A music arrangement method based on environmental monitoring data

Abstract

The present invention belongs to the field of intelligent music arrangement technology, and is particularly a music arrangement method based on environmental monitoring data, comprising the following steps: S1, obtaining the original signal of weather conditions in nature, preprocessing the natural signal, and extracting weather condition characteristic parameters; S2, selecting and adapting music clips, and extracting characteristic parameters of audio signals; S3, establishing a parameter mapping model between two characteristic parameter sets; S4, predicting and synthesizing new music clips. The present invention creatively takes the weather factors in the environment as the primary influencing factor of intelligent music arrangement, and conducts intelligent adaptation and generation based on the weather conditions in the environment for finished, high-quality music works that already have certain aesthetic value, thereby avoiding the defects of low enjoyability and insufficient artistry of music clips in existing artificial intelligence music arrangement algorithms. The intelligent music adaptation based on the aesthetic needs of modern people is of great benefit to the promotion and use of the present invention.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent music arrangement, in particular to a music arrangement method based on environmental monitoring data. Background Art

[0002] With the development of computer science and the rise of the artificial intelligence industry, artificial intelligence arrangement and composition has become a hot topic in the field of artificial intelligence. The application of artificial intelligence in music can be divided into two aspects, namely artificial intelligence arrangement and composition. Artificial intelligence composition has gone through three stages: random composition, logical composition and intelligent composition. Among them, the algorithm of intelligent composition is more complex, and the generated music has a higher level of intelligence.

[0003] Existing artificial intelligence composition technology relies on few factors. There are artificial intelligence composition and arrangement methods that use weather factors in the environment as the main influencing factors. Most of them use language factors such as lyrics and lyrics emotions. By classifying and judging the emotions of related emotional words, the style of the generated music is predicted to achieve the purpose of changing the final style of the music; or use brain wave signals (such as a brain wave composition method disclosed in patent CN108628450A), skin electrode signals (a music-assisted emotional regulation intelligent method that matches the caffeine content of beverages disclosed in CN113274002A, in which the physiological detection module detects muscle and skin electrode signals to feedback human emotions, which is convenient for matching with the caffeine content in the beverage. caffeine content is combined to generate and adjust music) and other digital signals as influencing factors to generate music, convert these digital signals that have nothing to do with music into audio signals, perform parameterized prediction and music analysis and generation; or predict and generate new music fragments based on the existing music library (such as a music arrangement method and system based on deep learning disclosed in patent CN109785818A, and an intelligent music arrangement method and system disclosed in patent CN109448684A), that is: by establishing a deep learning model and repeatedly training the built model, the music fragments are predicted and generated on the basis of a large amount of music in the music library. When constructing an artificial intelligence neural network, it is mainly by means of constructing a recurrent neural network and a long-term memory network. The construction of the above two neural networks shows good performance in terms of time series. Continuously training and enriching artificial intelligence neural networks can achieve the purpose of generating more complex music, such as from the synthesis of monophonic music to the prediction and generation of polyphonic music. On the other hand, in terms of artificial intelligence music arrangement, at this stage, it is possible to configure chords according to a single-tone melody, or predict the next tone according to the gene frequency of the previous tone. (For example, the music composition method of patent CN110634465A: mobile terminal, data processing method and music composition system provide users with music elements related to intelligent audio configuration) However, the music generated by these algorithms is difficult to apply to actual production and life.

[0004] To sum up, the existing research results on artificial intelligence arrangement and composition are difficult to be directly applied to real life and widely appreciated by the public. Given the singleness of the music generated, its appreciation and aesthetic value are not high, it is difficult to apply and promote it in production and life. Summary of the invention

[0005] 1. Technical issues to be solved

[0006] In view of the shortcomings of the prior art, the present invention provides a music arrangement method based on environmental monitoring data, which solves the problem that the existing research results on artificial intelligence music arrangement and composition are difficult to be directly applied to real life and widely appreciated by the public. Due to the singleness of the generated music, the appreciation and aesthetic value are not high, and it is difficult to apply and promote it in production and life.

[0007] (II) Technical solution

[0008] The present invention realizes the process of generating associations between environmental factors and musical elements, and achieves the application effect of generating music clips of corresponding emotions or styles according to changes in weather conditions. It avoids the defects of low enjoyability and insufficient artistry of music clips in existing artificial intelligence arrangement algorithms, adapts and recreates existing and established music clips, ensures the beauty of music, and has important significance for promotion and use in actual living environments. The present invention creatively links weather elements in the natural environment with musical elements, and applies them to the construction of single and boring artificial landscapes and smart cities in actual life, which will create a higher level of aesthetic experience for people's lives and create a good humanistic environment. At the same time, the present invention, by means of a method for constructing a parameter mapping model, establishes a diversified correspondence between weather condition feature parameters and feature parameters of music clips, predicts the time domain and frequency domain feature parameters corresponding to audio signals under specific weather conditions according to this diversified parameter mapping model, and predicts and generates music clips on this basis. This diversified parameter mapping model can achieve as close as possible to the various musical elements (rhythm, mode, loudness, speed, timbre and polyphonic duet, etc.) to reflect the complex and changeable weather conditions, so as to achieve the generation of music clips that are suitable for the complex and changeable weather conditions, avoiding the problem of generating a single and boring track under a single corresponding relationship. In terms of music adaptation, by means of the method of judging and adapting the major and minor modes of the original music clips, the music emotions are generally divided into two categories: positive emotions and negative emotions, and on this basis, they are divided step by step. This change method can generally express most of the complex human emotions, and more music theory knowledge is applied to the field of artificial intelligence arrangement, ensuring the artistic beauty of the music clips adapted and recreated. The above-mentioned numerous advantages that can be achieved have a good promoting effect on the actual application and promotion process of the present invention.

[0009] In order to achieve the above-mentioned purpose, the present invention specifically adopts the following technical solutions:

[0010] A music arrangement method based on environmental monitoring data comprises the following steps:

[0011] S1. By means of environmental load monitoring, which is an important part of structural health monitoring, the original signal of weather conditions in nature is obtained, and the natural signal is preprocessed to extract characteristic parameters of weather conditions;

[0012] The method of obtaining natural signals is: using the sensor equipment of a small meteorological observation station to obtain weather factors in the natural environment, thereby obtaining the original signal of weather conditions, wherein the weather factors include temperature, humidity, wind load signal, rainfall and light intensity;

[0013] The natural signal preprocessing method is: performing noise reduction and frame division processing on the acquired natural signal;

[0014] Extracting weather condition characteristic parameters includes extracting statistical characteristics, time domain characteristic parameters and frequency domain characteristic parameters of weather signals; wherein, statistical characteristics include mean and variance of signals; time domain characteristic parameters need to extract short-time energy and average amplitude of signals; frequency domain characteristic parameters need to extract and obtain short-time power spectrum of signals;

[0015] S2, selecting and adapting music clips, performing short-time time domain analysis and short-time frequency domain analysis on the audio signal, and extracting characteristic parameters of the audio signal;

[0016] The specific method of selecting and adapting music clips is: judging the major and minor scales and the emotional color of the selected music, and then using audio production software to adapt and recreate the opposite emotional color, wherein the recreation adopts the PN recreation method, and the music clips with positive emotions are recorded as positive music, and if there are richer accompaniment parts, they can be recorded as PA, that is, positive accompaniment; on the contrary, the music clips and accompaniment parts with negative emotions can be recorded as NM and NA; wherein, when processing audio clips, the clips with the most vivid and prominent emotional colors in the music are selected as much as possible to establish the music library;

[0017] The characteristic parameters of the extracted audio signal include short-time average amplitude, frequency, short-time energy and power spectrum density parameters;

[0018] S3, based on the statistical feature analysis and spectral feature analysis of the natural signal obtained in step S1 and the audio signal obtained in step S2, a reference point is selected to establish a parameter mapping model between the two feature parameter sets; taking temperature as an example, 15°C is used as the reference point of the parameter model, ≥15°C corresponds to the spectral parameters of the PM and PA music clips, and <15°C corresponds to the spectral parameters of the NM and PA music clips;

[0019] S4. Predict and synthesize a new music clip according to the spectrum parameters of the music clip mapped by the parameter mapping model, and export and play the new music clip through an audio output device.

[0020] The prediction and generation method of the audio clip is as follows: when processing the original audio clip, the audio clip must first be preprocessed, that is, frame processing and windowing processing, and then the audio clip is subjected to pitch detection to obtain pitch period information, and the frame-by-frame prediction coefficient a is obtained through linear prediction analysis. i , synthesize the music clip according to the obtained prediction coefficient α and pitch parameter through linear prediction analysis; when performing data superposition in signal synthesis, the overlap-addition method is used to overlap and connect the data frames into a continuous and smooth data stream to avoid interruption or jump of the audio signal. When performing data superposition in signal synthesis, the overlap-addition method is used to overlap and connect the data frames into a continuous and smooth data stream to avoid interruption or jump of the audio signal. Loudness, speed and pitch are three important factors for human perception of music, so the adaptation of music will focus on the speed and pitch change processing of audio signals and conduct intelligent adaptation.

[0021] (III) Beneficial effects

[0022] Compared with the prior art, the present invention provides a music arrangement method based on environmental monitoring data, which has the following beneficial effects:

[0023] The present invention creatively takes the weather factor in the environment as the primary influencing factor of intelligent arrangement, and performs intelligent adaptation and generation based on the weather conditions in the environment for finished, high-quality music works that already have certain aesthetic value, thereby avoiding the defects of low enjoyability and insufficient artistry of music clips in existing artificial intelligence arrangement algorithms. The intelligent music adaptation based on the aesthetic needs of modern people is of great benefit to the promotion and use of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a flow chart of the technical solution of the present invention;

[0025] Figure 2 A processing flow chart of the algorithm of the present invention applied to the actual processing process;

[0026] Figure 3 A schematic diagram of the P and N segments of the music score for PN rewriting using the music "Twinkle Twinkle Little Star" of the present invention as an example;

[0027] Figure 4 A flow chart is generated for the audio segment analysis of the present invention. DETAILED DESCRIPTION

[0028] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0029] Example

[0030] The algorithm flow of this embodiment can be mainly divided into two modules:

[0031] Environmental load monitoring module: With the help of environmental load monitoring, an important part of structural health monitoring, the original signal of weather conditions in nature is obtained. The processing operation flow of the original signal of weather conditions is as follows: signal preprocessing, time domain feature parameter extraction, frequency domain feature parameter extraction. The above processing flow is performed on multiple original signals that can describe weather conditions to obtain relevant feature parameters. (such as Figure 1 (shown)

[0032] Music library establishment and audio signal synthesis output module: The main operation process of this module is as follows: music adaptation and music library establishment, time domain and frequency domain analysis of audio signals, extraction of characteristic parameters of audio signals, establishment of parameter mapping model, and prediction and synthesis of audio signals by characteristic parameters. (e.g. Figure 1 (shown)

[0033] like Figure 1-4 As shown, a music arrangement method based on environmental monitoring data proposed in one embodiment of the present invention includes the following steps:

[0034] S1. By means of environmental load monitoring, which is an important part of structural health monitoring, the original signal of weather conditions in nature is obtained, and the natural signal is preprocessed to extract characteristic parameters of weather conditions;

[0035] Environmental load monitoring: With the help of sensor equipment of small meteorological observation stations, weather factors in the natural environment, such as temperature, humidity, wind load signal, rainfall, light intensity, etc., are obtained to obtain the original signal of weather conditions;

[0036] Signal preprocessing: Frame the acquired natural signal to facilitate short-term time domain and frequency domain processing in the next process;

[0037] Extract weather condition characteristic parameters: extract statistical characteristics, time domain characteristic parameters and frequency domain characteristic parameters of weather signals respectively; statistical characteristics include the mean and variance of the signal; time domain characteristic parameters need to extract the short-time energy and average amplitude of the signal; frequency domain characteristic parameters need to extract and obtain the short-time power spectrum of the signal;

[0038] S2, selecting and adapting music clips, performing short-time time domain analysis and short-time frequency domain analysis on the audio signal, and extracting characteristic parameters of the audio signal;

[0039] Specifically, the selected music is judged in terms of major and minor scale and emotional color, and then the audio production software is used to adapt and recreate the opposite emotional color (PN re-creation method, the music clips with positive emotions are recorded as positive music, if there are richer accompaniment parts, they can be recorded as PA, that is, positive accompaniment; on the contrary, the music clips and accompaniment parts with negative emotions can be recorded as NM and NA). When processing audio clips, try to select the most vivid and prominent emotional color clips in the music to establish the music library. The music library can also be enriched according to needs, and correspondingly, a more complex mapping model will be used in the parameter mapping model establishment process in the next step of the process.

[0040] S3, based on the statistical feature analysis and spectral feature analysis of the natural signal obtained in step S1 and the audio signal obtained in step S2, a reference point is selected to establish a multivariate segmented mapping model between the two aspects; (taking temperature as an example, 15°C is used as the reference point of the parameter model, ≥15°C corresponds to the spectral parameters of the PM and PA music clips, and <15°C corresponds to the spectral parameters of the NM and PA music clips)

[0041] Specifically, audio signal analysis: short-time time domain analysis and short-time frequency domain analysis are performed on the audio signal. The relevant feature parameters are extracted, and the feature parameters required are similar to the feature parameters of natural signals. In this model, we use the spectral digital features of the audio signal, which are mainly: short-time average amplitude, frequency, short-time energy, and power spectrum density.

[0042] S4. Predict and synthesize a new music clip according to the spectrum parameters of the music clip mapped by the parameter mapping model, and export and play the new music clip through an audio output device.

[0043] Specifically, the principle of audio clip synthesis is to generate music clips under current environmental conditions based on adapted music clips. In essence, it is to predict and synthesize audio signals based on the frequency domain feature data of the original audio clips. The frequency domain data of the original music clips and the characteristic parameters of the weather conditions in the natural environment are used to establish a parameter mapping model for reflection. When processing the original audio clips, the audio clips must first be preprocessed, that is, frame processing and windowing processing. On this basis, the audio clips are subjected to pitch detection to obtain pitch period information, and the frame-by-frame prediction coefficient a is obtained through linear prediction analysis. i , synthesize the music clip according to the obtained prediction coefficient α and fundamental pitch parameters through linear prediction analysis. (Fundamental pitch: general sound is composed of a series of vibrations with different frequencies and amplitudes emitted by the sound-emitting body. Among these composite vibrations, the vibration with the lowest frequency, the sound emitted by it is called fundamental pitch, and the rest are overtones. The pitch of music is determined by the fundamental pitch frequency). When superimposing data in signal synthesis, the overlap-addition method is used to overlap and connect frames of data into a continuous and smooth data stream to avoid interruption or jump of audio signals. Loudness, speed and pitch are three important factors for human perception of music, so the adaptation of music will focus on the speed and pitch change processing of audio signals and make intelligent adaptations.

[0044] The main principle of the speed change algorithm is to shorten or lengthen the original music clip without changing the fundamental frequency and formant of the music clip. The basic principle of pitch change processing of music clips is to adapt the fundamental frequency of the music clip, and at the same time, the formant is processed accordingly to achieve the purpose of pitch change. Therefore, we introduce the telescopic length ratio (i.e. the ratio of the duration of the music clip to be obtained after the speed change processing to the duration of the original music clip) and the increase and decrease ratio of the fundamental frequency (i.e., the ratio of the fundamental frequency of the music clip to be generated to the fundamental frequency of the original music clip, which is for each music clip after the frame processing, and the music clip will be synthesized in the subsequent generation process) to adapt. A mapping relationship is established between these two quantities that determine the change of music and the weather factors in the natural environment. This mapping relationship can be established with the help of an artificial intelligence neural network, or a parameter mapping relationship model can be established.

[0045] In some embodiments, Figure 2As shown, the algorithm in the present invention is applied to the actual processing process. In the figure, device 1 is a small weather station, which is used to obtain relevant characteristic signals of weather conditions, including temperature, humidity, light intensity, and wind load intensity. The original signal obtained by the small weather station is transmitted to device 3, and device 3 is a computer that can run the algorithm in the present invention. The original signal obtained in device 1 will first be pre-processed in the program of device 3, and then the spectrum analysis will be performed and the characteristic parameters will be extracted. Here, device 1 extracts the mean of the two factors of temperature and precipitation as influencing factors to adapt the music characteristics. The short-term mean of temperature is denoted as T, and the short-term mean of rainfall is denoted as P.

[0046] Figure 2 The device 2 in the process flow chart of the implementation case is a wav format file obtained by adapting and producing a given music work. Taking the music "Twinkle Twinkle Little Star" as an example, firstly, PN is rewritten according to the emotions expressed by the music, as mentioned above. Figure 3 As shown in the score of the P and N segments in the middle. It can be seen from the score that the main melody of the music segment is roughly the same. When we adapted it, we adjusted the intervals of individual bars. The resulting dissonant intervals changed the emotional style of the entire music. Other parts cooperated with the corresponding accompaniment to produce the corresponding foil effect. The finished music segment was imported into device 3. After being imported into device 3, it was processed according to the following flow chart.

[0047] Figure 2 Processing flow chart Device 3 analyzes and processes the acquired weather characteristic parameters and selects a music clip.

[0048] At the same time, the music clip is preprocessed, that is, frame processing and windowing processing. The total time before and after the adaptation of the original music clip is recorded as t and t', and the fundamental frequency of the frame-by-frame clip extracted from the music clip is recorded as f and f'.

[0049] Introducing telescopic length ratio And the fundamental frequency increase and decrease ratio A mapping relationship is established between parameters K, G and the mean temperature T and mean rainfall P obtained and processed in device 1. This mapping relationship can be established with the help of an artificial intelligence neural network, or a parameter mapping model can be established. Here, the parameter mapping model is used to construct the relationship. The details are as follows:

[0050]

[0051]

[0052]

[0053] K P With K Nis the ratio coefficient of the P segment and the N segment. Based on the analysis of the current weather conditions, these two music segments are selected for adaptation. a, b, and m in the above formula are all constants, which are determined by the auditory effect after the speed of the music segment is changed. The recommended range of K is approximately The speed of the music within this range is more suitable for the current music clip. If it exceeds this range, the beauty of the music clip will be lost.

[0054] G=c(T-20℃)+nlgP+1…………………………………Relationship model of tone shifting processing

[0055] In the above formula, c and n are both constants, which are determined by experiments and the numerical range of G. The determination of the numerical range of G needs to take into account the range of the musical sounds of certain instruments themselves, as well as the range to ensure the fullness of their timbre. And when changing, the auditory frequency range of the human ear should be considered to avoid the generation of music clips that cannot be captured by the human ear during the process of changing the generated music, forming a superficial interruption of music. For example, the violin, a musical instrument used in the above-mentioned "Twinkle Twinkle Little Star" music clip, 200Hz-300Hz affects the fullness of the violin's timbre, 6kHz-10kHz is the brightness of the timbre, and so on.

[0056] After the parameter mapping model is established, the program in device 3 follows Figure 4 The flow chart predicts and generates the music clip, and transmits it to the audio output device of device 4 to complete the generation and playback of the music clip.

[0057] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A music arrangement method based on environmental monitoring data, characterized in that: The following steps are involved: S1. By means of environmental load monitoring in structural health monitoring, the original signal of weather conditions in nature is obtained, and the original signal is preprocessed to extract characteristic parameters of weather conditions; S2, selecting and adapting music clips, performing short-time time domain analysis and short-time frequency domain analysis on the audio signal, and extracting characteristic parameters of the audio signal; S3, based on statistical feature analysis and spectrum feature analysis of the natural signal obtained in step S1 and the audio signal obtained in step S2, selecting a reference point and establishing a parameter mapping model between the two feature parameter sets; S4. Predict and synthesize a new music clip according to the spectrum parameters of the music clip mapped by the parameter mapping model, and export and play the new music clip through an audio output device.

2. The method for composing music based on environmental monitoring data according to claim 1, characterized in that: In step S1, the method of obtaining natural signals is: using the sensor equipment of a small meteorological observation station to obtain weather factors in the natural environment, thereby obtaining the original signal of the weather condition, wherein the weather factors include temperature, humidity, wind load signal, rainfall and light intensity.

3. The method for composing music based on environmental monitoring data according to claim 1, characterized in that: In the step S1, the natural signal preprocessing method is: performing noise reduction processing and frame segmentation processing on the acquired natural signal.

4. The method for composing music based on environmental monitoring data according to claim 1, characterized in that: In step S1, extracting weather condition characteristic parameters includes extracting statistical characteristics, time domain characteristic parameters and frequency domain characteristic parameters of weather signals; wherein the statistical characteristics include the mean and variance of the signal; the time domain characteristic parameters need to extract the short-time energy and average amplitude of the signal; the frequency domain characteristic parameters need to extract and obtain the short-time power spectrum of the signal.

5. The method for composing music based on environmental monitoring data according to claim 1, characterized in that: In step S2, the specific method of selecting and adapting the music clip is: judging the major and minor scale and the emotional color of the selected music, and then using audio production software to adapt and re-create the emotional color opposite to it; wherein, when processing the audio clip, the clip with the most vivid and prominent emotional color in the music is selected to establish a music library.

6. The method for composing music based on environmental monitoring data according to claim 5, characterized in that: The re-creation adopts the PN re-creation method. The music clips with positive emotions are recorded as positive music. If there are richer accompaniment parts, it is recorded as PA, that is, positive accompaniment; on the contrary, the music clips and accompaniment parts with negative emotions are recorded as NM and NA.

7. The method for composing music based on environmental monitoring data according to claim 1, characterized in that: In step S2, the characteristic parameters of the audio signal extracted include short-time average amplitude, frequency, short-time energy and power spectrum density parameters.

8. The method for composing music based on environmental monitoring data according to claim 1, characterized in that: In step S4, the audio segment is predicted and generated in the following manner: when processing the original audio segment, the audio segment is first preprocessed, i.e., frame processing and windowing processing are performed, and then the audio segment is subjected to pitch detection to obtain pitch period information, and the prediction coefficients for each frame are obtained through linear prediction analysis. , through linear prediction analysis based on the prediction coefficients already obtained The music clip is synthesized with the fundamental pitch parameters; when data is superimposed during signal synthesis, the overlap-addition method is used to overlap and connect the data frames into a continuous and smooth data stream to avoid interruption or jump of the audio signal.

Images