CN116153278A - Method, system and storage medium for predicting beating points of electric drum - Google Patents
Method, system and storage medium for predicting beating points of electric drum Download PDFInfo
- Publication number
- CN116153278A CN116153278A CN202310136962.9A CN202310136962A CN116153278A CN 116153278 A CN116153278 A CN 116153278A CN 202310136962 A CN202310136962 A CN 202310136962A CN 116153278 A CN116153278 A CN 116153278A
- Authority
- CN
- China
- Prior art keywords
- point information
- tapping point
- future
- prediction model
- predicted
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10H—ELECTROPHONIC MUSICAL INSTRUMENTS; INSTRUMENTS IN WHICH THE TONES ARE GENERATED BY ELECTROMECHANICAL MEANS OR ELECTRONIC GENERATORS, OR IN WHICH THE TONES ARE SYNTHESISED FROM A DATA STORE
- G10H3/00—Instruments in which the tones are generated by electromechanical means
- G10H3/12—Instruments in which the tones are generated by electromechanical means using mechanical resonant generators, e.g. strings or percussive instruments, the tones of which are picked up by electromechanical transducers, the electrical signals being further manipulated or amplified and subsequently converted to sound by a loudspeaker or equivalent instrument
- G10H3/125—Extracting or recognising the pitch or fundamental frequency of the picked up signal
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10H—ELECTROPHONIC MUSICAL INSTRUMENTS; INSTRUMENTS IN WHICH THE TONES ARE GENERATED BY ELECTROMECHANICAL MEANS OR ELECTRONIC GENERATORS, OR IN WHICH THE TONES ARE SYNTHESISED FROM A DATA STORE
- G10H3/00—Instruments in which the tones are generated by electromechanical means
- G10H3/12—Instruments in which the tones are generated by electromechanical means using mechanical resonant generators, e.g. strings or percussive instruments, the tones of which are picked up by electromechanical transducers, the electrical signals being further manipulated or amplified and subsequently converted to sound by a loudspeaker or equivalent instrument
- G10H3/14—Instruments in which the tones are generated by electromechanical means using mechanical resonant generators, e.g. strings or percussive instruments, the tones of which are picked up by electromechanical transducers, the electrical signals being further manipulated or amplified and subsequently converted to sound by a loudspeaker or equivalent instrument using mechanically actuated vibrators with pick-up means
- G10H3/146—Instruments in which the tones are generated by electromechanical means using mechanical resonant generators, e.g. strings or percussive instruments, the tones of which are picked up by electromechanical transducers, the electrical signals being further manipulated or amplified and subsequently converted to sound by a loudspeaker or equivalent instrument using mechanically actuated vibrators with pick-up means using a membrane, e.g. a drum; Pick-up means for vibrating surfaces, e.g. housing of an instrument
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Acoustics & Sound (AREA)
- Electrophonic Musical Instruments (AREA)
Abstract
The invention provides a method, a system and a storage medium for predicting a beating point of an electric drum, comprising the following steps: acquiring actual tapping point information at the current moment and predicted tapping point information corresponding to the current moment, judging whether the actual tapping point information is identical to the predicted tapping point information, and if so, using a first prediction model to acquire future tapping point information; if the time difference is different, acquiring a time difference between the initial time and the current time, judging whether the time difference exceeds a preset time period, and if the time difference exceeds the preset time period, acquiring future tapping point information by using a first prediction model; if the number of times is not exceeded, acquiring the number of times that the actual tapping point information is different from the predicted tapping point information, judging whether the number of times exceeds the preset number of times, and if the number of times is not exceeded, acquiring future tapping point information by using a first prediction model; if the weight of the first prediction model is exceeded, the weight of the first prediction model is adjusted to obtain a second prediction model, and the second prediction model is used for obtaining future number of future tapping point information. The method and the device can relieve the problem of delay of the electric drum.
Description
Technical Field
The present disclosure relates to the field of predicting electric drum beating points, and in particular, to a method, a system, and a storage medium for predicting electric drum beating points.
Background
The drum kit has varied rhythms and high performance skills of drummers, so that the drum kit rapidly becomes an indispensable instrument in rock music and is welcomed by vast teenagers and friends. At present, people are in a rapid life rhythm, the difficulty of hands on the drum set is relatively high, players are required to have a certain foundation and exercise for a long time, and people can hardly spend a great deal of time to finish the exercise of the drum set. With the development of society, the drum has the same effect as the drum kit because of the quick operation, so that the drum kit is quickly replaced by the drum kit. However, the working principle of the drum is different from that of the drum kit, the drum is a sensor for receiving signals, and when a drummer plays by using the drum, a certain time delay exists, so that the experience of a listener is affected.
In the related art, the electric drum has a plurality of beatable points, and a prediction model suitable for predicting the beatable points of the electric drum corresponding to a plurality of tracks is stored, and the prediction model is obtained by training according to the beatable points corresponding to the stored tracks. When a performer plays a certain track, a striking point which needs to be struck at the next moment is already obtained according to the prediction model, and the striking point is ready to be struck. That is, when a player plays a stored track, according to the prediction model, when the player beats a beat point corresponding to the current performance, the electric drum is ready to predict the beat point at the next moment, so that when the player only needs to just start to contact the beat point at the next moment, the electric drum immediately emits a sound which needs to be emitted by the beat point. However, the performer may have errors or play the non-stored tracks in the playing process, and the requirements of people may not be met only according to the existing prediction model, so that the existing problem of the delay of the electric drum cannot be solved.
Disclosure of Invention
In order to alleviate the problem of drum delay, the embodiment of the application provides a method, a system and a storage medium for predicting a knock point of a drum.
In a first aspect, the present embodiment provides a method for predicting a knock point of an electric drum, the method including:
acquiring actual beating point information at the current moment and predicted beating point information corresponding to the current moment, judging whether the actual beating point information is identical to the predicted beating point information, and if so, using a first prediction model corresponding to the current performance attribute to acquire future numerical value future beating point information which represents the future moment of a performer and needs to be beaten;
if the time difference is different, obtaining the time difference between the initial time and the current time representing the performance start, judging whether the time difference exceeds a preset time period, and if so, obtaining future numerical value future tapping point information by using the first prediction model;
if the number of times is not exceeded, acquiring the number of times that the actual tapping point information is different from the predicted tapping point information in the time difference, judging whether the number of times exceeds a preset number of times, and if the number of times does not exceed the preset number of times, acquiring future numerical value future tapping point information by using a first prediction model;
and if the number of times exceeds the preset number, adjusting the weight of the first prediction model based on the actual tapping point information and the predicted tapping point information so as to obtain a second prediction model corresponding to the replaced performance attribute, and using the second prediction model to obtain future numerical value future tapping point information.
In some embodiments, the obtaining future number of future tap point information characterizing a future time of the performer that needs to be tapped using the first predictive model corresponding to the current performance attribute includes:
acquiring a past tapping point information set corresponding to the current moment, wherein the past tapping point information set comprises the actual tapping point information and a plurality of past preset numerical value tapping point information taking the current moment as a reference;
and using the past tapping point information set as input, and replacing future numerical value future tapping point information by using a first prediction model, wherein the first prediction model is obtained by training based on the tapping point information corresponding to the stored tracks in the electric drum.
In some of these embodiments, said using the first predictive model to obtain future numerical future tap point information comprises:
and replacing the predicted tapping point information with the actual tapping point information to obtain an updated past tapping point information set, and using the past tapping point information set as input, and using a first prediction model to obtain future numerical value future tapping point information.
In some of these embodiments, the adjusting the weights of the first predictive model based on the actual tapping point information and the predicted tapping point information comprises:
and acquiring information difference between the actual beating point information and the predicted beating point information, acquiring weight difference between the weight in the first prediction model and the weight in the prediction model corresponding to the actual performance according to the information difference, and adjusting the weight of the first prediction model according to the weight difference.
In some of these embodiments, the first prediction model includes a first numerical layer weight, and obtaining, from the information difference, a weight difference between weights in the first prediction model and weights in a prediction model corresponding to an actual performance includes:
taking the current moment as a reference point, acquiring a middle tapping point information set which is obtained by subtracting a layer of weight from a previous first numerical value in a first prediction model and corresponds to the previous moment;
and obtaining a weight error of the last layer weight of the first prediction mode at the last moment according to the intermediate tapping point information set and the information difference, wherein the weight error is the weight difference.
In some of these embodiments, the value after the preset value is added by one is not less than the future value.
In some of these embodiments, determining that the actual tapping point information is not identical to the predicted tapping point information further comprises generating a cue signal indicative of performance errors of the performer.
In a second aspect, the present embodiment provides a system for predicting a knock point of an electric drum, the system including a processing module, where the processing module includes an obtaining unit, a judging unit, a first predicting unit, an adjusting unit, and a second predicting unit; wherein,,
the acquisition unit is used for acquiring actual beating point information at the current moment and predicted beating point information corresponding to the current moment;
the judging unit is used for judging whether the actual tapping point information is the same as the predicted tapping point information;
the first prediction unit is configured to obtain future number of future tapping point information that characterizes future time points of the performer by using a first prediction model corresponding to the current performance attribute when the actual tapping point information is the same as the predicted tapping point information;
the acquisition unit is further used for acquiring a time difference between the initial time and the current moment representing the performance when the actual tapping point information is different from the predicted tapping point information;
the judging unit is further used for judging whether the time difference exceeds a preset time period;
the first prediction unit is used for obtaining future numerical value future tapping point information by using the first prediction model when the time difference exceeds a preset time period;
the acquisition module is further configured to acquire the number of times that the actual tapping point information is different from the predicted tapping point information in the time difference when the time difference does not exceed a preset time period;
the judging module is also used for judging whether the times exceeds preset times or not;
the first prediction unit is further configured to obtain future number of future tapping point information by using a first prediction model if the number of times does not exceed a preset number of times;
the adjusting unit is configured to adjust the weight of the first prediction model based on the actual tapping point information and the predicted tapping point information if the number of times exceeds a preset number of times, so as to obtain a second prediction model corresponding to the replaced performance attribute;
the second prediction unit is configured to obtain future number of future tap point information using the second prediction model.
In some of these embodiments, the processing module further includes a prompting unit for generating a prompting signal representing a performance error of the performer after the actual tapping point information is different from the predicted tapping point information.
In a third aspect, embodiments of the present application provide a storage medium having stored thereon a computer program executable on a processor, the computer program implementing a method of predicting drum tapping according to the first aspect when executed by the processor.
By adopting the method, the actual beating point information and the predicted beating point information are compared to determine whether the actual beating point information is identical to the predicted beating point information, and under the identical condition, the beating point information at the future moment is obtained by using a first prediction model obtained through training according to the beating point information corresponding to the stored track; under the condition of non-uniformity, when the time difference is larger than the preset time period, the fact that the non-uniformity is caused by the play of the performer is indicated, and the first prediction model is needed to be continuously used for obtaining the tapping point information at the future moment. When the time difference is not greater than the preset time period, the relationship between the different times and the preset times is needed to be further generated according to the time difference, when the times are not greater than the predicted times, the fact that the different conditions are caused by the play of performers is indicated, and the first prediction model is needed to be continuously used for obtaining the beating point information at the future moment; when the number is larger than the predicted number, it is indicated that the non-uniformity is caused by the fact that the performer does not play the stored track, and at this time, a second prediction model is obtained on the basis of the first prediction model to obtain tapping point information at a future time. By selecting the corresponding prediction model according to different conditions, the accuracy of future beating point information can be improved, so that the corresponding beating point is ready, and when a performer just begins to contact the beating point, the electric drum immediately emits sound required to be emitted by the beating point, thereby being beneficial to alleviating the problem of delay of the electric drum.
Drawings
Fig. 1 is a schematic diagram of the working principle of the electric drum according to the present embodiment.
Fig. 2 is a flowchart of a method for predicting a knock point of an electric drum according to the present embodiment.
Fig. 3 is a flowchart of a method for adjusting weights of a first prediction model according to the present embodiment.
Fig. 4 is a block diagram of a predictive electric drum striking point system according to the present embodiment.
Detailed Description
For a clearer understanding of the objects, technical solutions and advantages of the present application, the present application is described and illustrated below with reference to the accompanying drawings and examples. However, it will be apparent to one of ordinary skill in the art that the present application may be practiced without these details. It will be apparent to those having ordinary skill in the art that various changes can be made to the embodiments disclosed herein and that the general principles defined herein may be applied to other embodiments and applications without departing from the principles and scope of the present application. Thus, the present application is not limited to the embodiments shown, but is to be accorded the widest scope consistent with the scope claimed herein.
Embodiments of the present application are described in further detail below with reference to the drawings attached hereto.
The electric drum is a sensor for receiving signals, and when a drum stick is knocked on a drum plate receiver, the sensor is transmitted to a drum sound source machine which is specially used for storing drum sounds by using a wire, and then the read and output tone is sent to a sound box, a loudspeaker for sounding and is heard. The electric drum comprises a hi-hat, a hanging cymbal, three dings, a military drum, a bottom drum, a pedal and a drum stick. Each cymbal and each drum are provided with a corresponding striking point, the position of the striking point is the position of the trigger, and when the drum stick strikes each striking point, the electric drum can generate corresponding sound. Fig. 1 is a schematic diagram of the working principle of the electric drum according to the present embodiment. As shown in fig. 1, after a drum stick strikes a drum or a small cymbal, a sensor in a trigger at the striking point generates a voltage signal, and transmits the voltage signal to a sound source of the electric drum through a data line and converts the voltage signal into a digital MIDI signal, and the sound source plays a prefabricated sound sampling file according to striking point information carried by the MIDI signal and force information corresponding to the striking point, so that people can hear the sound of the electric drum.
Fig. 2 is a flowchart of a method for predicting a knock point of an electric drum according to the present embodiment. As shown in fig. 2, a method for predicting a knock point of an electric drum includes the steps of:
step S100, acquiring actual tapping point information at the current time and predicted tapping point information corresponding to the current time, and determining whether the actual tapping point information is identical to the predicted tapping point information.
Each tapping point of the electric drum is correspondingly provided with a specific and unique number, the actual tapping point information represents the tapping point number and the tapping point strength of the tapping at the current moment, and the actual tapping point information comprises the actual tapping point number and the actual tapping point strength; the predicted tapping information characterizes a tapping number and a tapping force required to be tapped at a current time predicted at a past time, and includes the predicted tapping number and the predicted tapping force. The past time may be the time immediately before the current time or may be several times longer than the current time. In order to obtain more accurate predicted tapping point information, the predicted tapping point information in this embodiment characterizes tapping point information that needs to be tapped at the current time predicted at the previous time.
Because the same prediction model is used for the tapping point number and the tapping force, when the tapping point number can be accurately predicted, the tapping force can be accurately predicted certainly, namely when the actual tapping point number is the same as the predicted tapping point number, the actual tapping point information is the same as the predicted tapping point information. Therefore, it is only necessary to determine whether or not the actual tapping point number is the same as the predicted tapping point number, and it can be determined whether or not the actual tapping point information is the same as the predicted tapping point information.
A shooting device is arranged beside the electric drum, a pressure sensing device is arranged at the hand-held position of a drum rod of the electric drum, and the shooting device and the pressure sensing device are connected with a processing module. When the drum stick is placed on the electric drum, the pressure sensing device can only sense extremely small force, namely the force given by the drum head, the preset pressure value set by the pressure sensing device is not reached at the moment, the pressure sensing device cannot send out pressure sensing signals, namely the pressure sensing device cannot send out pressure sensing signals to the processing module, and the processing module cannot send shooting signals to the shooting device, so that the shooting device is in a sleep state continuously.
When the performer picks up the drum stick, the hand of the performer can give certain force to the drum stick, the pressure sensing device can feel certain force, the preset pressure value set by the pressure sensing device can be reached, the pressure sensing device can send a pressure sensing signal to the processing module, the processing module receives the pressure sensing signal and sends a shooting signal to the shooting device, the shooting device starts shooting work to obtain an image representing that the performer knocks the electric drum, the image is sent to the image processing device, and the image processing device is connected with the shooting device and the processing model. The pressure sensing device is provided with a pressure preset value when sending a pressure sensing signal, so that other objects can be prevented from touching the pressure sensing device and mistakenly sending the pressure sensing signal to the processing module, and the condition that the shooting device does useless shooting is reduced.
And after the image processing device receives the image, extracting the characteristics of the image to obtain the position coordinates of the drum stick in the image, and sending the position coordinates to the processing module. The processing module stores coordinate areas corresponding to the tapping points respectively, and the obtained coordinate positions are compared with each coordinate area. When the coordinate area contains the coordinate position, indicating that the coordinate position belongs to the coordinate area; when the coordinate region does not contain the coordinate position, it is indicated that the coordinate position does not belong to the coordinate region. After the coordinate area to which the coordinate position belongs is determined, determining the tapping point corresponding to the coordinate area, and determining the tapping point number corresponding to the tapping point according to the tapping point, thereby obtaining the actual tapping point information at the current moment. In addition, the drum is a sensor for receiving signals, and therefore, the drum can obtain the striking strength of each striking point by itself.
The predicted tapping point information is obtained from a prediction model at a past time, and the number and the strength corresponding to the tapping points at the current time are obtained from the prediction model by taking the number and the strength corresponding to the number of the tapping points which are closest to the current time and have occurred and are the same as the number of the tapping points as the input information of the prediction model. By comparing the number corresponding to the actual tapping point with the number corresponding to the predicted tapping point, the judgment of whether the actual tapping point information is identical to the predicted tapping point information can be completed. When the number corresponding to the actual beating point is the same as the number corresponding to the predicted beating point, indicating that the actual beating point information is the same as the predicted beating point information; when the number corresponding to the actual tapping point is different from the number corresponding to the predicted tapping point, the actual tapping point information is indicated to be different from the predicted tapping point information.
In step S200, if the actual tapping point information is the same as the predicted tapping point information, the future number of future tapping point information, which represents the future time of the performer and needs to be tapped, is obtained by using the first prediction model corresponding to the current performance attribute.
Since there is no tapping point information at the past time when the performer just starts playing, there is no prediction of several tapping point information that the performer just needs to perform. When the performer has tapped the same number of input information corresponding to the predictive model, a number of tap point information which is closest to the present time and has occurred can be obtained, and the tap point information is used as input of the predictive model to predict the tap point information at the future time.
Using a first predictive model corresponding to the current performance attribute to obtain future number of future tap point information characterizing a future time of the performer that is to be tapped includes: acquiring a past tapping point information set corresponding to the current moment, wherein the past tapping point information set comprises actual tapping point information and past preset numerical value tapping point information taking the current moment as a reference; and predicting future number of future tapping point information by using a first prediction model with a past tapping point information set as an input, wherein the first prediction model is trained based on the tapping point information corresponding to the stored tracks in the electric drum.
When a performer plays the tracks, the performer defaults that the stored tracks in the electric drum are selected for playing, namely, the current performance attribute corresponds to a stored track. Wherein the stored tracks all have respective tapping point information, and a deep learning model can be used to predict the tapping point information at a future time. In this embodiment, the first prediction model may use a BP neural network to divide the dynamics into different dynamics levels, where each dynamics level corresponds to a numerical value, so each piece of tapping point information may be equivalently a two-dimensional array, two elements in the two-dimensional array are a tapping point number and a tapping point dynamics, and the tapping point information of all the stored tracks is trained using the BP neural network.
Wherein the BP neural network is trained using the hit point information of the stored track to obtain a first predictive model. The input of the first prediction model is a past tapping point information set corresponding to the current time, the past tapping point information set including tapping point information of the current time and predicted numerical value tapping point information with the current time as a reference past time direction. The number of input tap point information of the first predictive model is a preset number plus one. The output of the first predictive model is tapping point information that needs to be tapped at a future time, and the number of the tapping point information is a future value.
In order to better predict the tapping point information at the future time, the value after the preset value is added by one should not be smaller than the future value. I.e. a smaller number of non-occurring tap point information is predicted using a larger number of already occurring tap point information. When the future value is not less than two, the first prediction model can predict the tapping point information at a future more distant moment, the tapping point information can provide reference for the tapping point possibly used later, and when the time of the processing model responding to different requests is longer than the time of two adjacent tapping, the future value can be set to be not less than two so as to reserve more preparation time for the processing device. When the future value is one, it indicates that the first predictive model predicts only tapping point information at the next moment. Because the predicted tapping point information corresponding to the moment closer to the current moment is more accurate, the corresponding tapping point is ready, and the delay of the electric drum is relieved, when the time of the processing module responding to different requests is not longer than the time of two adjacent tapping, the future value can be set to be one, so that the accuracy of the predicted tapping point information is improved. The present embodiment is based on prediction accuracy, where the future value is set to one.
Each stored track corresponds to a plurality of tapping point information according to a certain sequence, the tapping point information in each stored track is taken as a reference point at each moment, the tapping point information corresponding to the preset value added with one value and the future value is taken as the input quantity and the output quantity of the first prediction model, so that a plurality of groups of input quantity and output quantity corresponding to each stored track are obtained, and each input quantity and each output quantity are in one-to-one correspondence. And (3) arbitrarily selecting 70% of input quantity and output quantity from a plurality of groups of input quantity and output quantity corresponding to all stored tracks to train the BP neural network model so as to obtain a first prediction model. After each time the performer finishes the tapping operation of the tapping point, the processing module stores the tapping information of the time. Therefore, by acquiring the past tapping point information set corresponding to the current time from the processing model and using the past tapping point information set as the input quantity of the first prediction model, the future number of future tapping point information which characterizes the future time of the performer and needs to be tapped can be obtained.
Step S300, if the actual tapping point information is different from the predicted tapping point information, obtaining a time difference between the initial time and the current time representing the performance start, judging whether the time difference exceeds a preset time period, and if so, obtaining future numerical values of future tapping point information by using a first prediction model.
The processing module records each moment in time during which the performer plays the track. The initial time corresponds to the moment when the performer starts to play a track, and the time difference can be obtained by subtracting the initial time from the current time, namely the time period when the performer corresponding to the current moment plays the track. The above-mentioned preset time period is used to distinguish whether the performer plays the stored track, for example, the track actually played by the performer does not belong to the stored track, and the tapping point information of the track in the period not exceeding the time period is identical to the stored tapping point information, but there are no identical two tracks, and it is possible to distinguish whether the performer is playing the stored track in the preset time period. If the actual tapping point information is different from the predicted tapping point information, and if the period of time is exceeded, the actual tapping point information defaults to the error of the performer, so that the performer still needs to continue playing the current song, and therefore, the tapping point information at the future moment should be predicted by using the first prediction model continuously. The preset time period is obtained through big data analysis.
Wherein predicting future values of future tap point information using the first predictive model includes: the predicted tap point information is replaced with actual tap point information to obtain an updated set of past tap point information, and a first predictive model is used to obtain future numerical future tap point information with the set of past tap point information as input.
The actual tapping information is characterized by the tapping information in the past tapping information set, and the actual tapping information is different from the predicted tapping information due to errors of the performer, so that the predicted tapping information is correct and the actual tapping information is wrong. Therefore, when future tapping point information corresponding to the next time is predicted using the past tapping point information set at the present time, the actual tapping point information in which an error occurs should be replaced with the predicted tapping point information, thereby obtaining a correct past tapping point information set, and the newly obtained past tapping point information set is input into the first prediction model to predict the future tapping point information corresponding to the next time. That is, the use of the correct past set of tapping point information to predict future tapping point information may improve the accuracy of the future tapping point information as compared to the use of a past set of tapping point information containing incorrect tapping point information, thereby making the corresponding tapping point ready, so that the electric drum immediately emits the sound that the tapping point needs to emit when the performer just begins to touch the tapping point, which is advantageous for alleviating the problem of delay of the electric drum.
At other time points in the future, as long as the past point information set contains the actual point information of the player playing the wrong, the correct predicted point information should be used to replace the actual point information of the wrong occurrence, so as to obtain a new past point information set, and the past point information set is used as input to predict the future point information by using the first prediction model.
In addition, generating a prompt signal representing performance errors of the performer is further included after judging that the actual striking point information is different from the predicted striking point information.
If the actual beating point information is different from the predicted beating point information, the processing module defaults to be a performance error of the performer, and at the moment, the processing module generates a prompting signal and displays the prompting information of the performance error in a display screen of the electric drum so as to remind the performer that the performance error possibly exists.
In step S400, if the time difference does not exceed the preset time period, the number of times that the actual tapping point information is different from the predicted tapping point information in the time difference is obtained, and whether the number of times exceeds the preset number of times is determined, if not, a first prediction model is used to obtain future numerical value future tapping point information.
Even if most of the tracks on the market are stored in the electric drum, when the tracks are updated in real time with the progress of society, the electric drum cannot store all the tracks on the market. When the time difference does not exceed the preset time period, the performer may make a mistake to cause a play error, and it is also possible that the performer plays a track that is not stored. It is also necessary to acquire the number of times that the actual tapping point information is different from the predicted tapping point information within the time difference. Because in case the actual tapping point information is inconsistent with the predicted tapping point information, the display screen can display prompt information to remind the performer of the situation that performance errors possibly exist, so that the performer can adjust in time under the situation that performance errors exist, and the situation that errors continuously occur in the follow-up process is reduced. Therefore, if the number of times does not exceed the preset number of times, it indicates that the performer is only playing the stored track for some number of striking points to be wrong. And at the moment, the first prediction model is continuously used for predicting future tapping point information corresponding to the subsequent moment respectively.
And step S500, if the number of times exceeds the predicted number of times, adjusting the weight of the first prediction model based on the actual tapping point information and the predicted tapping point information to obtain a second prediction model corresponding to the replaced performance attribute, and using the second prediction model to obtain future numerical value future tapping point information.
Because the display screen is provided with the prompt information to remind the player, if the number of times exceeds the predicted number of times, the player is shown that the played track is not the stored track, and the model is required to be adjusted on the basis of the first prediction model at the moment so as to obtain a new prediction model to better predict the currently played track. Since training a new model takes a lot of time, and playing is focused on real-time, retraining a new predictive model every time an unrecorded track is played can also take a lot of effort. Retraining a new predictive model to predict the current track is not practical. However, the first predictive model is trained from several tracks, the first predictive model having the basic function of predicting future tapping point information from a set of past tapping point information for different tracks. The weight in the first prediction model is adjusted on line to obtain the second prediction model, so that the time is less, the real-time performance can be met, the prediction model which is more in line with the current track can be obtained according to the actual situation, compared with the use of the first prediction model, the future beating point information obtained through the second prediction model is more accurate, the corresponding beating point is ready, and therefore, when a performer just begins to contact the beating point, the electric drum immediately emits sound which needs to be emitted by the beating point, and the problem of delay of the electric drum is solved. Wherein the first prediction model and the second prediction model both include a first numerical layer weight.
Fig. 3 is a flowchart of a method for adjusting weights of a first prediction model according to the present embodiment. As shown in fig. 3, adjusting the weights of the first predictive model based on the actual tapping point information and the predicted tapping point information includes the steps of:
step S501, an information difference between the actual tapping information and the predicted tapping information is acquired.
Step S502, taking the current moment as a reference point, obtaining a middle tapping point information set which is obtained by subtracting a layer of weight from a previous first numerical value in the first prediction model and corresponds to the previous moment.
Step S503, obtaining a weight error of the last layer weight of the first prediction model at the last moment according to the intermediate tapping point information set and the information difference, wherein the weight error is the weight difference.
Step S504, the weight of the first prediction model is adjusted according to the weight difference.
And considering timeliness, only the weight of the last layer in the first prediction model is adjusted on line, so that a second prediction model is obtained. The information difference between the actual information and the predicted information corresponding to the current moment is obtained by subtracting the predicted tapping point information from the actual tapping point information, wherein if the information difference corresponding to other moments is related in the past tapping point information set, the information difference corresponding to each moment is obtained by subtracting the predicted tapping point information from the actual tapping point information, and the information differences are combined into a past tapping point information difference set which can be regarded as a matrix.
After the trained first prediction model is obtained, the weight value among all layers in the first prediction model can be obtained, so that the current moment is taken as a reference point, and after the weight distribution of the first numerical value minus one layer before the previous moment in the first prediction model is carried out on the past tapping point information set corresponding to the previous moment, the middle tapping point information set input to the last layer can be obtained, and the middle tapping point information set can be regarded as a matrix. The method comprises the steps of multiplying the violations of a middle tapping point information set by a past tapping point information difference set to obtain a weight error of the last layer of weight of a first prediction model at the last moment, taking the weight error as the weight difference of the last layer of weight of the current moment, and adding the weight difference to the weight value of the last layer of the original first prediction model to obtain the weight value of the last layer of the second prediction model, wherein the weight value of the first layer of weight subtracted from the previous first value of the second prediction model is the same as the weight value of the first layer of weight subtracted from the previous first value of the first prediction model. After the second predictive model is obtained, the future number of future tap point information is then predicted using the second predictive model.
Fig. 4 is a block diagram of a predictive electric drum striking point system according to the present embodiment. As shown in fig. 4, a frame diagram of a predictive drum striking point system includes a processing module including an acquisition unit, a judgment unit, a first prediction unit, an adjustment unit, and a second prediction unit.
And the acquisition unit is used for acquiring the actual tapping point information at the current moment and the predicted tapping point information corresponding to the current moment. And a judging unit for judging whether the actual tapping point information is identical to the predicted tapping point information. And the first prediction unit is used for obtaining future numerical value future tapping point information which represents the future moment of the performer and needs to be tapped by using a first prediction model corresponding to the current performance attribute when the actual tapping point information is the same as the predicted tapping point information. And the acquisition unit is also used for acquiring the time difference between the initial time and the current moment when the representation performance starts when the actual tapping point information is different from the predicted tapping point information. And the judging unit is also used for judging whether the time difference exceeds a preset time period. And the first prediction unit is used for obtaining future numerical value future tapping point information by using the first prediction model when the time difference exceeds a preset time period. The acquisition module is further used for acquiring the times of different actual tapping point information and predicted tapping point information in the time difference when the time difference does not exceed the preset time period. The judging module is also used for judging whether the times exceed the preset times. The first prediction unit is further configured to obtain future number of future tapping point information by using the first prediction model if the number of times does not exceed the preset number of times. And the adjusting unit is used for adjusting the weight of the first prediction model based on the actual tapping point information and the predicted tapping point information if the number of times exceeds the preset number of times so as to obtain a second prediction model corresponding to the replaced performance attribute. And a second prediction unit for obtaining future number of future tap point information using the second prediction model.
The processing module further comprises a prompting unit which is used for generating a prompting signal representing performance errors of the performer after the actual striking point information is different from the predicted striking point information.
The present application provides a computer readable storage medium having a computer program stored thereon, which when run on a computer, causes the computer to perform the relevant content of the foregoing method embodiments.
It should be understood that, although the steps in the flowcharts of the figures are shown in order as indicated by the arrows, these steps are not necessarily performed in order as indicated by the arrows. The steps are not strictly limited in order and may be performed in other orders, unless explicitly stated herein.
The foregoing is only a partial embodiment of the present application, and it should be noted that, for a person skilled in the art, several improvements and modifications can be made without departing from the principle of the present application, and these improvements and modifications should also be considered as the protection scope of the present application.
Claims (10)
1. A method of predicting a knock point of an electric drum, the method comprising:
acquiring actual beating point information at the current moment and predicted beating point information corresponding to the current moment, judging whether the actual beating point information is identical to the predicted beating point information, and if so, using a first prediction model corresponding to the current performance attribute to acquire future numerical value future beating point information which represents the future moment of a performer and needs to be beaten;
if the time difference is different, obtaining the time difference between the initial time and the current time representing the performance start, judging whether the time difference exceeds a preset time period, and if so, obtaining future numerical value future tapping point information by using the first prediction model;
if the number of times is not exceeded, acquiring the number of times that the actual tapping point information is different from the predicted tapping point information in the time difference, judging whether the number of times exceeds a preset number of times, and if the number of times does not exceed the preset number of times, acquiring future numerical value future tapping point information by using a first prediction model;
and if the number of times exceeds the preset number, adjusting the weight of the first prediction model based on the actual tapping point information and the predicted tapping point information so as to obtain a second prediction model corresponding to the replaced performance attribute, and using the second prediction model to obtain future numerical value future tapping point information.
2. The method of claim 1, wherein using the first predictive model corresponding to the current performance attribute to obtain future number of future tap point information indicative of a future time of a performer that needs to be tapped comprises:
acquiring a past tapping point information set corresponding to the current moment, wherein the past tapping point information set comprises the actual tapping point information and a plurality of past preset numerical value tapping point information taking the current moment as a reference;
and taking the past tapping point information set as input, and using a first prediction model to obtain future numerical value future tapping point information, wherein the first prediction model is obtained by training based on the tapping point information corresponding to the stored tracks in the electric drum.
3. The method of claim 2, wherein the using the first predictive model to obtain future values of future tap point information comprises:
and replacing the predicted tapping point information with the actual tapping point information to obtain an updated past tapping point information set, and using the past tapping point information set as input, and using a first prediction model to obtain future numerical value future tapping point information.
4. The method of claim 1, wherein the adjusting weights of the first predictive model based on the actual tapping point information and the predicted tapping point information comprises:
and acquiring information difference between the actual beating point information and the predicted beating point information, acquiring weight difference between the weight in the first prediction model and the weight in the prediction model corresponding to the actual performance according to the information difference, and adjusting the weight of the first prediction model according to the weight difference.
5. The method of claim 4, wherein the first predictive model includes a first numerical layer weight, wherein obtaining a weight difference between weights in the first predictive model and weights in a predictive model corresponding to an actual performance from the information difference includes:
taking the current moment as a reference point, acquiring a middle tapping point information set which is obtained by subtracting a layer of weight from a previous first numerical value in a first prediction model and corresponds to the previous moment;
and obtaining a weight error of the last layer weight of the first prediction model at the last moment according to the intermediate tapping point information set and the information difference, wherein the weight error is the weight difference.
6. The method of claim 2, wherein the value after the preset value is added by one is not less than the future value.
7. The method of claim 1, wherein determining that the actual tapping point information is not the same as the predicted tapping point information further comprises generating a cue signal indicative of a performance error of the performer.
8. The system for predicting the beating points of the electric drum is characterized by comprising a processing module, wherein the processing module comprises an acquisition unit, a judging unit, a first predicting unit, an adjusting unit and a second predicting unit; wherein,,
the acquisition unit is used for acquiring actual beating point information at the current moment and predicted beating point information corresponding to the current moment;
the judging unit is used for judging whether the actual tapping point information is the same as the predicted tapping point information;
the first prediction unit is configured to obtain future number of future tapping point information that characterizes future time points of the performer by using a first prediction model corresponding to the current performance attribute when the actual tapping point information is the same as the predicted tapping point information;
the acquisition unit is further used for acquiring a time difference between the initial time and the current moment representing the performance when the actual tapping point information is different from the predicted tapping point information;
the judging unit is further used for judging whether the time difference exceeds a preset time period;
the first prediction unit is used for obtaining future numerical value future tapping point information by using the first prediction model when the time difference exceeds a preset time period;
the acquisition module is further configured to acquire the number of times that the actual tapping point information is different from the predicted tapping point information in the time difference when the time difference does not exceed a preset time period;
the judging module is also used for judging whether the times exceeds preset times or not;
the first prediction unit is further configured to obtain future number of future tapping point information by using a first prediction model if the number of times does not exceed a preset number of times;
the adjusting unit is configured to adjust the weight of the first prediction model based on the actual tapping point information and the predicted tapping point information if the number of times exceeds a preset number of times, so as to obtain a second prediction model corresponding to the replaced performance attribute;
the second prediction unit is configured to obtain future number of future tap point information using the second prediction model.
9. The system of claim 8, wherein the processing module further comprises a prompting unit for generating a prompting signal indicative of a performance error of the performer after the actual tapping point information is different from the predicted tapping point information.
10. A computer readable storage medium having stored thereon a computer program executable on a processor, wherein the computer program when executed by the processor implements a method of predicting electric drum tapping as claimed in any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310136962.9A CN116153278A (en) | 2023-02-20 | 2023-02-20 | Method, system and storage medium for predicting beating points of electric drum |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310136962.9A CN116153278A (en) | 2023-02-20 | 2023-02-20 | Method, system and storage medium for predicting beating points of electric drum |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116153278A true CN116153278A (en) | 2023-05-23 |
Family
ID=86359751
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310136962.9A Pending CN116153278A (en) | 2023-02-20 | 2023-02-20 | Method, system and storage medium for predicting beating points of electric drum |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116153278A (en) |
-
2023
- 2023-02-20 CN CN202310136962.9A patent/CN116153278A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9773480B2 (en) | Electronic music controller using inertial navigation-2 | |
| CN111052223B (en) | Playback control method, playback control device and recording medium | |
| US7633003B2 (en) | Performance control apparatus and program therefor | |
| JPWO2018168953A1 (en) | Signal supply device, keyboard device, and program | |
| US11120780B2 (en) | Emulation of at least one sound of a drum-type percussion instrument | |
| JP2015045829A (en) | Touch detection device, touch detection method, and program | |
| US8785759B2 (en) | Electric keyboard musical instrument, method executed by the same, and storage medium | |
| WO2007135791A1 (en) | Data computing unit for music game, data computing program for music game, and data computing method for music game | |
| JP3509545B2 (en) | Performance information evaluation device, performance information evaluation method, and recording medium | |
| CN116153278A (en) | Method, system and storage medium for predicting beating points of electric drum | |
| US10805475B2 (en) | Resonance sound signal generation device, resonance sound signal generation method, non-transitory computer readable medium storing resonance sound signal generation program and electronic musical apparatus | |
| US7838754B2 (en) | Performance system, controller used therefor, and program | |
| JP2525853B2 (en) | Continuous hit processing device for electronic musical instruments | |
| JP4628725B2 (en) | Tempo information output device, tempo information output method, computer program for tempo information output, touch information output device, touch information output method, and computer program for touch information output | |
| JP3624780B2 (en) | Music control device | |
| JP6507519B2 (en) | Touch detection device, method, and program, electronic musical instrument | |
| JP4046226B2 (en) | Electronic piano | |
| JP6572916B2 (en) | Pronunciation control device and method, program | |
| JP3624773B2 (en) | Reading practice device, reading practice method, and computer-readable recording medium recording a program for realizing the method | |
| CN116830179A (en) | Information processing system, electronic musical instrument, information processing method and machine learning system | |
| WO2023080080A1 (en) | Performance analysis method, performance analysis system, and program | |
| JP3953071B2 (en) | Electronic musical instruments | |
| JP5560581B2 (en) | Electronic musical instruments | |
| CN116189635A (en) | High-precision low-delay response method for electric drum and storage medium | |
| JP3879759B2 (en) | Electronic musical instruments |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |