered. Within the response waiting time, receive the response information returned by the reservation recipient based on the confirmation notification; The operation option selected by the reservation object and the receiving time of the response information are parsed from the response information, and the parsed operation option and the receiving time are associated with the target reservation record to obtain the response parsing result.

7. The method for identifying and releasing reserved meeting room vacancy resources according to claim 1, characterized in that, If the response information is not returned, a re-inspection instruction is generated based on the suspected idle event, including: When the response timer for the confirmation notification expires and no response is returned, and the delivery status of the confirmation notification is "delivered", a non-response determination result is generated. Based on the non-response determination result, query the online status of the acoustic monitoring device and the update status of the non-semantic acoustic features; When the online status is online and the update status is normal update, extract the meeting room identifier and detection time from the suspected idle event, and extract the comparison results between the non-semantic acoustic features and the unmanned environmental noise baseline; A re-inspection instruction is generated based on the meeting room identifier, the detection time, and the comparison result. The re-inspection instruction includes a re-inspection trigger time, a re-inspection duration, re-inspection collection items, and a re-inspection threshold. The re-inspection threshold is consistent with the threshold corresponding to the comparison result, and the re-inspection collection items correspond to the constituent items of the non-semantic acoustic features. The re-inspection instruction is associated with the suspected idle event to obtain the associated re-inspection instruction. The associated re-inspection instruction is used to instruct the acoustic monitoring device to extract the non-semantic acoustic features again according to the re-inspection collection item. When the online status is offline or the update status is interrupted, the meeting room is marked as unconfirmable, the generation of the re-inspection instruction is terminated, and the unconfirmable status is output.

8. The method for identifying and releasing reserved meeting room vacancy resources according to claim 1, characterized in that, According to the re-inspection command, the acoustic monitoring device is controlled to perform re-inspection data acquisition to obtain the re-inspection results, including: According to the re-inspection instruction, the acoustic monitoring device is issued acquisition control parameters, which include acquisition time, acquisition window, acquisition feature items and acquisition number; The acoustic monitoring device acquires the sound signal of the conference room within the acquisition window at the acquisition time, and performs feature statistics on the sound signal according to the acquisition feature items to extract the re-examination acoustic features; Based on the comparison results in the suspected idle events associated with the re-inspection command, the re-inspection acoustic features are compared with the unmanned environmental noise baseline using the same aperture to obtain the re-inspection comparison results; When the re-inspection comparison result indicates that the re-inspection acoustic features are still within the range defined by the unmanned environmental noise baseline, a re-inspection result indicating idleness is generated; When the sound pressure statistics in the re-examined acoustic features exceed the range defined by the unmanned environmental noise baseline, a re-examination result representing the use is generated; When the acoustic monitoring device interrupts acquisition within the acquisition window or the update of the re-examination acoustic features is interrupted, a re-examination result characterized as unconfirmable is generated. The re-inspection results include the re-inspection acoustic features, the re-inspection comparison results, and the re-inspection identifier of the conference room. The re-inspection identifier is determined by the conference room identifier and the detection time in the suspected idle event.

9. The method for identifying and releasing reserved meeting room vacancy resources according to claim 1, characterized in that, The reservation status of the meeting room is updated according to the meeting room resource release instruction. If the acoustic monitoring device is offline or the non-semantic acoustic features are not updated, the meeting room is marked as unconfirmable and the generation of the meeting room resource release instruction is terminated. The status update result is output, including: A release determination result is generated based on the release intention indicated by the response information or the idle conclusion represented by the re-inspection result, and a conference room resource release instruction is generated when the release determination result indicates release; The resource release command for the meeting room is verified based on the meeting room's attributes. If the meeting room's attributes are in reminder mode only, the resource release command is aborted, and only the reminder result is output. When the meeting room attribute is in automatic release mode, the meeting room reservation interface is called according to the meeting room resource release instruction to cancel the remaining reservation time slots of the meeting room; Based on the result returned by the reservation interface, the reservation status of the meeting room is updated from occupied to available, thus obtaining the reservation status update result; When the acoustic monitoring device is offline, the network is interrupted, the non-semantic acoustic features are not updated, or the re-inspection result is unconfirmable, the meeting room will be marked as unconfirmable, and the generation of the meeting room resource release instruction and the update of the reservation status will be terminated. The status update result is generated and output based on the reservation status update result and the unconfirmed status. The status update result includes the meeting room identifier, the reservation status, the release duration, and the number of suspected idle times.

10. A system for identifying and releasing reserved meeting room vacancy resources, characterized in that, The system for performing the method according to any one of claims 1 to 9, the system comprising: An acoustic monitoring device, installed in the conference room, includes a sound acquisition module, a sound pressure detection module, a processing module, a communication module, and a power supply module. The sound acquisition module is used to acquire the ambient sound signal in the conference room. The sound pressure detection module and the processing module are used to extract non-semantic acoustic features based on the ambient sound signal. The communication module is used to upload the non-semantic acoustic features. The baseline management unit is used to establish the unmanned environmental noise baseline corresponding to the conference room based on the non-semantic acoustic features, and to update the unmanned environmental noise baseline according to the time period and equipment operating status. An idle identification unit is used to generate a suspected idle event of the conference room based on the comparison result between the non-semantic acoustic features and the noisy background noise baseline. The reservation association unit is used to associate the meeting room reservation records with the suspected idle events to obtain the target reservation record; The notification interaction unit is used to send a confirmation notification to the corresponding reservation object according to the target reservation record, and to receive the response information returned by the reservation object based on the confirmation notification; The re-inspection control unit is used to generate a re-inspection command based on the suspected idle event when the response information is not returned, and to control the acoustic monitoring device to perform re-inspection acquisition based on the re-inspection command to obtain the re-inspection result; The release execution unit is used to generate a meeting room resource release instruction based on the response information or the re-inspection result, update the reservation status of the meeting room according to the meeting room resource release instruction, mark the meeting room as unconfirmable and terminate the generation of the meeting room resource release instruction when the acoustic monitoring device is offline or the non-semantic acoustic features are not updated, and output the status update result.

Images