A closed-loop music composition generating system based on physical environment acoustics parameters.

Abstract

The invention relates to a closed-loop system that measures the physical acoustic properties of a listening environment in real time and directly converts the resulting data into music composition parameters. This system utilizes sound and music processing technologies, room and ambient acoustics measurement, digital signal processing, generative artificial intelligence, symbolic music generation, and real-time sound synthesis. The system processes ambient sound received from a microphone or microphone array and a source sound reference sent to a loudspeaker to calculate acoustic parameters such as reverberation time, musical clarity, signal-to-noise ratio, frequency transmission characteristics, ambient noise, and low-frequency resonance.The calculated acoustic parameters are converted into a normalized acoustic characteristic vector along with confidence coefficients, and this is then translated by the acoustic-composition decision engine into structural variables of the music such as tempo, note density, polyphony, internote interval, harmonic shift rate, octave spread, instrument selection, articulation, and dynamic range. The generated composition is validated against spectral masking, resonance, temporal overlap, and musical integrity criteria; only updates that provide measurable technical improvement are applied, and composition parameters are updated during playback via closed-loop feedback.

Description

- 1 - TARIFF CLOSED BASED ON PHYSICAL ENVIRONMENT ACOUSTICS PARAMETERS A system that produces CYCLE MUSIC COMPOSITION. TECHNICAL FIELD The invention relates to sound and music processing technologies, room and ambient acoustics measurement, digital signal processing, generative artificial intelligence, symbolic music production, and reality. It relates to areas of timed sound synthesis. Specifically, the invention involves a listening 10 audio data received from the environment via a microphone or array of microphones processing, calculation of the physical acoustic properties of the environment and the calculated by converting acoustic properties into music composition parameters It relates to a system that produces music adapted to the environment. The system includes reverberation time, musical clarity, band-based noise level, 15 signal-to-noise ratio, frequency transmission characteristics, and low-frequency resonance By using acoustic parameters such as tempo, note density, and polyphony in music, Inter-note onset time, harmonic change rate, octave distribution, instrument The choice determines the characteristics of timbre, articulation, and dynamic range. The invention Application areas include digital music platforms, mobile music and audio 20 applications, in-car infotainment and audio systems, smart speakers and home audio systems, television and media playback systems, games and virtual environments applications, personalized sound production services for concerts, events and indoor venues The venue features sound systems. The invention allows for the addition of only equalizers, filters, gain boosters, or volume enhancements to an existing music signal. Unlike systems that apply level correction, the physical acoustics of the environment It reconstructs the compositional structure of the music according to its conditions. The system must include at least one processor, memory, data storage units, and a microphone or microphone headset. array, speaker or audio output unit and edge (EDGE) calculation capability 30 to be run on one or more user devices It has been improved. - 2 - PREVIOUS TECHNIQUE In existing systems, in order to reduce the acoustic effects of the listening environment... Generally, automatic equalizer and frequency response correction, room impulse response Measurement, speaker level and delay adjustment, phase and time alignment, bass 5 management, noise level-dependent volume adjustment, dynamic range compression, Active room correction and digital signal processing methods are used. For example, Sonos Trueplay lets sound through walls and furniture via the microphone. and measures how it reflects off other surfaces and determines the speaker output's location. It adjusts according to the environment. However, the process applied is essentially 10 It is aimed at fine-tuning the sound output of the speaker; the notes, rhythm, of the musical piece, It does not reproduce the harmonic or instrumental structure. Dirac Live analyzes the impulse response of the room and speaker system. It applies correction filters for distortions in the frequency and time domains. Solutions like Dirac Live Bass Control and Active Room Treatment utilize low frequencies (15). It can also regulate behavior and interaction between speakers. However, this systems enable more accurate reproduction of existing content within the room. It aims to rethink the compositional structure of music according to the acoustic environment. It does not create. The Bose ADAPTiQ uses a measurement microphone to measure the sound characteristics of the environment. It measures and adjusts the output settings of the sound system accordingly. This The approach is also fundamentally related to the calibration of the sound system. Audyssey, Samsung SpaceFit Sound, LG AI Room Calibration, and similar solutions also help with room calibration. or a rival group of technologies that regulate audio output according to listening position These solutions can be evaluated. The common aspect of these solutions is that they allow playback while preserving the existing music. 25 They need to fix the chaining and the speaker output. In known room adjustment systems, measured ambient data is often filtered. coefficients, channel levels, loudspeaker delays, phase behavior, and It is used in adjusting frequency gains. Therefore, high In an environment with reverberation, the mixing of note events, intense 30 the polyphonic structure becoming unintelligible or certain bass notes being in the room - 3 - Problems such as overlapping resonances are at the composition level. It cannot be resolved. For example, if there is a strong room resonance around 63 Hz. Known systems can reduce the sound level at this frequency. However, with resonance... It does not shift the conflicting note to another octave, it does not change the note duration, bass 5 does not change its instrument, simultaneous instrument in the same frequency range It does not reduce the number, nor does it recreate the chord inversion or bass line. Similar Applying Auto-EQ in an environment where the RT60 value is high, as described above, affects the musical notation. because it does not change the intensity or the start time between notes Temporal overlap may not be completely eliminated. 10 Document number US20190080709A1, power spectral density. by utilizing their predictions, reverberation parameters such as T60 and direct- It relates to the calculation of the reverberated sound ratio. Document environment acoustics. While relevant to estimation, the calculated acoustic parameters of music It does not convert into composition variables and notes, rhythms, octaves or 15 It does not produce instrumentation. The difference of the invention is that it only measures acoustic parameters. not only measures it but also converts it into a confidence-weighted feature vector, and this Vector input includes note density, polyphony, interval between notes, octave, and instrument selection. It is the production. Document number US10467998B2; tempo, rhythm, pitch, chord, octave and 20 an automated music system where instrument selections are determined by probability tables It relates to the composition system. However, it also relates to music production parameters. RT60, C80, band-based SNR, or bass measured from a physical listening environment. The difference of the invention lies in its composition. In addition to user or musical experience identifiers, their decisions are directly related to 25 It is derived from physical acoustic measurements. In document US11839815B2, artificial neural networks are defined as previously used in humans. dynamically selects compatible music stems created by and It is confusing. The solution, however, involves adaptive audio mixing. New note events 30 depending on the RT60, resonance or musical clarity values ​​of the environment. and does not produce a compositional structure. The difference of the invention is that it is not previously recorded. - 4 - Instead of simply selecting or mixing stems, the choice depends on the ambient acoustics. It is the recreation of symbolic compositions or musical events. Document US9467791B2, object-based audio and spatial Using metadata, the content can be displayed in different speaker configurations and playback environments. It relates to the appropriate presentation of sound objects. The main purpose is the position and 5 It is the adaptation of mix information to the playback environment. The difference of the invention is the sound. Instead of spatially repositioning objects, acoustic measurements rhythm, note density, harmony, octave and instrumentation are redefined through movement. It is the determination of. Document number US5434922A, music and noise inside the vehicle 10 It compensates for ambient noise by measuring sound levels. The invention... The difference is that instead of adjusting the sound level or audio signal according to the noise level, instrumental and melodic recording of the composition according to the frequency distribution of the noise. It is changing the region. Document number US20230104111A1 describes how the user interacts with device sensors 15 by determining the parameters of the existing acoustic environment and using content acoustics It explains how to apply spatial filters and presets by comparing them. It is based on acoustic parameter measurement but is a confidence-weighted vector. through note density, polyphony, interval between notes, octave and Transformation to instrument selection, technical cost gateway and closed loop 20 Elements such as compositional updating are missing, and it is symbolic music production. It is not specified. Document number US20150179181A1 calibrates ambient acoustics. Measure signals and update filters; based on ambient noise peak frequencies. It describes adjusting the pitch or selecting music. Frequency response 25 It is based on the correction of RT60, C80, SNR and bass resonance score. quantitative transformation of composition parameters, confidence-weighted feature vector, Spectral masking, resonance, and temporal overlap cost validation. Elements such as closed-loop symbolic production are missing, and the note level needs to be recalculated. Composition is not specified. 30 The previous technique is generally grouped into two separate categories: ambient acoustics Systems that correct sound output by measuring, user request or musical - 5 - Systems that generate music based on parameters. The invention lies between these two fields. It establishes a direct and quantitative technical link: physical acoustic measurement, reliability. weighted acoustic feature vector, composition parameters, symbolic music Production, technical cost verification, and a closed-loop update chain. This Therefore, the invention is based on the known Auto-EQ, room calibration, and general artificial intelligence-based 5 It differs from other music production systems. A BRIEF DESCRIPTION OF THE INVENTION Sound and music processing technologies, room and ambient acoustics measurement, digital 10 signal processing, generative artificial intelligence, symbolic music generation, and real-time audio. In the field of synthesis, the physical acoustic properties of a listening environment are realistic. measuring over time and directly transferring the data obtained from the measurement to music invention relating to a closed-loop system that converts composition parameters The subject is closed-loop music based on physical environment acoustics parameters 15 A system for producing the composition has been developed. The developed system uses ambient sound captured from a microphone or array of microphones. The reverberation time is determined by processing the source sound reference sent to the speaker. musical clarity, signal-to-noise ratio, frequency transmission characteristics, ambient noise and calculates acoustic parameters such as low-frequency resonance. 20 The calculated acoustic parameters were normalized along with confidence coefficients. It is converted into an acoustic feature vector and acoustic-composition decision The engine controls the tempo, note density, polyphony, and internotes of the music. start time, harmonic change rate, octave distribution, instrument selection, This translates into structural variables such as articulation and dynamic range. 25 Thus, the system only adds equalizers, filters, gain boosters, or other effects to the existing audio output. It does not apply volume correction; it focuses on the rhythm, melody, harmony, and redesigning the instrumentation structure according to physical environmental conditions. It consists of: The produced composition; spectral masking, resonance, 30 technical aspects such as temporal overlap, musical integrity, volume level, and production delay It is verified according to the criteria. The new composition is only available. when it provides a specific technical improvement to the user according to the composition - 6 - is presented. During the sound playback, the environment is measured again and a permanent... If acoustic changes are detected, the composition parameters are closed-loop. It is updated using a feedback method. The system must include at least one processor, memory, data storage units, microphone, or microphone array, speaker or audio output unit and edge computing capability 5 developed to run on one or more user devices This involves measuring the acoustic environment and weighting the acoustic properties of the measurements with confidence. converting it into a vector, the composition parameters of that vector matching, creation of symbolic music, verification of sound output and the same as a closed-loop architecture based on the principle of re-measurement in the environment 10 He is working. DESCRIPTION OF THE FIGURES Figure 1. Closed-Loop Based on Physical Environment Acoustics Parameters 15 Music Composition Generating System Architecture The corresponding part numbers shown in the figures are given below. 100. Audio Output and User Device 20 110. Microphone and Reference Data Acquisition Module 120. Front Signal Processing Module 130. Acoustic Parameter Extraction Module 140. Environmental Classification and Temporal Tracking Module 150. Acoustic-Composition Decision Engine 30 160. User and Musical Restriction Module 170. Conditional Music Production Module 35 Module 180: Arrangement and Instrumentation - 7 - 190. Sound Synthesis and Rendering Module 200. Quality Control and Verification Module 210. Flow, Cache and Transition Management Module 5 220. Real-Time Feedback Module 230. Temporary Acoustic Property Buffer 240. Model, Rule, and Instrument Data Repository DETAILED DESCRIPTION OF THE INVENTION Sound and music processing technologies, room and ambient acoustics measurement, digital signal processing, generative artificial intelligence, symbolic music generation, and real-time audio. In the field of synthesis, the physical acoustic properties of a listening environment are determined by a microphone. or measures via a microphone array; measures audio data in the time domain, frequency domain operating in the field and time-frequency domain; medium reverberation, noise, resonance, 20 frequency attenuation, spectral masking, and musical clarity are characteristics that determine these properties. and the resulting acoustic properties are related to music composition parameters. transforming and creating a new musical structure; requires at least one processor, memory, and data. storage units, microphone or microphone array, speaker or audio output unit and one or more user devices with edge calculation capability 25 The invention, developed for use in the field of physical environment acoustics, A system that produces closed-loop music compositions based on sound parameters; Output and user device (100), microphone and reference data acquisition module (110), front signal processing module (120), acoustic parameter extraction module (130), environment Classification and temporal tracking module (140), acoustic-composition decision 30 engine (150), user and musical constraint module (160), conditional music generation module (170), editing and instrumentation module (180), sound synthesis and render module (190), quality control and verification module (200), streaming, caching and Transition management module (210), real-time feedback module (220), temporary Acoustic properties buffer (230) and model, rule and instrument data repository (240) 35 It includes. - 8 - In the invention, only the speaker output is measured as a result of ambient acoustic measurements. The frequency response of the signal is not altered. The system adapts to the measured acoustic conditions. According to this, the tempo of the music, its rhythmic subdivision, and the number of notes produced per unit of time. number of notes, number of simultaneous notes, note durations, silences between notes durations, melodic pitch and octave distribution, harmonic change frequency, chord 5 their conversions, instrument selection, instrument recording areas, timbre, and articulation characteristics, transient intensity, dynamic range, musical section and can alter the transition structure. The fundamental transformation performed by the system follows this general relationship: It can be expressed as: 10 C(t) = f(A(t)) Here, A(t) is the ambient acoustics characteristics measured at a specific time t; C(t) the target composition of the music will be created in accordance with the same environment. It expresses its parameters. 15 An example of an acoustic feature vector. A = [RT60_norm, C80_norm, SNR_k, BRS, conf, Δenv] It can be created in this form. The composition parameter vector is 20. C = [ND, Poly, IOI, HarmRate, DynRange, Oct, Inst] It can be defined in this way. The distinctive technical feature of the invention is A acoustic. feature vector directly or indirectly C composition parameter 25 Converting it to a vector and recreating the resulting composition in the same environment It is updated in a closed-loop system by measurement. The invention relates to closed-loop acoustics based on physical environment acoustic parameters. Sound output and user device (100) included in the system that produces music composition; mobile device, computer, in-car infotainment system, smart speaker, television, 30 in the form of a headphone system, a home audio system, or a standalone audio production device This can be accomplished. The module's main tasks are; initiated by the user. - 9 - Creating an acoustic adaptive music production session, considering music genre, duration, and energy. Gathering user preferences such as level, vocal preference, and allowed instruments, broadcasting the test signal to be used in active calibration through the loudspeaker, Digital samples of the stolen source audio are used as a reference channel for data collection. Transfer to module, play approved music segments, microphone access, 5 Managing user permissions regarding data processing scope and privacy preferences. device processing capacity, available memory, network latency, and number of audio output channels. This involves reporting technical skills such as these. The following example structure can be created at the start of the session: Session = {Sid, fs, bit, Nmic, Nout, Tmax, Upref, Cap} Here, Sid displays the session ID, fs the sampling frequency, and bit the bit depth. Nmic represents the number of microphones, Nout represents the number of audio output channels, and Tmax represents the maximum allowed limit. end-to-end latency, Upref user preferences, and Cap device processing capacity are 15. This refers to the module initiating active calibration and passive monitoring. control signal to switch or select safe operating mode It produces. The commissioning of the produced composition is only subject to quality control. Approval received from module (200) and flow management received from module (210) This is done after the timing information is received. User 20 is used as input. preferences, device capabilities, approved audio segments, and calibration commands The source or test signal is received; the output is a digital signal emitted into the acoustic environment. The source audio reference, session information, and user preference package are generated. The invention relates to closed-loop acoustics based on physical environment acoustic parameters. Microphone and reference data collection system included in the music composition production system 25 module (110), time domain sound representing the physical acoustic behavior of the medium simultaneously samples and the source audio reference sent to the audio output device. It collects them as follows: single microphone, stereo microphone, or multi-microphone array. It is available for use. Microphone measurement can be modeled as follows: 30 m(t) = h(t) ∗ s(t) + n(t) - 10 - Here, s(t) is the source audio signal sent to the speaker, h(t) is the speaker-ambient signal. microphone acoustic path, n(t) ambient noise, and m(t) by the microphone It represents the total measured signal. Each measurement frame Frame = {mic_id, t_s, n_idx, gain, clip} 5 It can be associated with the data structure. Here, “mic_id” is the microphone ID, t_s timestamp, n_idx sample order, gain microphone gain, and clip saturation. or indicates a clipping indicator. The module stores microphone samples in 16-bit, 24-bit, or floating-point format. to collect, 44.1 kHz, 48 kHz or another sampling rate supported by the device. operating at that frequency, hardware or software for the microphone and source channel adding timestamps, channel order and geometry of multiple microphones recording, gathering microphone gain, orientation, and saturation information, testing marking the start and end times of the signal and short-circuiting the raw data 15 It can perform operations to transfer resources to the buffer structure. Resource and... Delay between microphone channels is cross-correlation or GCC-PHAT. (Generalized Cross-Correlation Phase Transform) Correlation can be estimated using the Phase Transformation (PHA) method. τ = argmax_τ Σ [M(f) S*(f) / |M(f) S*(f)|] e^{j2πfτ} The module includes not only audio samples but also the measurement mode for each frame. band-based power sufficiency and time synchronization status of the source signal and also generates microphone validation information. This information is then used for subsequent reliability testing. 25 It is used in coefficient calculations. Acoustic ambient sound and digital input are used. Source audio reference and session configuration are being retrieved; synchronized as output. Raw microphone frames, source reference frames, and measurement metadata. It is produced. The invention is a closed-loop system based on physical environment acoustics parameters. pre-signal processing module (120) located in the system that produces music composition, Raw microphone data can be made usable for acoustic parameter extraction. - 11 - This involves the direct use of the microphone signal; microphone frequency The answer is electrical noise, source-microphone delay, speaker echo, device. multi-stage movement and clipping can lead to erroneous results. It is applied to the process. The preferred procedure is as follows: removal of the direct current component, 5 Calibration compensation is applied based on the microphone frequency response, source and time of microphone signals GCC-PHAT or cross-correlation Alignment is performed using the source reference NLMS (Normalized Least Mean Square (Normalized Least Mean Square) or frequency domain Acoustic echo cancellation with block adaptive filter, residual echo Wiener 10 filtering, spectral suppression, or MMSE (Minimum Mean Square Error – Reducing the signal using the Minimum Mean Squared Error (MMS) method, with Hann windows, and Separation into overlapping frames, STFT (Short-Time Fourier Transform) Creating time-frequency representations using the Fourier Transform (with time) and 1 / 1 or extraction of band energies via a 1 / 3 octave filter bank. 15 The direct current component can be removed in the following way, for example: x_dc(t) = x(t) − (1 / N) Σ x(n) Calibration compensation is performed using the microphone's known frequency response H_mic(f) 20 X_cal(f) = X(f) / H_mic(f) It can be applied in this way. In acoustic echo reduction, estimated echo and error are considered. signal 25 ŷ(t) = ŵ(t) ∗ s(t), e(t) = m(t) − ŷ(t) It can be calculated as follows. NLMS filter update is an example. ŵ(t+1) = ŵ(t) + (μ / (||s(t)||² + ε)) e(t) s(t) 30 This can be done as follows: Short-time Fourier transform - 12 - X(k, m) = Σ_{n=0}^{N−1} x(n + mH) w(n) e^{−j2πkn / N} It can be implemented through this relationship. When using multiple microphones, delay-and-sum and MVDR (Minimum Variance) are used. Distortionless Response (GSC 5) (Generalized Sidelobe Canceller) Beamforming can be implemented. Simultaneously, user speech or independent communication is possible. To prevent incorrect updating of the adaptive filter when external noise is present, use a dual system. Speech detection is available. Microphone saturation for each processed frame. or the absence of digital clipping, ensuring sufficient signal availability in the relevant frequency bands. energy content, source-microphone coherence being above the lower threshold, time synchronization error remaining within the allowed range and double speech or sudden The conditions for controlling whether the coup attempt should be significant enough to disrupt the parameter calculations are being checked. These checks are performed. As a result of these checks, a valid or invalid option is determined for each frame. A flag and frame confidence score is generated. Synchronized raw 15 is used as input. Microphone and source reference frames are being captured; the output is cleaned. time domain signal, spectrogram, mel-spectrogram, octave band energies, phase Information and a framework confidence score are generated. The invention relates to closed-loop acoustics based on physical environment acoustic parameters. Acoustic parameter extraction module 20 in the music composition generating system (130), through preprocessed signal, impulse response and time-frequency representations It calculates physical and statistical parameters that represent the environment. Preference In the implemented core application, the decision engine uses RT60 in at least the mid-frequency bands. (Reverberation Time 60 – 60 decibels) or equivalent decay Duration, C80 (Clarity 80 – 80 milliseconds of musical clarity) or equivalent musical clarity 25 indicator, band-based SNR (Signal-to-Noise Ratio) or noise masking vector with low frequency resonance centers and bass The bass resonance score (BRS) is transmitted. Each parameter... P = {value, confidence} 30 - 13 - the measured value and the confidence coefficient together It can be held. In the RT60 calculation, the energy decay of the impulse response is separated into frequency bands. curve with Schroeder backward integration E(t) = ∫_t^∞ h²(τ) dτ It can be calculated as follows. The regression slope is calculated using α. RT60 = −60 / (α 20 log₁₀ e) 10 It can be calculated as follows. If the regression fit coefficient is low, the measurement It can be rejected or marked as low confidence. SNR in the calculation, noise power is from quiet intervals, residual after echo cancellation. from the signal, minimum statistical method or sound activity detection 15 This can be predicted from the result. Frequency transfer function and coherence In the calculation, bands with low coherence are considered a reliable measurement of transmission. It is not accepted. For the bass resonance score, the center frequency of each resonance is f_r, and the peak frequency is... Height A_r, bandwidth BW_r, quality factor Q_r and time stability T_r 20 It can be calculated as follows: BRS = Σ_r w_r · A_r · Q_r · T_r The fundamental frequency of each symbolic note event is 25 f₀ = 440 · 2^{(midi−69) / 12} The resonance risk value of the note can be calculated as follows. R_note = Σ_r A_r · exp(−(f₀ − f_r)² / (2 σ_r²)) - 14 - This can be determined by the cost of temporal overlap. C_overlap = Σ_i Σ_j≠i max(0, d_i − IOI_{ij}) / d_i It can be calculated as follows. The parameter confidence coefficient is 5. conf = g(coh, E_src, R², rep, frame_valid) coherence, source signal energy, regression fit, measurement in this form. It can be generated from repeatability and framework validity scores. 10 as input. cleaned signal, impulse response, spectrogram, octave band energies, and frame. Security information is being retrieved; output includes RT60, C80, D50, SNR, and transfer function. Coherence, BRS, resonance frequencies, noise profile, and reliability for each parameter. a coefficient is generated. The invention is a closed-loop system based on physical environment acoustics parameters. environmental classification and temporality within the system that produces music compositions The monitoring module (140) cleans, normalizes, and stabilizes the acoustic parameters. combining them into an ordered feature vector and tracking the change of the environment over time. It does so. For incomplete or unreliable values, the previous reliable measurement is used. Adjacent band interpolation, moving median, or model-based missing value 20 Completion can be applied. Outliers are measured using the z-score, median, and absolute deviation. The quadrants can be identified by their openness or by the isolation forest. Preferred normalized core variables A_norm = [RT60_n, C80_n, SNR_n, BRS_n, conf, Δenv] 25 It can be generated in this way. The final kernel feature vector. V = [A_norm, conf_joint, Δrate] They can be combined in this way. Here, conf_joint is the combined confidence factor, Δrate represents the rate of change of the medium over time. The medium - 15 - classification rule-based decision tree, random forest, SVM (Support Vector) Support Vector Machine), gradient boosting, multilayer neural network, LSTM (Long Short-Term Memory) or Transformer This can be done with the model. The system uses a multi-label probability vector instead of a single class. P(class) = [p₁, p₂, …, p_K] It can produce temporal smoothing. V_smooth(t) = α V(t) + (1−α) V_smooth(t−1) 10 It can be implemented with a Kalman filter, HMM (Hidden Markov Model). Markov Model), CUSUM (Cumulative Sum) or change By identifying the point of origin, permanent changes can be distinguished from short-term events. The environment class is a supportive explanatory output; the composition decision is only for class 15. It is not provided according to the label. The decision engine uses fixed-order, normalized, and A confidence-weighted numerical feature vector is transmitted. Acoustic input is provided. Parameters and confidence coefficients are obtained; normalized output is provided. acoustic feature vector, ambient class probabilities, combined confidence value, and ambient A change score is generated. 20 The invention relates to closed-loop acoustics based on physical environment acoustic parameters. acoustic-composition decision in the system that produces music composition engine (150), physical environment measurements and source composition of music It is the core component that establishes a direct technical transformation between the variables. Decision engine; normalized acoustic feature vector, feature confidence coefficients, environment 25 class possibilities, available composition parameters, user and license restrictions, The device's resource limits and the musical structure of the previous segment are used as input. It is receiving. Target composition vector C* = [ND*, Poly*, IOI*, HarmRate*, DynRange*, Oct*, Inst*] - 16 - It can be structured in this way. Changes can be made to the preferred application. The technical aspects are carried out in the following order of priority: first, the note density is reduced; sufficient If not, the polyphony level is reduced; if temporal overlap continues, the notes... The start time between these two points is increased or the sustain time is shortened; this If the changes are insufficient, the tempo is adjusted in a controlled and limited manner; 5 Note, octave, and instrument selection based on resonance and masking costs. is being done. Note density, polyphony, interval between notes, harmonic shift Speed ​​and dynamic range can be expressed by the following relationships: ND = N_notes / T_seg Poly = max_t |{notes active at t}| IOI = mean(t_{i+1} − t_i) HarmRate = N_chord_changes / T_seg DynRange = L_max − L_min 15 Each candidate has a selection cost for sheet music and instrument. Cost(n,i) = w₁·C_mask(n,i) + w₂·C_res(n,i) + w₃·C_overlap(n,i) + w₄·C_dev(n,i) 20 It can be calculated as follows. The decision engine requires mandatory musical and licensing restrictions. among the candidates who provide argmin Cost(n,i) 25 It makes its choice. In nonlinear applications, fuzzy logic is very popular. layered neural network, gradient boosting regression, kernel regression, mixture-of- Experts may use reinforcement learning or multi-objective optimization. However, the core physical relationships cannot be explained by the generative model in any way. Modification is not permitted. The module's outputs are for general music production only. Not a command; the target is note density, polyphony, and the time interval between notes. - 17 - harmonic change rate, dynamic range, allowed octaves, instrument sequencing, and each These are the technical cost values ​​related to the selection. The input is a reliability-weighted acoustic feature. vector, environment profile, user or license restrictions, and current composition. The output received is the target composition vector and note-instrument cost. The matrix, allowed parameter ranges, and production decision package are generated. 5 The invention relates to closed-loop acoustics based on physical environment acoustic parameters. User and musical constraint module within the music composition generating system. (160), musical integrity when creating an acoustically suitable composition, It ensures the protection of usage rights and user preferences. Mandatory. Restrictions; the resource may not be modified without a license or usage right, permission 10 The use of an unspecified sound or instrument model, key, mode, and preserving the integrity of the tonal center, the physical or virtual sound of the instrument The range should not be exceeded, and the number of channels supported by the device should not be exceeded to ensure safety. This may include maintaining sound level and latency limits. Optional. Constraints include music genre, energy and emotional level, vocal preference, instrument preferences, 15 This can include target time and complexity levels. Constraints can be expressed using a set of requirements: S = {c | g_j(c) ≤ 0, h_k(c) = 0} Here, g_j represents the mandatory inequality constraints, and h_k represents the musical equality constraints. The module assesses whether the goals generated by the decision engine are feasible. checking that it is not; setting unimplemented targets to the nearest permitted value. It projects and sends a mandatory restriction mask to the production model. Input user preferences, licensing information, device limitations, and decision engine outputs 25 The output includes permitted parameter ranges, mandatory restriction masks, and production. A fitness set is generated. The invention relates to closed-loop acoustics based on physical environment acoustic parameters. conditional music production module in the music composition generating system (170), target composition vector from decision engine and constraint 30 Symbolic musical events using the suitability set from the module It produces. Symbolic event - 18 - e = {t_onset, dur, pitch, vel, inst, ch} It can be expressed as follows. The production methods that can be used are conditional. Transformer, RNN (Recurrent Neural Network) or LSTM, Variational autoencoder, diffusion-based symbolic model, graph-based harmony 5 It includes the model and the rule-supported generative model. In probabilistic generation, top-k, nucleus sampling, beam search, or temperature-controlled sampling It is applicable. The raw output of the generative model goes directly into the voice synthesis module. Not being sent. Raw output, acoustic properties A, user preferences U and license 10. the allowed space defined by the constraints L E_proj = Π_{A,U,L}(E_raw) It is projected as such. During projection, a low 15 exceeding the note density. Important events can be extracted, and accompanying sounds exceeding the polyphony limit can be reduced. Notes with high resonance cost can be shifted to another octave, reducing noise. Masking costs are high; instruments can be changed, over-chords Changes can be combined, events outside the key or authority. It can be corrected or reproduced. Melodic significance for each note is 20. score S_mel = α·role + β·duration + γ·position It can be calculated as follows; in grade reduction, events with lower importance scores are prioritized first (25). The main melody is preserved by removing certain elements. Generative artificial intelligence overcomes physical limitations. Not the determining factor; a valid sequence of events within physical and musical constraints. It is the generating component. The target composition vector is used as input, along with the allowed parameters. Intervals and previous music segments are taken; acoustic and musical output is provided. Symbolic notes, chords, and rhythmic sequences are produced that conform to the constraints. 30 The invention relates to closed-loop acoustics based on physical environment acoustic parameters. arrangement and instrumentation involved in the system that produces musical compositions - 19 - module (180) distributes symbolic events to instrument and channel layers; octave, timbre, articulation, and sustain values ​​according to the acoustic costs of the environment. It regulates the available pitch range, fundamental frequency, and for each instrument. harmonic energy distribution, spectral centroid, attack, decay, sustain, and Release characteristics, transient prominence, timbre brightness, and source and license 5 Information can be stored. Noise and resonance costs for the instrument. Cost_inst(i) = w_n·C_noise(i) + w_r·C_res(i) It can be calculated as follows. The module converts a high-cost instrument into another one. changing the instrument, melody or bass part to another permitted octave to carry, shorten long sustain times according to the risk of overlap, same frequency reducing simultaneous sounds in that region, changing the chord inversion, staccato, Choose legato or another articulation and set the license limits to 15 with user preference. It can perform preservation operations. Octave or instrument change, Not only based on spectral gain; but also on note resonance risk and noise masking. cost, temporal overlap cost, and deviation from initial composition This is done by evaluating them together. Symbolic musical events and musical notation are used as input. The instrument cost matrix and constraint mask are obtained; acoustic 20 is the output. Reduced costs for instrument or channel planning, octave spread and rendering. An organized symbolic composition is produced. The invention relates to closed-loop acoustics based on physical environment acoustic parameters. sound synthesis and rendering module within the music composition generating system (190), organized symbolic music data into audible sound stream 25 It transforms. Methods that can be used include sample-based synthesis and waveform charting. synthesis, FM (Frequency Modulation) or subtractive synthesis, physical modeling, neural vocoder, neural codec-based production and It involves diffusion-based sound production. The sound is not produced all at once for the entire piece; It can be produced in segments consisting of one or more dimensions. 30 Segment data structure - 20 - Seg = {id, t_start, t_end, events, audio, meta} It can include these areas. Total end-to-end delay. T_e2e = T_meas + T_feat + T_dec + T_gen + T_synth + T_buf 5 It can be calculated as follows: If the value of T_e2e exceeds the T_max limit. The module allows for a shorter production horizon, a lower complexity model, and prior... to the verified template or simply to reproduce the problematic channel It can pass. Symbolic composition and synthesis arranged as input 10 Parameters are received; output is in PCM, WAV, FLAC or compressed format. The rendered audio segment and segment metadata are generated. The invention relates to closed-loop acoustics based on physical environment acoustic parameters. Quality control and verification within the music composition production system. module (200), technical, generated composition and rendered sound segment, 15 It is verified according to musical and acoustic criteria. Technical controls are digital. clipping and true-peak, program loudness, DC component, channel or phase matching, Clicking, popping, and corrupted segments, codec or packaging errors during production. It includes the delay. Musical controls target BPM (Beats Per Minute – Beats per minute), note density and polyphony, key or mode 20 harmony, chord progressions, melodic continuity, instrument range, and intersegmental harmony It includes rhythmic harmony. The difference between the target and the actual composition. ΔC = ||C* − C_realized|| 25 It can be calculated as follows. The technical acceptance gate is measurable as follows: It is based on costs. Spectral masking cost. C_mask = Σ_k max(0, N_k − M_k) / Σ_k N_k 30 resonance cost - 21 - C_res = Σ_n R_note(n) · A_n cost of temporal overlap C_overlap = Σ_i Σ_j≠i max(0, d_i − IOI_{ij}) / d_i 5 transit cost C_trans = ||C_new − C_prev||_w It can be defined as the total technical cost. C_total = λ₁ C_mask + λ₂ C_res + λ₃ C_overlap + λ₄ C_trans It is in this form. 15 for the new composition to be approved. C_total(new) < C_total(prev) − δ_min musical integrity ≥ θ_mus T_e2e ≤ T_max The conditions can be checked. If the conditions are not met, only the problematic channel can be reinstalled. being produced, another octave or instrument is being selected, note density The composition is either reduced or the last verified composition is preserved. This structure, The system not only produces a different sound; it also provides a measurable comparison to the previous output. This necessitates providing technical improvements. 25 rendered files are used as input. audio, target composition vector, environment profile, and previously validated The costs of the composition are being collected; the approved audio segment is being processed as output, again. Production demand or a decision to maintain the existing segment is being generated. The invention relates to closed-loop acoustics based on physical environment acoustic parameters. Streaming, caching, and transition management within a music composition generating system 30 module (210) transmits the sound segments that have passed quality control to the user device. It transmits seamlessly and with low latency. The module; the next segment - 22 - pre-generate and cache, use dual buffer or ring buffer, RTP (Real-time Transport Protocol), WebSockets use HTTP-based streaming or on-device data bus to streamline packet order. and managing lost packets, applying jitter buffer, beat, meter or musical. Crossing at the section boundary, crossing between old and new segments 5 implementing and stealing a secure backup segment during network or production delays. They are able to carry out their transactions. Traverse y(t) = α(t) · x_new(t) + (1 − α(t)) · x_old(t) 10 It can be implemented in this way. The module determines which beat or measure the new composition belongs to. feedback module and user device that it will be activated at the limit It reports that parameter updates cause sudden discontinuities in the audio stream. It ensures that it does not create. Approved audio segments as input and 15 Timing information is being collected; the output is uninterrupted audio streaming, playback time. The stamps and information for the next measurement window are generated. The invention relates to closed-loop acoustics based on physical environment acoustic parameters. real-time feedback included in the music composition generating system module (220) re-measures the environment while the generated music is playing and the new 20 whether acoustic conditions necessitate a composition update It determines. The weighted average between the new feature vector V_new and the reference vector V_ref change ΔV = Σ_i w_i |V_new,i − V_ref,i| It can be calculated as follows: Mahalanobis distance in correlated features. d_M = √((V_new − V_ref)ᵀ Σ⁻¹ (V_new − V_ref)) 30 - 23 - It can be used to prevent a single sudden event from generating an update. hysteresis, sequential window validation, minimum stability time, median filter, CUSUM or change point detection can be applied. Composition update only ΔV > θ_Δ N_consec ≥ N_min C_total(new) < C_total(prev) − δ_min It is accepted when all conditions are met. The new targets are 10-phase. implementation C(t) = (1 − β) C_prev + β C* This can be achieved with [company name]. Even if the acoustic profile has changed, the new composition is 15. If the technical cost is not reduced sufficiently, the current composition is maintained. This The decision eliminates unnecessary model runs, energy consumption, memory usage, and network usage. It reduces traffic. New microphone measurement as input, reference acoustic profile, Current and candidate composition costs are collected; profile update is the output. command, request to produce a new composition or preserve an existing composition 20 The decision is being made. The invention relates to closed-loop acoustics based on physical environment acoustic parameters. a temporary acoustic buffer located within the system that produces music compositions (230), short-term data required for acoustic analysis and recently validated acoustics It temporarily stores profiles. In the preferred application, the public buffer is 25. its structure is being used, and as new samples become available, they are built upon the old raw samples. It is being written. Data lifecycle T_keep ≤ T_analysis 30 - 24 - It can be limited to. Limited-time raw microphone frames in the buffer, final acoustic characteristic vectors, parameter confidence coefficients, measurement mode, and time. Stamp, reference profile version, final composition parameters, and environment. The change score can be maintained. Sent to the cloud or edge system. De-identified feature pack 5 FeatPkg = {SessionAlias, RT60_k, C80, SNR_k, BRS, EnvironmentClass, Confidence, Timestamp} It can be in this form. Raw and processed short-term measurements as input. 10 The output includes the current acoustic profile, a historical comparison window, and A feature package is being developed that complies with privacy restrictions. The invention relates to closed-loop acoustics based on physical environment acoustic parameters. model, rule and instrument data included in the system that produces music composition repository (240), technical configurations used in the operation of the system and 15 It stores versioned data. The repository contains raw user microphone recordings. It is not necessary. Spectral analysis and environmental classification models in the warehouse, temporal tracking model parameters, acoustic-composition transformation weights, instrument sound ranges and spectral energy profiles, note-frequency mappings, Musical production templates, usage rights and licensing information, quality control thresholds 20 Device-class specific delay and resource limits may apply. In the updated core implementation, θ_Δ, δ_min, N_min, λ₁…λ₄, w₁…w₄, α, β, T_max, and device-based thresholds are also stored in a versioned manner. These values different depending on media type, device class, music genre, or user profile. Configuration sets can be stored in this format. The system administrator input is 25. configurations, training results, model updates, and instrument metadata Data is collected; version-controlled model or rule parameters are given as output. Bridges and instrument features are manufactured. The end-to-end workflow of the system consists of the following main process steps: It consists of. The user can acoustically 30 through the sound output and user device (100). It initiates a music production process adapted to the environment. The system includes a microphone and... Number of speakers, sampling frequency, device processing capacity, maximum allowed - 25 - session-specific techniques such as latency, user preferences, and data processing permissions The system determines the parameters. The system ensures the reliability of the existing acoustic profile and the device. or changes in microphone position, frequency content of the source signal and environment Active calibration, passive monitoring, or hybrid methods are used to assess the level of change. It selects one of the measurement modes. For first use or current profile 5 Active calibration when reliability is lost, and during normal music playback. Passive monitoring can be implemented. Microphone and reference data acquisition module (110), microphone data taken from the environment The digital source audio signal is sent to the speaker simultaneously with the signal. It collects source and microphone signals; timestamp, cross-correlation 10 or time-wise using delay estimation methods such as GCC-PHAT. is aligned. Front signal processing module (120); microphone calibration compensation, removal of the direct current component, resampling, band limiting, and It performs source-microphone time alignment operations. speaker the source component can be filtered using methods such as NLMS or frequency domain adaptive filtering. The signal is separated from the microphone signal; then STFT, spectrogram, and octave band are determined. Energy representations are being created. Acoustic parameter extraction module (130) extracts at least RT60 from the processed signal. or equivalent reverberation time, C80 or equivalent musical clarity value, band It calculates the SNR and low-frequency bass resonance score. The calculated 20 each acoustic parameter, source energy adequacy of the measurement, coherence, regression with a confidence factor, taking into account compatibility and repeatability They are correlated. The parameters are normalized to a common scale to determine the reverberation. risk, loss of clarity, bass resonance risk, band-based noise density, combined safety An acoustic property vector is generated that includes the coefficient and the rate of change of the medium. 25 Environment classification and temporal tracking module (140), sequential acoustic properties By comparing the vectors, we can determine whether the measured change is a short-term noise event or It determines whether there has been a permanent change in environment. Acoustic-composition decision engine (150), normalized acoustics Features include target note density, polyphony level, inter-note interval, 30 harmonic change rate, dynamic range, melodic octave, and instrument selection. It converts to parameters. Changes in the preferred application. - 26 - Firstly, note density, then polyphony and the interval between notes. is being carried out; if these are insufficient, the pace is being controlled. The selection of notes, octaves, and instruments is being modified based on resonance and spectral factors. masking, temporal overlap, and costs of deviation from the existing composition They are evaluated together. User and musical constraint module (160); music genre, 5 key or authority, target duration, instrument preferences, license and usage rights, Constraints such as device processing capacity and maximum allowable latency affect the decision engine. It applies this to the output. Conditional music production module (170), target composition parameters According to this, symbolic notes produce rhythm, melody, harmony, and accompaniment. Productive 10 The raw output of the model undergoes acoustic and musical constraint testing. The Arrangement and Instrumentation Module (180) arranges note events using instruments. It distributes the notes into its channels; notes that resonate with each other are transferred to another octave. can transport, shorten sustain times or masking costs It can select low-end instruments. Sound synthesis and rendering module (190), 15 verified symbolic composition into playable sound segments It transforms. Quality control and verification module (200), loudness of the generated sound, true-peak, clipping, target BPM, note density, polyphony, tonal integrity, and It checks compliance with production delay criteria. Also, the beginning 20 spectral masking, resonance for the new composition with its composition, Temporal overlap and segment transition costs are calculated. New The composition is determined solely by the minimum technical requirements based on the existing composition. It facilitates recovery, meets the threshold of musical integrity, and allows for delay. It is approved if it does not exceed the limit. Approved audio segment, stream, cache and 25 The beat is transmitted to the user device via the transition management module (210), It is played continuously within the limits of the measure or musical section. Real-time feedback module (220) monitors the environment during audio playback. It is re-measuring. The threshold difference between the new acoustic profile and the reference profile is... exceeding its value, the change continuing over a certain number of consecutive measurements, and 30 The proposed new composition has a measurable technical equivalent to the existing composition. If improvements are achieved, the composition parameters are updated. This - 27 - If the conditions are not met, the current composition is maintained and the system becomes passive. It continues to monitor. Thus, unnecessary reproduction, processor usage, Energy consumption and network traffic are prevented. In the preferred implementation example, the system initially uses a logarithmic sine wave. Performing active acoustic calibration using scanning, music playback 5 During this process, passive tracking based on the source audio reference is applied. Active Calibration results in at least mid-frequency reverberation time and musical clarity. value, band-based signal-to-noise ratio, and low-frequency resonance map are calculated. The calculated values, along with measurement confidence coefficients, are then presented. It is converted into a normalized acoustic feature vector. Normalized 10 the determined acoustic feature vector, defined transformation functions and decision rules regarding note density, polyphony level, and the beginning of intervals between notes duration, harmonic change rate, melodic register range, octave, and instrument selection. It is converted into parameters. For example, having a high reverberation time. In a medium, the density of notes and the number of simultaneous sounds can be reduced, with notes being 15. The time interval between them can be increased, or instruments that use longer sustain can be used more It can be replaced with instruments that have a short decay character. Certain If resonance is found at low frequencies, the corresponding bass notes are transferred to another location. It can be shifted to an octave, note durations can be shortened, or it can be achieved through resonance. A different instrument that does not conflict can be selected. The ambient noise is at a certain level. In cases where frequency bands are concentrated, then melody and dominant instruments It can be transferred to less masked frequency regions. The system processes raw microphone data on the device or EDGE unit, the data is only stored in a short-term temporary buffer memory and then... the deletion, the non-transcription of the conversation content, and the identification of the person 25 not being determined, RT60, C80, SNR and resonance instead of raw sound to the central system transmitting anonymous acoustic characteristics such as score, microphone access from the user obtaining permission, transmitting data via encrypted communication, and storing it for a limited period of time. It anticipates situations where it is necessary to transmit the raw audio outside the device. open information, appropriate legal processing requirements, access control, encrypted storage and 30 Deletion mechanisms are being implemented. - 28 - The system's audio measurement and synthesis layer uses common audio sampling frequencies. Active calibration can be performed on devices compatible with various bit depths. In structures where this function is implemented, the controlled test signal is normal listening. timing in a way that will not disrupt the experience and within the relevant audio output limits. It is anticipated that it will be suitable. Total decision delay, technical cost improvement 5 The rate, the number of unnecessary repetitions, and processor or energy usage can be measured. This can be observed as technical results. The impact of added technical elements, reliability without using weighted feature vectors, without applying technical cost gates, and separately with the results obtained without using closed-loop feedback. They can be compared. 10 20 30

Claims

- 29 - SYSTEMS 1. Sound and music processing technologies, room and ambient acoustics measurement, digital signaling. processing, generative artificial intelligence, symbolic music production, and real-time audio. Physical acoustic properties of a listening environment in the synthesis field 5 measuring audio data via microphone or array of microphones; recording measured audio data over time operating in the frequency domain and time-frequency domain; the reverberation of the medium, noise, resonance, frequency attenuation, spectral masking, and musical clarity. determining the characteristics and resulting acoustic properties of music compositions It creates a new musical structure by converting it into its parameters; at least one processor, 10 memory, data storage units, microphone or microphone array, speaker or one or more audio output units and edge (EDGE) calculation capabilities The invention is designed to be run on multiple user devices. Closed-loop music based on physical environment acoustics parameters It is the system that produces the composition, its feature is; sound output and user device (100), 15 microphone and reference data acquisition module (110), pre-signal processing module (120), acoustic parameter extraction module (130), ambient classification and temporal tracking module (140), acoustic-composition decision engine (150), User and musical constraint module (160), conditional music production module (170), Editing and instrumentation module (180), sound synthesis and rendering 20 module (190), quality control and verification module (200), flow, cache and Transition management module (210), real-time feedback module (220), temporary acoustic properties buffer (230) with model, rule and instrument data repository (240) is characterized by its inclusion.

2. A system that produces closed-loop music compositions according to claim 1, 25 Features include: mobile devices, computers, in-car infotainment systems, and smart speakers. television, headphone system, home audio system or standalone audio production User-initiated acoustic devices that can be implemented in the form of a device. The adaptive music production session consists of music genre, duration, and energy level. Active 30, which takes into account user preferences such as vocal preference and allowed instruments. broadcasting the test signal to be used in calibration through a loudspeaker, transmitting digital samples of the source audio being played as a reference channel, - 30 - Playing approved music segments, with microphone access and privacy preferences. manages user permissions related to device processing capacity, memory, and network. Audio output and audio output channels are technical capabilities that indicate features such as latency and number of audio output channels. It is characterized by containing user devices (100).

3. A system that produces closed-loop music compositions according to claim 1, 5 Its characteristic is time-domain sound that represents the physical acoustic behavior of the medium. simultaneously samples and the source audio reference sent to the audio output device. collecting as, single microphone, stereo microphone or multi-microphone array capable of cross-referencing the delay between the source and microphone channels. Estimated by correlation or GCC-PHAT method, 10 measurements for each frame. mode, band-based power sufficiency of the source signal, time synchronization Microphone and reference data collection that generates status and microphone validity information. It is characterized by containing module (110).

4. A system that produces closed-loop music compositions according to claim 1, Its feature is the removal of the direct current component, resulting in a 15-bit response based on the microphone frequency. Calibration compensation, time alignment of source and microphone signals, NLMS or acoustic echo removal with frequency domain block adaptive filter, no longer echo reduction, Hann window framing, time-frequency representation with STFT and performs band energy extraction processes via an octave filter bank, A valid or invalid flag and a frame confidence score of 20 are assigned to each processed frame. It is characterized by containing the pre-signal processing module (120) that produces the signal.

5. A system that produces closed-loop music compositions according to claim 1, Its characteristic is; at least RT60 or equivalent decay time in the mid-frequency bands, C80 or equivalent musical clarity indicator, band-based SNR or noise level. Masking vector with low frequency resonance centers and bass resonance 25 The score is calculated by combining each parameter's measured value and confidence coefficient. the parameter confidence coefficient maintains the coherence of the source signal. energy, regression fit, measurement repeatability, and frame validity by including an acoustic parameter extraction module (130) that generates scores from it. It is characterized by 30 6. A system that produces closed-loop music compositions according to claim 1, Its feature is a fixed-order feature that cleans and normalizes acoustic parameters. - 31 - incomplete or previous reliable measurement for unreliable values ​​or interpolating, identifying outliers, normalizing Risk of reverberation, loss of clarity, risk of bass resonance, band-based noise intensity. Feature vector 5 including combined confidence coefficient and environmental change rate. the environment that creates and distinguishes permanent changes from short-term events. It is characterized by containing a classification and temporal tracking module (140).

7. A system that produces closed-loop music compositions according to Claim 1, The feature is; the normalized acoustic feature vector of the target note intensity, polyphony level, internote interval, harmonic change rate, dynamics 10 converting intervals into melodic octave and instrument selection parameters, The changes were first in note density, then in polyphony and the interval between notes, If that's not enough, they perform on tempo, each candidate learning sheet music and instrument. spectral masking, resonance, temporal overlap, and current acoustic-15 which considers the costs of deviations from composition together It is characterized by containing a composition decision engine (150).

8. A system that produces closed-loop music compositions according to Claim 1, The characteristic is that the source is not modified without a license or usage rights. The use of unauthorized sound or instrument models is prohibited, key, preserving the integrity of the mode and tonal center, the instrument's sound range is 20 not exceeding the number of device channels, ensuring a safe sound level, and Mandatory constraints such as maintaining latency limits; music genre, energy level, vocal preference, instrument preferences, target duration and complexity level, etc. Applying optional restrictions, unimplementable targets to the nearest authorized one. 25 by including the user and musical constraint module (160) that projects value. It is characteristic.

9. A system that produces closed-loop music compositions according to Claim 1, This feature uses the target composition vector and the fitness set. acoustics are the raw output of the generative model that produces symbolic musical events. 30 permitted, defined by features, user preferences, and license restrictions. projecting into space, extracting low-importance events that exceed note density, notes with high resonance cost that reduce accompanying sounds exceeding the polyphony limit - 32 - instruments that carry another octave and have high noise masking costs It is characterized by containing a conditional music production module (170) that modifies.

10. A system that produces closed-loop music compositions according to Claim 1, Its characteristic feature is the distribution of symbolic events across instrument and channel layers, octave. timbre, articulation and sustain values ​​are adjusted according to the acoustic costs of the environment. regulating, replacing a high-cost instrument with another instrument, a long sustain that carries the melody or bass part to another permitted octave shortening their durations based on the risk of overlap and peers in the same frequency region Editing and instrumentation module that reduces timed sounds (180) It is characterized by its inclusion. 10 11. A system that produces closed-loop music compositions according to Claim 1, Its feature is to convert organized symbolic musical data into an audible audio stream. converting sound into segments consisting of one or more measures generating, the total end-to-end delay exceeding the maximum allowable delay limit If it exceeds this limit, it will lead to a shorter production horizon and a lower complexity model. or voice synthesis that can only proceed to reproduce the problematic channel. and is characterized by containing the render module (190).

12. A system that produces closed-loop music compositions according to Claim 1, Its feature is the technical aspect of the generated composition and rendered sound segment. Verified according to musical and acoustic criteria, spectral masking cost, 20 through resonance cost, temporal overlap cost and transition cost calculating the total technical cost, the new composition is only available. providing minimal technical improvements according to the composition, musical integrity if it meets the threshold and does not exceed the permitted delay limit It is characterized by containing the quality control and verification module (200) that approves. 25 13. A system that produces closed-loop music compositions according to Claim 1, Its feature is to transmit quality-controlled audio segments to the user device. transmitting seamlessly and with low latency, anticipating the next segment. Generating and caching, transitioning between beats, measures, or musical sections, implementing cross-functionality between old and new segments and network or production 30 secure backup segment stealing stream, cache and transition management in latency It is characterized by containing module (210). - 33 - 14. A system that produces closed-loop music compositions according to Claim 1, This new feature re-measures the environment while the generated music is playing. Hysteresis calculates the weighted variation between the vector and the reference vector. implementing sequential window validation or change point detection, The composition update is only possible with the new acoustic profile and reference profile 5. If the difference between them exceeds the threshold value, the change occurs across successive measurements. continuation and the new composition providing measurable technical improvement real-time feedback that accepts the conditions when both are met. It is characterized by containing module (220).

15. A system that produces closed-loop music compositions according to Claim 1, 10 Its feature is that it provides the short-term data and recently validated results necessary for acoustic analysis. limited acoustic profiles are temporarily concealed using a ring buffer structure. Temporary raw microphone frames, final acoustic characteristic vectors, parameters confidence coefficients, measurement mode, timestamp, reference profile version, latest 15 in the central system that keeps track of composition parameters and environmental change score a temporary acoustic properties buffer that only transmits an anonymous acoustic properties package (230) is characterized by its inclusion.

16. A system that produces closed-loop music compositions according to Claim 1, Features include spectral analysis and environmental classification models, and temporal tracking. model parameters, acoustic-composition conversion weights, instrument 20 sound ranges and spectral energy profiles, note-frequency mappings, musical production templates, usage rights and license information, quality control thresholds Device class-specific latency and resource limits in a versioned manner. the model, rule and It is characterized by containing an instrument data repository (240). 25 17. A system that produces closed-loop music compositions according to Claim 1, The feature is that raw microphone data is processed on the device or edge unit, and the data is then processed. It is only stored in a short-term temporary buffer memory and then deleted, The content of the conversation was not transcribed and the person's identity was not determined. Instead of raw audio, the central system uses 30 parameters such as RT60, C80, SNR, and resonance score. Anonymous acoustic characteristics are transmitted, and the user grants permission to access the microphone. temporary where data is received, transmitted via encrypted communication, and stored for a limited time. - 34 - acoustic properties buffer (230) and model, rule and instrument data repository (240) It is characterized by its inclusion.

18. A system that produces closed-loop music compositions according to Claim 1, Its feature is active acoustics initially using logarithmic sine wave scanning. Calibration was performed, the source sound reference was 5 during music playback. Passive monitoring based on active calibration results in at least medium frequency. reverberation time, musical clarity value, band-based signal-to-noise ratio, and low-frequency resonance map is calculated, the calculated values normalized acoustic characteristic vector with measurement confidence coefficients converted audio output and user device (100), microphone and reference data 10 by including the collection module (110) and the acoustic parameter extraction module (130) It is characteristic.

19. A system that produces closed-loop music compositions according to Claim 1, Its feature is note density and equivalence in an environment with a long reverberation time. the number of timed sounds is reduced, the time between notes is increased or 15 Instruments that use long sustain have a shorter decay characteristic. replaced with instruments, resonance found at certain low frequencies in this case, the relevant bass notes are shifted to another octave, and the note durations the environment where it is shortened or a different instrument that does not coincide with the resonance is chosen If the noise is concentrated in certain frequency bands, the melody and 20 where dominant instruments are shifted to less masked frequency regions acoustic-composition decision engine (150), conditional music production module (170) and is characterized by including the regulation and instrumentation module (180).

20. A system that produces closed-loop music compositions according to Claim 1, The feature is that the new composition is determined only by the 25 compared to the existing composition. ensuring minimum technical improvements, meeting the threshold of musical integrity, and permission It is approved if the given delay limit is not exceeded; otherwise, the conditions will be met. only the problematic channel is reproduced, no other octave or instrument selected, note density reduced, or the last verified composition 30 by including the quality control and verification module (200) in which it is protected It is characteristic.

Images