WO2019196301A1 - 电子装置、基于深度学习的乐谱识别方法、系统及存储介质 - Google Patents

电子装置、基于深度学习的乐谱识别方法、系统及存储介质 Download PDF

Info

Publication number
WO2019196301A1
WO2019196301A1 PCT/CN2018/102113 CN2018102113W WO2019196301A1 WO 2019196301 A1 WO2019196301 A1 WO 2019196301A1 CN 2018102113 W CN2018102113 W CN 2018102113W WO 2019196301 A1 WO2019196301 A1 WO 2019196301A1
Authority
WO
WIPO (PCT)
Prior art keywords
music
model
score
recognition
pitch
Prior art date
Application number
PCT/CN2018/102113
Other languages
English (en)
French (fr)
Inventor
刘奡智
王健宗
肖京
Original Assignee
平安科技(深圳)有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 平安科技(深圳)有限公司 filed Critical 平安科技(深圳)有限公司
Publication of WO2019196301A1 publication Critical patent/WO2019196301A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12Classification; Matching
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/08Learning methods
    • G06N3/088Non-supervised learning, e.g. competitive learning
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/21Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/04Architecture, e.g. interconnection topology
    • G06N3/044Recurrent networks, e.g. Hopfield networks
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/04Architecture, e.g. interconnection topology
    • G06N3/045Combinations of networks
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/04Architecture, e.g. interconnection topology
    • G06N3/047Probabilistic or stochastic networks
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/10Character recognition
    • G06V30/30Character recognition based on the type of data
    • G06V30/304Music notations
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/40Document-oriented image-based pattern recognition

Definitions

  • the present application relates to the field of deep learning, and in particular, to an electronic device, a music learning method based on deep learning, a system, and a storage medium.
  • the present application provides an electronic device, a depth learning-based music score recognition method, and a storage medium, which can accurately identify the quality of a musical piece, and the method is simple, flexible, and practical.
  • the present application provides an electronic device including a memory and a processor coupled to the memory, the processor for performing deep learning-based music score recognition stored on the memory
  • the program when the depth learning based music score recognition program is executed by the processor, implements the following steps:
  • the scores of the music intensity are analyzed, and whether the scores marked with the music strength meet the predefined music standards are determined;
  • the score of the music to be discriminated is qualified, or, if not, the score of the music to be discriminated is determined to be unqualified.
  • the present application further provides a music learning method based on deep learning, the method comprising the following steps:
  • the scores of the music intensity are analyzed, and whether the scores marked with the music strength meet the predefined music standards are determined;
  • the score of the music to be discriminated is qualified, or, if not, the score of the music to be discriminated is determined to be unqualified.
  • the present application further provides a music learning system based on deep learning, the system comprising an acquisition module, an identification module, an analysis module, and a determination module;
  • the acquiring module is configured to obtain a music element in a musical score of a music quality to be discriminated, and preprocess the acquired music element to generate a corresponding music feature matrix;
  • the identification module is configured to substitute the music feature matrix into a predetermined music strength annotation model for recognition, and output a music score marked with music strength;
  • the analysis module is configured to analyze a music score marked with music strength according to a predetermined music recognition model, and determine whether the music score marked with the music strength meets a predefined music standard;
  • the determining module is configured to determine that the score of the music to be discriminated is qualified if the score of the standard musical intensity is determined to meet the predefined music standard, or, if not, determine that the score of the music to be discriminated is unqualified.
  • the present application further provides a computer readable storage medium storing a depth learning based music score recognition program, the depth learning based music score recognition program being at least one processed Executing to cause the at least one processor to perform the following steps:
  • the scores of the music intensity are analyzed, and whether the scores marked with the music strength meet the predefined music standards are determined;
  • the score of the music to be discriminated is qualified, or, if not, the score of the music to be discriminated is determined to be unqualified.
  • the electronic device, the deep learning-based music score recognition method, the system and the storage medium proposed by the present application preprocess the acquired music elements by acquiring music elements in the musical score of the music quality to be discriminated. Generating a corresponding music feature matrix; substituting the music feature matrix into a predetermined music velocity annotation model for recognition, outputting a music score marked with music strength; analyzing a music score marked with music strength according to a predetermined music recognition model, and determining the label Whether the score of the musical intensity conforms to the predefined music standard; if it is met, it is determined that the score of the music to be discriminated is qualified, or, if not, the score of the music to be discriminated is determined to be unqualified.
  • the quality of the musical piece can be accurately identified, and the method is simple, flexible and practical.
  • FIG. 1 is a schematic diagram of an optional hardware architecture of an electronic device proposed by the present application.
  • FIG. 2 is a schematic diagram of a program module of a music score recognition program based on deep learning in an embodiment of an electronic device of the present application
  • FIG. 3 is a flow chart of an implementation of a preferred embodiment of a music score recognition method based on deep learning in the present application.
  • the electronic device 10 may include, but is not limited to, the memory 11, the processor 12, and the network interface 13 being communicably connected to each other through the communication bus 14. It should be noted that FIG. 1 only shows the electronic device 10 having the components 11-14, but it should be understood that not all illustrated components may be implemented, and more or fewer components may be implemented instead.
  • the memory 11 includes at least one type of computer readable storage medium including a flash memory, a hard disk, a multimedia card, a card type memory (for example, SD or DX memory, etc.), a random access memory (RAM), and a static memory.
  • the memory 11 may be an internal storage unit of the electronic device 10, such as a hard disk or memory of the electronic device 10.
  • the memory 11 may also be an outsourced storage device of the electronic device 10, such as a plug-in hard disk equipped on the electronic device 10, a smart memory card (SMC), and a secure digital (Secure Digital, SD) ) cards, flash cards, etc.
  • the memory 11 can also include both an internal storage unit of the electronic device 10 and an outsourced storage device thereof.
  • the memory 11 is generally used to store an operating system installed in the electronic device 10 and various types of application software, such as a music recognition program based on deep learning. Further, the memory 11 can also be used to temporarily store various types of data that have been output or are to be output.
  • Processor 12 may be a Central Processing Unit (CPU), controller, microcontroller, microprocessor, or other data processing chip in some embodiments.
  • the processor 12 is typically used to control the overall operation of the electronic device 10.
  • the processor 12 is configured to run program code or processing data stored in the memory 11, such as a running deep learning-based music score recognition program or the like.
  • the network interface 13 may include a wireless network interface or a wired network interface, which is typically used to establish a communication connection between the electronic device 10 and other electronic devices.
  • Communication bus 14 is used to implement a communication connection between components 11-13.
  • Figure 1 shows only the electronic device 10 with components 11-14 and a deep learning based score recognition program, but it should be understood that not all illustrated components may be implemented, alternative implementations may be more or less Component.
  • the electronic device 10 may further include a user interface (not shown in FIG. 1), and the user interface may include a display, an input unit such as a keyboard, wherein the user interface may further include a standard wired interface, a wireless interface, and the like.
  • the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED touch device, or the like. Further, the display may also be referred to as a display screen or display unit for displaying information processed in the electronic device 10 and a user interface for displaying visualizations.
  • the scores of the music intensity are analyzed, and whether the scores marked with the music strength meet the predefined music standards are determined;
  • the score of the music to be discriminated is qualified, or, if not, the score of the music to be discriminated is determined to be unqualified.
  • the music element is pitch and music intensity
  • the step of pre-processing the acquired music element to generate a corresponding music feature matrix comprises: obtaining the pitch and a predefined vibration frequency value (Predefined with 128 pitches, each pitch has three representations) to match, matching the vibration frequency values corresponding to each pitch;
  • the vibration frequency value after matching is identified by a predefined pitch identification method (for example, the predefined pitch identification mode is, C1, 0, 0 means that there is no sound in the C1 frequency segment, and C1, 0, 1 indicates that the frequency is in the C1 frequency range. Short tone, C1,1,1 indicates the extension of the C1 frequency band);
  • a two-dimensional matrix is generated according to the obtained vibration frequency value and the number of acquired pitches, wherein one dimension of the two-dimensional matrix represents the number of pitches and the identifier of the pitch, and the other dimension represents a predefined time interval.
  • the music strength annotation model and the music recognition model are pre-trained generated confrontation networks (GAN), and the GAN network includes a Generative Model and a Discriminative Model.
  • the generation model is used to annotate the musical strength
  • the discriminant model is used to identify whether the musical score conforms to the music standard.
  • the Generative Model is a convolution-based neural network (CNN), and the discriminant model is a recognition model obtained by training the convolution based neural network;
  • the Generative Model is an LSTM long and short memory neural network
  • the discriminant model is a recognition model obtained based on the LSTM long and short memory neural network training
  • the recognition model generally outputs a value of a probability function value.
  • the probability function value conforms to a normal probability distribution, and the recognition result conforms to a preset criterion. If the probability function value does not conform to the normal probability distribution, the recognition result is represented. Does not meet the preset criteria.
  • the generation of music is taken as an example to illustrate the principle of GAN.
  • the generation model is a music intensity annotation network, which receives a random sound Z, and the sound intensity is marked by this sound, and is recorded as G(Z).
  • the recognition model is a discriminant network that discriminates whether the intensity of the annotated music is "consistent with the performance scene.” Its input parameter is X, X represents a music with the strength of the music, and the output D(X) represents the probability that X is the strength of the music in accordance with the performance scene. If it is 1, it means that 100% is the real music that matches the performance scene.
  • the strength of the label, and the output is 0, it means that it is impossible to be a true annotation of the musical intensity of the performance scene.
  • the goal of generating the network is to generate a true mark of the music that matches the performance scene to deceive the discriminant network.
  • the goal of discriminating the network is to separate the music generated by the generated network and label the music strength.
  • the generation network and the discriminant network constitute a dynamic "game process".
  • the electronic device proposed by the present application preprocesses the acquired music elements by acquiring music elements in the musical score of the music quality to be discriminated, and generates a corresponding music feature matrix;
  • the determined music velocity annotation model is identified, and the music score marked with music intensity is output;
  • the music score marked with the music intensity is analyzed according to the predetermined music recognition model, and whether the music score marked with the music strength meets the predefined music standard; if yes, Then, it is determined that the score of the music to be discriminated is qualified, or, if not, the score of the music to be discriminated is determined to be unqualified.
  • the quality of the musical piece can be accurately identified, and the method is simple, flexible and practical.
  • the deep learning-based music score recognition program of the present application may be described by a program module having the same function according to different functions implemented by the respective parts.
  • FIG. 2 is a schematic diagram of a program module of a music score recognition program based on deep learning in an embodiment of the electronic device of the present application.
  • the music recognition program based on deep learning may be divided into an acquisition module 201, an identification module 202, an analysis module 203, and a determination module 204 according to different functions implemented by the respective parts.
  • the program module referred to in the present application refers to a series of computer program instruction segments capable of performing a specific function, and is more suitable than the program to describe the execution process of the depth learning-based music score recognition program in the electronic device 10.
  • the functions or operational steps implemented by the modules 201-204 are similar to the above, and are not described in detail herein, by way of example, for example:
  • the obtaining module 201 is configured to obtain a music element in a musical score of the music quality to be discriminated, and preprocess the acquired music element to generate a corresponding music feature matrix;
  • the identification module 202 is configured to substitute the music feature matrix into a predetermined music strength annotation model for recognition, and output a music score marked with music strength;
  • the analyzing module 203 is configured to analyze the music score marked with the music intensity according to the predetermined music recognition model, and determine whether the music score marked with the music strength meets the predefined music standard;
  • the determining module 204 is configured to determine that if the music score marked with the music strength meets the predefined music standard, determine that the music score of the music to be determined is qualified, or determine that if the music score marked with the music strength does not meet the predefined music standard, It is determined that the score of the music to be discriminated is unqualified.
  • the present application also provides a music learning method based on deep learning.
  • the depth learning based music score recognition method includes the following steps:
  • Step S301 Acquire a music element in a musical score of the music quality to be discriminated, and preprocess the acquired music element to generate a corresponding music feature matrix;
  • Step S302 substituting the music feature matrix into a predetermined music strength annotation model for recognition, and outputting a music score marked with music strength;
  • Step S303 analyzing a music score marked with music strength according to a predetermined music recognition model, and determining whether the music score marked with the music strength meets a predefined music standard;
  • Step S304 if it is met, it is determined that the score of the music to be discriminated is qualified, or, if not, the score of the music to be discriminated is determined to be unqualified.
  • the music element is pitch and music intensity
  • the step of pre-processing the acquired music element to generate a corresponding music feature matrix comprises: obtaining the pitch and a predefined vibration frequency value (Predefined with 128 pitches, each pitch has three representations) to match, matching the vibration frequency values corresponding to each pitch;
  • the vibration frequency value after matching is identified by a predefined pitch identification method (for example, the predefined pitch identification mode is, C1, 0, 0 means that there is no sound in the C1 frequency segment, and C1, 0, 1 indicates that the frequency is in the C1 frequency range. Short tone, C1,1,1 indicates the extension of the C1 frequency band);
  • the music strength annotation model and the music recognition model are pre-trained generated confrontation networks (GAN), and the GAN network includes a Generative Model and a Discriminative Model.
  • the generation model is used to annotate the musical strength
  • the discriminant model is used to identify whether the musical score conforms to the music standard.
  • the Generative Model is a convolution-based neural network (CNN), and the discriminant model is a recognition model obtained by training the convolution based neural network;
  • the Generative Model is an LSTM long and short memory neural network
  • the discriminant model is a recognition model obtained based on the LSTM long and short memory neural network training
  • the recognition model generally outputs a value of a probability function value.
  • the probability function value conforms to a normal probability distribution, and the recognition result conforms to a preset criterion. If the probability function value does not conform to the normal probability distribution, the recognition result is represented. Does not meet the preset criteria.
  • the generation of music is taken as an example to illustrate the principle of GAN.
  • the generation model is a music intensity annotation network, which receives a random sound Z, and the sound intensity is marked by this sound, and is recorded as G(Z).
  • the recognition model is a discriminant network that discriminates whether the intensity of the annotated music is "consistent with the performance scene.” Its input parameter is X, X represents a music with the strength of the music, and the output D(X) represents the probability that X is the strength of the music in accordance with the performance scene. If it is 1, it means that 100% is the real music that matches the performance scene.
  • the strength of the label, and the output is 0, it means that it is impossible to be a true annotation of the musical intensity of the performance scene.
  • the goal of generating the network is to generate a true mark of the music that matches the performance scene to deceive the discriminant network.
  • the goal of discriminating the network is to separate the music generated by the generated network and label the music strength.
  • the generation network and the discriminant network constitute a dynamic "game process".
  • the deep learning-based music score recognition method proposed by the present application preprocesses the acquired music elements by acquiring music elements in the musical scores of the music quality to be discriminated, and generates a corresponding music feature matrix;
  • the music feature matrix is substituted into a predetermined music velocity annotation model for recognition, and the music score marked with music intensity is output; the music score marked with the music intensity is analyzed according to the predetermined music recognition model, and whether the music score marked with the music strength meets the predefined music is determined.
  • the standard if it is met, it is determined that the score of the music to be discriminated is qualified, or, if not, the score of the music to be discriminated is determined to be unqualified.
  • the quality of the musical piece can be accurately identified, and the method is simple, flexible and practical.
  • the present application further provides a computer readable storage medium on which a deep learning-based musical score recognition program is stored, and the deep learning-based musical score recognition program is executed by a processor to:
  • the scores of the music intensity are analyzed, and whether the scores marked with the music strength meet the predefined music standards are determined;
  • the score of the music to be discriminated is qualified, or, if not, the score of the music to be discriminated is determined to be unqualified.
  • the specific embodiment of the computer readable storage medium of the present application is substantially the same as the above embodiments of the electronic device and the deep learning based music score recognition method, and will not be described herein.
  • the foregoing embodiment method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be through hardware, but in many cases, the former is better.
  • Implementation Based on such understanding, the technical solution of the present application, which is essential or contributes to the prior art, may be embodied in the form of a software product stored in a storage medium (such as ROM/RAM, disk,
  • the optical disc includes a number of instructions for causing a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the methods described in various embodiments of the present application.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Data Mining & Analysis (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Health & Medical Sciences (AREA)
  • Software Systems (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Computational Linguistics (AREA)
  • Mathematical Physics (AREA)
  • Multimedia (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Signal Processing (AREA)
  • Probability & Statistics with Applications (AREA)
  • Auxiliary Devices For Music (AREA)

Abstract

本申请公开了一种电子装置、基于深度学习的乐谱识别方法及存储介质,通过获取待判别音乐质量的乐谱中的音乐元素,将获取的音乐元素进行预处理,生成对应的音乐特征矩阵;将所述音乐特征矩阵代入预先确定的音乐力度标注模型进行识别,输出标注了音乐力度的乐谱;根据预先确定的音乐识别模型分析标注了音乐力度的乐谱,确定标注了音乐力度的乐谱是否符合预定义的音乐标准;若符合,则确定待判别音乐质量的乐谱合格,或者,若不符合,则确定待判别音乐质量的乐谱不合格。能够准确地识别出音乐作品的质量,且该方法简单灵活实用性强。

Description

电子装置、基于深度学习的乐谱识别方法、系统及存储介质
本申请要求于2018年4月9日提交中国专利局、申请号为2018103124305,发明名称为“电子装置、基于深度学习的乐谱识别方法及存储介质”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本申请涉及深度学习领域,尤其涉及一种电子装置、基于深度学习的乐谱识别方法、系统及存储介质。
背景技术
目前,分析音乐作品的好坏,需要从专业角度讨论音乐最基本的元素和结构,例如和声、配器、旋律、调式、律动的特点等,这些通常是由专业的音乐人士根据多年积累的经验进行分析的,而对于普通的音乐爱好者或者音乐初学者来说,如何分析自己创作的音乐作品的好坏,存在一定的难度,严重影响学习兴趣以及学习效率。
发明内容
有鉴于此,本申请提出一种电子装置、基于深度学习的乐谱识别方法及存储介质,能够准确地识别出音乐作品的质量,且该方法简单灵活实用性强。
首先,为实现上述目的,本申请提出一种电子装置,所述电子装置包括存储器、及与所述存储器连接的处理器,所述处理器用于执行所述存储器上存储的基于深度学习的乐谱识别程序,所述基于深度学习的乐谱识别程序被所述处理器执行时实现如下步骤:
获取待判别音乐质量的乐谱中的音乐元素,将获取的音乐元素进行预处理,生成对应的音乐特征矩阵;
将所述音乐特征矩阵代入预先确定的音乐力度标注模型进行识别,输出标注了音乐力度的乐谱;
根据预先确定的音乐识别模型分析标注了音乐力度的乐谱,确定标注了音乐力度的乐谱是否符合预定义的音乐标准;
若符合,则确定待判别音乐质量的乐谱合格,或者,若不符合,则确定待判别音乐质量的乐谱不合格。
此外,为实现上述目的,本申请还提供一种基于深度学习的乐谱识别方法,该方法包括如下步骤:
获取待判别音乐质量的乐谱中的音乐元素,将获取的音乐元素进行预处理,生成对应的音乐特征矩阵;
将所述音乐特征矩阵代入预先确定的音乐力度标注模型进行识别,输出标注了音乐力度的乐谱;
根据预先确定的音乐识别模型分析标注了音乐力度的乐谱,确定标注了音乐力度的乐谱是否符合预定义的音乐标准;
若符合,则确定待判别音乐质量的乐谱合格,或者,若不符合,则确定待判别音乐质量的乐谱不合格。
此外,为实现上述目的,本申请还提供一种基于深度学习的乐谱识别系统,该系统包括获取模块、识别模块、分析模块以及确定模块;
所述获取模块用于获取待判别音乐质量的乐谱中的音乐元素,将获取的音乐元素进行预处理,生成对应的音乐特征矩阵;
所述识别模块用于将所述音乐特征矩阵代入预先确定的音乐力度标注模型进行识别,输出标注了音乐力度的乐谱;
所述分析模块用于根据预先确定的音乐识别模型分析标注了音乐力度的乐谱,确定标注了音乐力度的乐谱是否符合预定义的音乐标准;
所述确定模块用于在确定标准了音乐力度的乐谱若符合预定义的音乐标准后,则确定待判别音乐质量的乐谱合格,或者,若不符合,则确定待判别 音乐质量的乐谱不合格。
此外,为实现上述目的,本申请还提供一种计算机可读存储介质,所述计算机可读存储介质存储有基于深度学习的乐谱识别程序,所述基于深度学习的乐谱识别程序可被至少一个处理器执行,以使所述至少一个处理器执行如下步骤:
获取待判别音乐质量的乐谱中的音乐元素,将获取的音乐元素进行预处理,生成对应的音乐特征矩阵;
将所述音乐特征矩阵代入预先确定的音乐力度标注模型进行识别,输出标注了音乐力度的乐谱;
根据预先确定的音乐识别模型分析标注了音乐力度的乐谱,确定标注了音乐力度的乐谱是否符合预定义的音乐标准;
若符合,则确定待判别音乐质量的乐谱合格,或者,若不符合,则确定待判别音乐质量的乐谱不合格。
相较于现有技术,本申请所提出的电子装置、基于深度学习的乐谱识别方法、系统及存储介质,通过获取待判别音乐质量的乐谱中的音乐元素,将获取的音乐元素进行预处理,生成对应的音乐特征矩阵;将所述音乐特征矩阵代入预先确定的音乐力度标注模型进行识别,输出标注了音乐力度的乐谱;根据预先确定的音乐识别模型分析标注了音乐力度的乐谱,确定标注了音乐力度的乐谱是否符合预定义的音乐标准;若符合,则确定待判别音乐质量的乐谱合格,或者,若不符合,则确定待判别音乐质量的乐谱不合格。能够准确地识别出音乐作品的质量,且该方法简单灵活实用性强。
附图说明
图1是本申请提出的电子装置一可选的硬件架构的示意图;
图2是本申请电子装置一实施例中基于深度学习的乐谱识别程序的程序模块示意图;
图3是本申请基于深度学习的乐谱识别方法较佳实施例的实施流程图。
本申请目的的实现、功能特点及优点将结合实施例,参照附图做进一步说明。
具体实施方式
为了使本申请的目的、技术方案及优点更加清楚明白,以下结合附图及实施例,对本申请进行进一步详细说明。应当理解,此处所描述的具体实施例仅用以解释本申请,并不用于限定本申请。基于本申请中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例,都属于本申请保护的范围。
需要说明的是,在本申请中涉及“第一”、“第二”等的描述仅用于描述目的,而不能理解为指示或暗示其相对重要性或者隐含指明所指示的技术特征的数量。由此,限定有“第一”、“第二”的特征可以明示或者隐含地包括至少一个该特征。另外,各个实施例之间的技术方案可以相互结合,但是必须是以本领域普通技术人员能够实现为基础,当技术方案的结合出现相互矛盾或无法实现时应当认为这种技术方案的结合不存在,也不在本申请要求的保护范围之内。
参阅图1所示,是本申请提出的电子装置一可选的硬件架构示意图。本实施例中,电子装置10可包括,但不仅限于,可通过通信总线14相互通信连接存储器11、处理器12、网络接口13。需要指出的是,图1仅示出了具有组件11-14的电子装置10,但是应理解的是,并不要求实施所有示出的组件,可以替代的实施更多或者更少的组件。
其中,存储器11至少包括一种类型的计算机可读存储介质,计算机可读存储介质包括闪存、硬盘、多媒体卡、卡型存储器(例如,SD或DX存储器等)、随机访问存储器(RAM)、静态随机访问存储器(SRAM)、只读存储器(ROM)、电可擦除可编程只读存储器(EEPROM)、可编程只读存储器 (PROM)、磁性存储器、磁盘、光盘等。在一些实施例中,存储器11可以是电子装置10的内部存储单元,例如电子装置10的硬盘或内存。在另一些实施例中,存储器11也可以是电子装置10的外包存储设备,例如电子装置10上配备的插接式硬盘,智能存储卡(Smart Media Card,SMC),安全数字(Secure Digital,SD)卡,闪存卡(Flash Card)等。当然,存储器11还可以既包括电子装置10的内部存储单元也包括其外包存储设备。本实施例中,存储器11通常用于存储安装于电子装置10的操作系统和各类应用软件,例如基于深度学习的乐谱识别程序等。此外,存储器11还可以用于暂时地存储已经输出或者将要输出的各类数据。
处理器12在一些实施例中可以是中央处理器(Central Processing Unit,CPU)、控制器、微控制器、微处理器、或其他数据处理芯片。处理器12通常用于控制电子装置10的总体操作。本实施例中,处理器12用于运行存储器11中存储的程序代码或者处理数据,例如运行的基于深度学习的乐谱识别程序等。
网络接口13可包括无线网络接口或有线网络接口,网络接口13通常用于在电子装置10与其他电子设备之间建立通信连接。
通信总线14用于实现组件11-13之间的通信连接。
图1仅示出了具有组件11-14以及基于深度学习的乐谱识别程序的电子装置10,但是应理解的是,并不要求实施所有示出的组件,可以替代的实施更多或者更少的组件。
可选地,电子装置10还可以包括用户接口(图1中未示出),用户接口可以包括显示器、输入单元比如键盘,其中,用户接口还可以包括标准的有线接口、无线接口等。
可选地,在一些实施例中,显示器可以是LED显示器、液晶显示器、触控式液晶显示器以及OLED触摸器等。进一步地,显示器也可称为显示屏或显示单元,用于显示在电子装置10中处理信息以及用于显示可视化的用户界 面。
在一实施例中,存储器11中存储的基于深度学习的乐谱识别程序被处理器12执行时,实现如下操作:
获取待判别音乐质量的乐谱中的音乐元素,将获取的音乐元素进行预处理,生成对应的音乐特征矩阵;
将所述音乐特征矩阵代入预先确定的音乐力度标注模型进行识别,输出标注了音乐力度的乐谱;
根据预先确定的音乐识别模型分析标注了音乐力度的乐谱,确定标注了音乐力度的乐谱是否符合预定义的音乐标准;
若符合,则确定待判别音乐质量的乐谱合格,或者,若不符合,则确定待判别音乐质量的乐谱不合格。
在一实施例中,所述音乐元素为音高以及音乐力度,所述将获取的音乐元素进行预处理,生成对应的音乐特征矩阵的步骤包括,将获取的音高与预定义的振动频率值(预定义有128个音高,每个音高均有三种表示方式)进行匹配,匹配出各个音高对应的振动频率值;
将匹配之后的振动频率值用预定义音高标识方式进行标识(例如预定义的音高标识方式为,C1,0,0表示在C1频率段没有音,C1,0,1表示在C1频率段的短促音,C1,1,1表示在C1频率段的延长音);
根据预定义的时间间隔周期获取标识了音高的振动频率值,以及分别获取在各个所述预定义的时间间隔周期内的音高的数量;
根据获取的振动频率值以及获取的音高的数量生成二维矩阵,其中,所述二维矩阵的一个维度表示音高的数量及音高的标识,另一个维度表示预定义的时间间隔。
在本实施例中,所述音乐力度标注模型及所述音乐识别模型均为预先训练完成的生成式对抗网络(GAN),所述GAN网络包括生成模型(Generative  Model)及判别模型(Discriminative Model),所述生成模型用于标注音乐力度,所述判别模型用于识别乐谱是否符合音乐标准。
进一步地,在本实施例中,所述生成模型(Generative Model)为基于卷积的神经网络(CNN),所述判别模型为基于所述卷积的神经网络训练得到的识别模型;
在另本实施例的一种实施方式中,所述生成模型(Generative Model)为LSTM长短记忆神经网络,所述判别模型为基于所述LSTM长短记忆神经网络训练得到的识别模型;
所述识别模型通常输出的值为一个概率函数值,通常该概率函数值符合正态概率分布,则表示识别结果符合预设的标准,该概率函数值不符合正态概率分布,则表示识别结果不符合预设的标准。
在本实施例中,假设以生成音乐为例说明GAN的原理,假设生成模型是一个音乐力度标注网络,它接收一个随机的声音Z,通过这个声音进行音乐力度的标注,记做G(Z)。识别模型是一个判别网络,判别标注的音乐力度是不是“符合演奏场景的”。它输入参数是X,X代表一首标注了音乐力度的音乐,输出D(X)代表X为符合演奏场景的音乐力度的概率,如果是1,就代表100%是真实的符合演奏场景的音乐力度的标注,而输出是0,就代表不可能是真实的符合演奏场景的音乐力度的标注。在训练过程中,生成网络的目标就是尽量生成真实的符合演奏场景的音乐力度的标注去欺骗判别网络。而判别网络的目标就是尽量把生成网络生成的标注了音乐力度的音乐分别出来,这样,生成网络和判别网络构成一个动态的“博弈过程”。在最理性的状态下,生成网络可以标注出以假乱真的音乐力度G(Z)。判别网络难以判定生成网络标注的音乐力度是不是符合真实演奏场景的,此时,D(G(Z))=0.5,符合正态分布。
由上述事实施例可知,本申请提出的电子装置通过获取待判别音乐质量的乐谱中的音乐元素,将获取的音乐元素进行预处理,生成对应的音乐特征矩阵;将所述音乐特征矩阵代入预先确定的音乐力度标注模型进行识别,输 出标注了音乐力度的乐谱;根据预先确定的音乐识别模型分析标注了音乐力度的乐谱,确定标注了音乐力度的乐谱是否符合预定义的音乐标准;若符合,则确定待判别音乐质量的乐谱合格,或者,若不符合,则确定待判别音乐质量的乐谱不合格。能够准确地识别出音乐作品的质量,且该方法简单灵活实用性强。
进一步需要说明的是,本申请的基于深度学习的乐谱识别程序依据其各部分所实现的功能不同,可用具有相同功能的程序模块进行描述。请参阅图2所示,是本申请电子装置一实施例中基于深度学习的乐谱识别程序的程序模块示意图。本实施例中,基于深度学习的乐谱识别程序依据其各部分所实现的功能的不同,可以被分割成获取模块201、识别模块202、分析模块203、确定模块204。由上面的描述可知,本申请所称的程序模块是指能够完成特定功能的一系列计算机程序指令段,比程序更适合于描述基于深度学习的乐谱识别程序在电子装置10中的执行过程。所述模块201-204所实现的功能或操作步骤均与上文类似,此处不再详述,示例性地,例如其中:
获取模块201用于获取待判别音乐质量的乐谱中的音乐元素,将获取的音乐元素进行预处理,生成对应的音乐特征矩阵;
识别模块202用于将所述音乐特征矩阵代入预先确定的音乐力度标注模型进行识别,输出标注了音乐力度的乐谱;
分析模块203用于根据预先确定的音乐识别模型分析标注了音乐力度的乐谱,确定标注了音乐力度的乐谱是否符合预定义的音乐标准;
确定模块204用于确定若标注了音乐力度的乐谱符合预定义的音乐标准,则确定待判别音乐质量的乐谱合格,或者用于确定若标注了音乐力度的乐谱不符合预定义的音乐标准,则确定待判别音乐质量的乐谱不合格。
此外,本申请还提出一种基于深度学习的乐谱识别方法,请参阅图3所示,所述基于深度学习的乐谱识别方法包括如下步骤:
步骤S301,获取待判别音乐质量的乐谱中的音乐元素,将获取的音乐元素进行预处理,生成对应的音乐特征矩阵;
步骤S302,将所述音乐特征矩阵代入预先确定的音乐力度标注模型进行识别,输出标注了音乐力度的乐谱;
步骤S303,根据预先确定的音乐识别模型分析标注了音乐力度的乐谱,确定标注了音乐力度的乐谱是否符合预定义的音乐标准;
步骤S304,若符合,则确定待判别音乐质量的乐谱合格,或者,若不符合,则确定待判别音乐质量的乐谱不合格。
在一实施例中,所述音乐元素为音高以及音乐力度,所述将获取的音乐元素进行预处理,生成对应的音乐特征矩阵的步骤包括,将获取的音高与预定义的振动频率值(预定义有128个音高,每个音高均有三种表示方式)进行匹配,匹配出各个音高对应的振动频率值;
将匹配之后的振动频率值用预定义音高标识方式进行标识(例如预定义的音高标识方式为,C1,0,0表示在C1频率段没有音,C1,0,1表示在C1频率段的短促音,C1,1,1表示在C1频率段的延长音);
根据预定义的时间间隔周期获取标识了音高的振动频率值,以及分别获取在各个所述预定义的时间间隔周期内的音高的数量;
根据获取的振动频率值以及获取的音高的数量生成二维矩阵,其中,所述二维矩阵的一个维度表示音高的数量及音高的标识,另一个维度表示预定义的时间间隔
在本实施例中,所述音乐力度标注模型及所述音乐识别模型均为预先训练完成的生成式对抗网络(GAN),所述GAN网络包括生成模型(Generative Model)及判别模型(Discriminative Model),所述生成模型用于标注音乐力度,所述判别模型用于识别乐谱是否符合音乐标准。
进一步地,在本实施例中,所述生成模型(Generative Model)为基于卷积的神经网络(CNN),所述判别模型为基于所述卷积的神经网络训练得到的识别模型;
在另本实施例的一种实施方式中,所述生成模型(Generative Model)为LSTM长短记忆神经网络,所述判别模型为基于所述LSTM长短记忆神经网络训练得到的识别模型;
所述识别模型通常输出的值为一个概率函数值,通常该概率函数值符合正态概率分布,则表示识别结果符合预设的标准,该概率函数值不符合正态概率分布,则表示识别结果不符合预设的标准。
在本实施例中,假设以生成音乐为例说明GAN的原理,假设生成模型是一个音乐力度标注网络,它接收一个随机的声音Z,通过这个声音进行音乐力度的标注,记做G(Z)。识别模型是一个判别网络,判别标注的音乐力度是不是“符合演奏场景的”。它输入参数是X,X代表一首标注了音乐力度的音乐,输出D(X)代表X为符合演奏场景的音乐力度的概率,如果是1,就代表100%是真实的符合演奏场景的音乐力度的标注,而输出是0,就代表不可能是真实的符合演奏场景的音乐力度的标注。在训练过程中,生成网络的目标就是尽量生成真实的符合演奏场景的音乐力度的标注去欺骗判别网络。而判别网络的目标就是尽量把生成网络生成的标注了音乐力度的音乐分别出来,这样,生成网络和判别网络构成一个动态的“博弈过程”。在最理性的状态下,生成网络可以标注出以假乱真的音乐力度G(Z)。判别网络难以判定生成网络标注的音乐力度是不是符合真实演奏场景的,此时,D(G(Z))=0.5,符合正态分布。
由上述事实施例可知,本申请提出的基于深度学习的乐谱识别方法通过获取待判别音乐质量的乐谱中的音乐元素,将获取的音乐元素进行预处理,生成对应的音乐特征矩阵;将所述音乐特征矩阵代入预先确定的音乐力度标注模型进行识别,输出标注了音乐力度的乐谱;根据预先确定的音乐识别模 型分析标注了音乐力度的乐谱,确定标注了音乐力度的乐谱是否符合预定义的音乐标准;若符合,则确定待判别音乐质量的乐谱合格,或者,若不符合,则确定待判别音乐质量的乐谱不合格。能够准确地识别出音乐作品的质量,且该方法简单灵活实用性强。
此外,本申请还提出一种计算机可读存储介质,所述计算机可读存储介质上存储有基于深度学习的乐谱识别程序,所述基于深度学习的乐谱识别程序被处理器执行时实现如下操作:
获取待判别音乐质量的乐谱中的音乐元素,将获取的音乐元素进行预处理,生成对应的音乐特征矩阵;
将所述音乐特征矩阵代入预先确定的音乐力度标注模型进行识别,输出标注了音乐力度的乐谱;
根据预先确定的音乐识别模型分析标注了音乐力度的乐谱,确定标注了音乐力度的乐谱是否符合预定义的音乐标准;
若符合,则确定待判别音乐质量的乐谱合格,或者,若不符合,则确定待判别音乐质量的乐谱不合格。
本申请计算机可读存储介质具体实施方式与上述电子装置以及基于深度学习的乐谱识别方法各实施例基本相同,在此不作累述。
上述本申请实施例序号仅仅为了描述,不代表实施例的优劣。
通过以上的实施方式的描述,本领域的技术人员可以清楚地了解到上述实施例方法可借助软件加必需的通用硬件平台的方式来实现,当然也可以通过硬件,但很多情况下前者是更佳的实施方式。基于这样的理解,本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来,该计算机软件产品存储在一个存储介质(如ROM/RAM、磁碟、光盘)中,包括若干指令用以使得一台终端设备(可以是手机,计算机,服务器,空调器,或者网络设备等)执行本申请各个实施例所述的方法。
以上仅为本申请的优选实施例,并非因此限制本申请的专利范围,凡是 利用本申请说明书及附图内容所作的等效结构或等效流程变换,或直接或间接运用在其他相关的技术领域,均同理包括在本申请的专利保护范围内。

Claims (20)

  1. 一种电子装置,其特征在于,所述电子装置包括存储器、及与所述存储器连接的处理器,所述处理器用于执行所述存储器上存储的基于深度学习的乐谱识别程序,所述基于深度学习的乐谱识别程序被所述处理器执行时实现如下步骤:
    获取待判别音乐质量的乐谱中的音乐元素,将获取的音乐元素进行预处理,生成对应的音乐特征矩阵;
    将所述音乐特征矩阵代入预先确定的音乐力度标注模型进行识别,输出标注了音乐力度的乐谱;
    根据预先确定的音乐识别模型分析标注了音乐力度的乐谱,确定标注了音乐力度的乐谱是否符合预定义的音乐标准;
    若符合,则确定待判别音乐质量的乐谱合格,或者,若不符合,则确定待判别音乐质量的乐谱不合格。
  2. 如权利要求1所述的电子装置,其特征在于,所述音乐元素为音高以及音乐力度,所述将获取的音乐元素进行预处理,生成对应的音乐特征矩阵的步骤,包括:
    将获取的音高与预定义的振动频率值进行匹配,匹配出各个音高对应的振动频率值;
    将匹配之后的振动频率值用预定义音高标识方式进行标识;
    根据预定义的时间间隔周期获取标识了音高的振动频率值,以及分别获取在各个所述预定义的时间间隔周期内的音高的数量;
    根据获取的振动频率值以及获取的音高的数量生成二维矩阵,其中,所述二维矩阵的一个维度表示音高的数量及音高的标识,另一个维度表示预定义的时间间隔。
  3. 如权利要求1所述的电子装置,其特征在于,所述音乐力度标注模型及所述音乐识别模型均为预先训练完成的生成式对抗网络;
    所述生成式对抗网络包括生成模型及判别模型;
    所述生成模型用于标注音乐力度,所述判别模型用于识别乐谱是否符合音乐标准。
  4. 如权利要求3所述的电子装置,其特征在于,所述生成模型为预先训练完成的基于卷积的神经网络,所述判别模型为基于所述卷积的神经网络训练得到的识别模型。
  5. 如权利要求4所述的电子装置,其特征在于,所述生成模型为预先训练完成的LSTM长短记忆神经网络,所述判别模型为基于所述LSTM长短记忆神经网络训练得到的识别模型。
  6. 一种基于深度学习的乐谱识别方法,其特征在于,所述方法包括如下步骤:
    获取待判别音乐质量的乐谱中的音乐元素,将获取的音乐元素进行预处理,生成对应的音乐特征矩阵;
    将所述音乐特征矩阵代入预先确定的音乐力度标注模型进行识别,输出标注了音乐力度的乐谱;
    根据预先确定的音乐识别模型分析标注了音乐力度的乐谱,确定标注了音乐力度的乐谱是否符合预定义的音乐标准;
    若符合,则确定待判别音乐质量的乐谱合格,或者,若不符合,则确定待判别音乐质量的乐谱不合格。
  7. 如权利要求6所述的基于深度学习的乐谱识别方法,其特征在于,所述音乐元素为音高以及音乐力度,所述将获取的音乐元素进行预处理,生成对应的音乐特征矩阵的步骤,包括:
    将获取的音高与预定义的振动频率值进行匹配,匹配出各个音高对应的振动频率值;
    将匹配之后的振动频率值用预定义音高标识方式进行标识;
    根据预定义的时间间隔周期获取标识了音高的振动频率值,以及分别获取在各个所述预定义的时间间隔周期内的音高的数量;
    根据获取的振动频率值以及获取的音高的数量生成二维矩阵,其中,所述二维矩阵的一个维度表示音高的数量及音高的标识,另一个维度表示预定义的时间间隔。
  8. 如权利要求6所述的基于深度学习的乐谱识别方法,其特征在于,所述音乐力度标注模型及所述音乐识别模型均为预先训练完成的生成式对抗网络;
    所述生成式对抗网络包括生成模型及判别模型;
    所述生成模型用于标注音乐力度,所述判别模型用于识别乐谱是否符合音乐标准。
  9. 如权利要求8所述的基于深度学习的乐谱识别方法,其特征在于,所述生成模型为预先训练完成的基于卷积的神经网络,所述判别模型为基于所述卷积的神经网络训练得到的识别模型。
  10. 如权利要求9所述的基于深度学习的乐谱识别方法,其特征在于,所述生成模型为预先训练完成的LSTM长短记忆神经网络,所述判别模型为基于所述LSTM长短记忆神经网络训练得到的识别模型。
  11. 一种基于深度学习的乐谱识别系统,其特征在于,所述系统包括获取模块、识别模块、分析模块以及确定模块;
    所述获取模块用于获取待判别音乐质量的乐谱中的音乐元素,将获取的音乐元素进行预处理,生成对应的音乐特征矩阵;
    所述识别模块用于将所述音乐特征矩阵代入预先确定的音乐力度标注模型进行识别,输出标注了音乐力度的乐谱;
    所述分析模块用于根据预先确定的音乐识别模型分析标注了音乐力度的乐谱,确定标注了音乐力度的乐谱是否符合预定义的音乐标准;
    所述确定模块用于在确定标注了音乐力度的乐谱若符合预定义的音乐标准后,则确定待判别音乐质量的乐谱合格,或者,若不符合,则确定待判别音乐质量的乐谱不合格。
  12. 如权利要求11所述的基于深度学习的乐谱识别系统,其特征在于,所述音乐元素为音高以及音乐力度,所述将获取的音乐元素进行预处理,生成对应的音乐特征矩阵的步骤,包括:
    将获取的音高与预定义的振动频率值进行匹配,匹配出各个音高对应的振动频率值;
    将匹配之后的振动频率值用预定义音高标识方式进行标识;
    根据预定义的时间间隔周期获取标识了音高的振动频率值,以及分别获取在各个所述预定义的时间间隔周期内的音高的数量;
    根据获取的振动频率值以及获取的音高的数量生成二维矩阵,其中,所述二维矩阵的一个维度表示音高的数量及音高的标识,另一个维度表示预定义的时间间隔。
  13. 如权利要求11所述的基于深度学习的乐谱识别系统,其特征在于,所述音乐力度标注模型及所述音乐识别模型均为预先训练完成的生成式对抗网络;
    所述生成式对抗网络包括生成模型及判别模型;
    所述生成模型用于标注音乐力度,所述判别模型用于识别乐谱是否符合音乐标准。
  14. 如权利要求13所述的基于深度学习的乐谱识别系统,其特征在于,所述生成模型为预先训练完成的基于卷积的神经网络,所述判别模型为基于所述卷积的神经网络训练得到的识别模型。
  15. 如权利要求14所述的基于深度学习的乐谱识别系统,其特征在于,所述生成模型为预先训练完成的LSTM长短记忆神经网络,所述判别模型为基于所述LSTM长短记忆神经网络训练得到的识别模型。
  16. 一种计算机可读存储介质,所述计算机可读存储介质存储有基于深度学习的乐谱识别程序,所述基于深度学习的乐谱识别程序可被至少一个处理器执行,以使所述至少一个处理器执行如下步骤:
    获取待判别音乐质量的乐谱中的音乐元素,将获取的音乐元素进行预处理,生成对应的音乐特征矩阵;
    将所述音乐特征矩阵代入预先确定的音乐力度标注模型进行识别,输出标注了音乐力度的乐谱;
    根据预先确定的音乐识别模型分析标注了音乐力度的乐谱,确定标注了音乐力度的乐谱是否符合预定义的音乐标准;
    若符合,则确定待判别音乐质量的乐谱合格,或者,若不符合,则确定待判别音乐质量的乐谱不合格。
  17. 如权利要求16所述的电子装置,其特征在于,所述音乐元素为音高以及音乐力度,所述将获取的音乐元素进行预处理,生成对应的音乐特征矩阵的步骤,包括:
    将获取的音高与预定义的振动频率值进行匹配,匹配出各个音高对应的振动频率值;
    将匹配之后的振动频率值用预定义音高标识方式进行标识;
    根据预定义的时间间隔周期获取标识了音高的振动频率值,以及分别获取在各个所述预定义的时间间隔周期内的音高的数量;
    根据获取的振动频率值以及获取的音高的数量生成二维矩阵,其中,所述二维矩阵的一个维度表示音高的数量及音高的标识,另一个维度表示预定义的时间间隔。
  18. 如权利要求16所述的电子装置,其特征在于,所述音乐力度标注模型及所述音乐识别模型均为预先训练完成的生成式对抗网络;
    所述生成式对抗网络包括生成模型及判别模型;
    所述生成模型用于标注音乐力度,所述判别模型用于识别乐谱是否符合音乐标准。
  19. 如权利要求18所述的电子装置,其特征在于,所述生成模型为预先训练完成的基于卷积的神经网络,所述判别模型为基于所述卷积的神经网络训练得到的识别模型。
  20. 如权利要求19所述的电子装置,其特征在于,所述生成模型为预先训练完成的LSTM长短记忆神经网络,所述判别模型为基于所述LSTM长短记忆神经网络训练得到的识别模型。
PCT/CN2018/102113 2018-04-09 2018-08-24 电子装置、基于深度学习的乐谱识别方法、系统及存储介质 WO2019196301A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810312430.5A CN108805000B (zh) 2018-04-09 2018-04-09 电子装置、基于深度学习的乐谱识别方法及存储介质
CN201810312430.5 2018-04-09

Publications (1)

Publication Number Publication Date
WO2019196301A1 true WO2019196301A1 (zh) 2019-10-17

Family

ID=64095488

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/102113 WO2019196301A1 (zh) 2018-04-09 2018-08-24 电子装置、基于深度学习的乐谱识别方法、系统及存储介质

Country Status (2)

Country Link
CN (1) CN108805000B (zh)
WO (1) WO2019196301A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111275043A (zh) * 2020-01-22 2020-06-12 西北师范大学 一种基于pcnn处理的纸质简谱电子化播放装置

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109908578B (zh) * 2019-01-28 2022-07-05 努比亚技术有限公司 一种游戏震感控制方法、终端及计算机可读存储介质
CN110288965B (zh) * 2019-05-21 2021-06-18 北京达佳互联信息技术有限公司 一种音乐合成方法、装置、电子设备及存储介质
CN110443127A (zh) * 2019-06-28 2019-11-12 天津大学 结合残差卷积结构和循环神经网络的乐谱图像识别方法
CN113112969B (zh) * 2021-03-23 2024-04-05 平安科技(深圳)有限公司 基于神经网络的佛教音乐记谱方法、装置、设备及介质

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105976803A (zh) * 2016-04-25 2016-09-28 南京理工大学 一种结合乐谱的音符切分方法
CN106340286A (zh) * 2016-09-27 2017-01-18 华中科技大学 一种通用的实时乐器演奏评价系统
CN106446952A (zh) * 2016-09-28 2017-02-22 北京邮电大学 一种乐谱图像识别方法及装置

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2010518459A (ja) * 2007-02-14 2010-05-27 ミューズアミ, インコーポレイテッド 配布オーディオファイル編集用ウェブポータル
CN107146598B (zh) * 2016-05-28 2018-05-15 浙江大学 一种多音色混合的智能演奏系统和方法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105976803A (zh) * 2016-04-25 2016-09-28 南京理工大学 一种结合乐谱的音符切分方法
CN106340286A (zh) * 2016-09-27 2017-01-18 华中科技大学 一种通用的实时乐器演奏评价系统
CN106446952A (zh) * 2016-09-28 2017-02-22 北京邮电大学 一种乐谱图像识别方法及装置

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111275043A (zh) * 2020-01-22 2020-06-12 西北师范大学 一种基于pcnn处理的纸质简谱电子化播放装置
CN111275043B (zh) * 2020-01-22 2021-08-20 西北师范大学 一种基于pcnn处理的纸质简谱电子化播放装置

Also Published As

Publication number Publication date
CN108805000B (zh) 2019-12-17
CN108805000A (zh) 2018-11-13

Similar Documents

Publication Publication Date Title
WO2019196301A1 (zh) 电子装置、基于深度学习的乐谱识别方法、系统及存储介质
CN106960219B (zh) 图片识别方法及装置、计算机设备及计算机可读介质
CN106683680B (zh) 说话人识别方法及装置、计算机设备及计算机可读介质
WO2022116420A1 (zh) 语音事件检测方法、装置、电子设备及计算机存储介质
WO2021232594A1 (zh) 语音情绪识别方法、装置、电子设备及存储介质
JP6541673B2 (ja) モバイル機器におけるリアルタイム音声評価システム及び方法
WO2019019256A1 (zh) 电子装置、身份验证的方法、系统及计算机可读存储介质
WO2020015153A1 (zh) 为歌词文本生成乐曲的方法、装置及计算机可读存储介质
WO2019136909A1 (zh) 基于深度学习的语音活体检测方法、服务器及存储介质
CN110675862A (zh) 语料获取方法、电子装置及存储介质
CN111461168A (zh) 训练样本扩充方法、装置、电子设备及存储介质
WO2019179028A1 (zh) 电子装置、基于动态图片的用户验证方法及存储介质
WO2019085338A1 (zh) 电子装置、基于图像的年龄分类方法、系统及存储介质
WO2022247403A1 (zh) 关键点检测方法、电子设备、程序及存储介质
CN113450822B (zh) 语音增强方法、装置、设备及存储介质
CN111477200A (zh) 乐谱文件生成方法、装置、计算机设备和存储介质
CN109410972B (zh) 生成音效参数的方法、装置及存储介质
CN110827789A (zh) 音乐生成方法、电子装置及计算机可读存储介质
JP2004030694A (ja) デジタル映像テクスチャー分析方法
CN111695405B (zh) 一种狗脸特征点的检测方法、装置、系统及存储介质
US20190213989A1 (en) Technologies for generating a musical fingerprint
CN115631748A (zh) 基于语音对话的情感识别方法、装置、电子设备及介质
CN115329125A (zh) 一种歌曲串烧拼接方法和装置
CN110580905B (zh) 识别装置及方法
JPWO2019187107A1 (ja) 情報処理装置、制御方法、及びプログラム

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18914856

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 25/01/2021)

122 Ep: pct application non-entry in european phase

Ref document number: 18914856

Country of ref document: EP

Kind code of ref document: A1