CN112634842A

CN112634842A - New music generation method based on dual-mode network wandering fusion

Info

Publication number: CN112634842A
Application number: CN202011474157.XA
Authority: CN
Inventors: 谭虎; 雷杰
Original assignee: Hunan Institute of Engineering
Current assignee: Hunan Institute of Engineering
Priority date: 2020-12-14
Filing date: 2020-12-14
Publication date: 2021-04-09
Anticipated expiration: 2040-12-14
Also published as: CN112634842B

Abstract

The invention discloses a new song generation method based on dual-mode network wandering fusion, which comprises the following steps: constructing an original music score set, and constructing two mode node sets of pitch and rhythm according to each music score; connecting any two nodes in each node set to form two mode networks; setting initial conditions for random walk of two networks; judging whether the current walking step length of each network reaches a limited maximum value or not; and according to the judgment result, randomly walking to a neighbor node and judging again, or finishing walking and fusing the obtained mode sequences to generate a new song. The melody is divided into two variable dimensions of pitch and rhythm, two types of mode networks are constructed and are respectively subjected to random walk, and the walking result is fused to realize artificial composition, so that the problem that the note sequence generated by the random walk of the note network lacks rhythm is mainly solved; the method adopts the note length of two measures to construct the mode node, and simultaneously considers the continuity and the variability of the melody.

Description

New music generation method based on dual-mode network wandering fusion

Technical Field

The invention relates to a composing method based on complex network random walk, in particular to a new music generation method based on dual-mode network walk fusion.

Background

Human life and production activities depend on a large number of complex systems, both natural and man-made, and for a given system, the connections and interaction patterns between the components can be represented by networks, the components of the system can be abstracted into nodes in the network, and the connections between the components can be abstracted into edges. In the last two decades, the research of complex networks has stimulated the interest of a large number of application fields, and certain theories and application achievements have been achieved in the art field. Current basic theory research mainly performs complex network modeling and network feature analysis comparison on different types of music, such as the paper of silver, sores, etc. published in "Physica" in 2004 and the paper of Ferretti published in "Multimedia Tools & Applications" in 2017.

Further, there have been preliminary attempts at application studies, which mainly focused on artificial music production, such as the one published by Liu, Chi, et al, Physica A2010, and the one published by Gomez, Lorimer, et al, International Conference on Nonlinear Dynamics of Electronic Systems, 2014.

In combination with the existing application research, artificial composition based on a complex network is basically realized by random walk in the network, and the difference is mainly reflected in different modes of modeling and random walk. Network modeling typically abstracts notes into nodes and then treats the order of their association in time as edges. After the note network is generated, a new note sequence is generated through random walk in a certain mode. Random walks based on the note network have rich note walk choices, can form more music changes, and have defects in the formation stage of a final new music score. Because the randomly generated note sequence can not form regular pitch variation and rhythm variation in most cases, the generated new music score has poor rhythm sense.

Disclosure of Invention

The invention mainly aims to solve the problem that a note sequence generated by random walk in a note network lacks rhythm sense, and provides a new song generation method based on dual-mode network walk fusion. In the construction of the mode node, the note length of two measures is adopted, and the continuity and variability of pitch fluctuation and rhythm speed are considered.

In order to achieve the purpose, the invention adopts the technical scheme that:

the embodiment of the invention provides a new song generation method based on dual-mode network wandering fusion, which comprises the following steps:

s100, constructing an original koji spectrum set; the music score set comprises a plurality of groups of classical music scores;

s200, defining an abstract method of pitch and rhythm change mode nodes, and constructing a pitch and rhythm change mode node set according to each music score in the step S100; connecting any two nodes in each node set to form two fully-connected mode networks;

s300, setting initial conditions of random walk in the network for the mode network obtained in the step S200;

s400, judging whether each mode network executes the wandering currently or not according to the initial conditions set in the step S300 and the current actual wandering state of each mode network;

and S500, selecting to randomly walk to a neighbor node without limit according to the judgment result in the step S400, or restoring all the node sequences which are walked to a pitch and rhythm change sequence, and fusing the two sequences to generate a new song.

Further, the process of constructing the original music score set in step S100 includes:

constructing an original koji spectrum set:

M＝{S(k)}|card(M)＝n； k＝1,2,...,n

wherein, M is original music score set, S (k) represents kth classic music score, n represents total number of music score in music score set, | card (M) ═ n represents that number of different music score in music score set M is n, namely, same music score is not constructed.

Further, the step S200 includes:

s201: defining an abstract method of pitch and rhythm change mode nodes;

s2011, sorting the pitches of all notes of the first two bars of the current music score from low to high; counting the pitches, and taking the sorted median pitches as reference pitches when the counting result is odd; when the counting result is even, the bigger pitch in the middle two pitches after sorting is taken as the reference pitch; defining a reference pitch as a 0-order pitch, and solving a relative pitch according to the scale difference value between all original pitches of the current music score and the reference pitch;

accordingly, the abstract method of the pitch change mode node is as follows: abstracting a change sequence of relative pitch in every two subsections into a node N (i), wherein the node set N is defined as N ═ { N (i) };

s2012, defining the duration of the quarter note as 1 unit duration, and converting the durations of other notes according to the duration proportion of the quarter note to the duration proportion of the other notes;

accordingly, the abstract method of the rhythm change mode node comprises the following steps: abstracting duration value change sequences of all notes in every two measures into nodes M (i), wherein a set M of the nodes is defined as M ═ { M (i) };

s202: constructing a pitch and rhythm mode network;

according to the abstract method in the step S201, pitch and rhythm mode nodes are constructed, and the method includes:

s2021, constructing mode nodes N (i) and M (i) by using the classical curved spectrum S (k) in the step S100 according to the node abstraction method in the step S201; starting from a first classical curved spectrum S (1), constructing N (1) and M (1) by using a change sequence of relative pitch and time values in 1 st and 2 nd nodes; then constructing N (2) and M (2) in the change sequence of the relative pitch and time values in the 3 rd and 4 th measure; and so on until the mode nodes corresponding to the last two subsections of the 1 st classical music score are constructed; similarly, mode nodes corresponding to the remaining classical curved spectrums are sequentially constructed from S (2) to S (N), so that construction of two mode node sets N and M is completed;

s2022, connecting any two node pairs N (i) and N (j), M (i) and M (j) in each node set by adopting a non-directional edge to construct a pitch mode network and a rhythm mode network which are respectively defined as G_NAnd G_M。

Further, the step S300 includes:

s301, setting a wandering node;

G_Nand G_MThe middle wandering start node is respectively defined as S_NAnd S_MSet by the user; the nodes to be walked at present are respectively defined as V_NAnd V_MThe initial nodes are respectively S_NAnd S_M；

S302: setting a walking step length;

for G_NAnd G_MSetting the same maximum walk step length as T_maxSet by the user; the step length of actual wandering in the two mode networks is respectively defined as T_NvAnd T_MvThe initial values are all 0.

Further, the step S400 of determining whether each network currently executes the wandering includes:

according to the maximum step length of the wandering set in step S302 and according to G_NCurrent actual wandering step length T_NvAnd the current node V to be walked_NJudging whether to execute the next step of wandering according to the number of the adjacent nodes, and the steps are as follows:

wherein (T)_Nv≤T_max) Represents: to at G_NStep length T of middle actual wandering_NvLess than or equal to the maximum walk step length T_maxThe proposition is judged to be true or false; t represents that the proposition judgment result is true, and the walking is executed; f represents that the proposition judgment result is false and the walking is finished;

also according to G_MCurrent actual wandering step length T_MvAnd the current node V to be walked_MAnd judging whether to execute the next walk according to the number of the adjacent nodes.

Further, the step S500 includes:

according to the situation that the wandering needs to be executed in the step S400, firstly, the current node to be wandered is wandered to a neighbor node randomly without limit; the unlimited random walk steps are:

randomly selecting one node from the nodes pointed by the current nodes to be walked as a node to be walked, wherein the node is defined as V_next(ii) a Defining the node sequence of the wandering as TV, and setting the initial sequence as S]And adding the wandering node V after each wandering_nextTo node sequenceIn TV, i.e. TV ═ TV, V_next](ii) a Updating the original node V which is currently walked to V_nextI.e. V ═ V_next(ii) a Will actually wander step length T_vIs updated to T_v+1, i.e. T_v＝T_v+1；

After the current wandering is executed, returning to the step S400 to judge whether to execute new wandering; according to the circulation, the wandering is executed and whether the wandering is continued or not is judged;

according to the situation of ending the wandering in step S400, all the wandering pitch pattern node sequences TV are processed according to the corresponding relationship between the pitch pattern nodes and the relative pitch change in step S202_NReduction to relative pitch sequence TV 'in turn'_N(ii) a And, according to the corresponding relationship between rhythm pattern nodes and rhythm change in step S202, all the wandering rhythm pattern node sequences TV are sequenced_MReduction to the rhythmic sequence TV'_M；

One reference pitch is re-given by the user and TV'_NSequence TV converted back to Normal pitch_N(ii) a Taking two bars as a unit, let TV_NAnd TV'_MRecombining the sequence of notes in the order of pitch and duration, thereby generating a new song; in an actual blend of every two measures, the last duration is assigned equally to the overtaken pitch if the number of pitches in the pitch pattern is greater than the number of durations in the cadence pattern, and the last overtaken duration is assigned to the last pitch if the number of pitches in the pitch pattern is less than the number of durations in the cadence pattern.

Compared with the prior art, the invention has the following beneficial effects:

compared with the prior art, the method provided by the invention has the advantages that only the pitch and the duration of notes are randomly created and fused in manual composition, two sections are adopted as basic construction units of the mode, the continuity and the variability of the pitch and the rhythm are considered, and the construction complexity of the mode network is reduced.

Furthermore, the invention sets a reference pitch according to the pitch distributions of the first two measures before constructing a pitch pattern for different operculums, thereby limiting the pitch fluctuation in the subsequently generated pitch sequence to a relatively small extent. The rhythm types of music score may be different, and in the actual rhythm mode node construction, the music score with the same rhythm type is usually selected from the music score set, so that the number difference between the pitch and the rhythm at the later stage is small, and the fusion is well completed.

Drawings

Fig. 1 is a flowchart of a method for generating a new song based on dual-mode network wandering fusion according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of constructing a pitch pattern network according to an embodiment of the present invention;

fig. 3 is a schematic diagram of constructing a rhythm pattern network according to an embodiment of the present invention;

fig. 4 is a graph showing a music score obtained by randomly performing a dual-mode network walk and fusing to generate a new music.

Detailed Description

In order to make the technical means, the creation characteristics, the achievement purposes and the effects of the invention easy to understand, the invention is further described with the specific embodiments.

In the description of the present invention, it should be noted that the terms "upper", "lower", "inner", "outer", "front", "rear", "both ends", "one end", "the other end", and the like indicate orientations or positional relationships based on those shown in the drawings, and are only for convenience of description and simplicity of description, but do not indicate or imply that the referred device or element must have a specific orientation, be constructed in a specific orientation, and be operated, and thus, should not be construed as limiting the present invention. Furthermore, the terms "first" and "second" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.

In the description of the present invention, it is to be noted that, unless otherwise explicitly specified or limited, the terms "mounted," "disposed," "connected," and the like are to be construed broadly, such as "connected," which may be fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.

Referring to fig. 1, the present invention provides a new song generation method based on dual-mode network wandering fusion, including the following steps:

In the step S100, an original music score set is constructed, where the music score set includes a plurality of groups of classical music scores, and a specific classical music score may be selected by a composer from the classical music scores of different countries and different composers, for example, so as to enrich the number of nodes and the migration path for selection in a pitch and rhythm pattern network constructed subsequently; in addition, the music score has diversified existing forms, and the music score in a printing or picture or midi format is selected by a composer; the embodiments of the present disclosure are not limited thereto.

Fig. 2 is a schematic diagram of the pitch pattern network constructed in step S200; referring to fig. 3, a schematic diagram of the rhythm pattern network constructed in step S200 is shown.

In the above step S500, "unrestricted" means that the node to be migrated is allowed to migrate to all pointed nodes, no matter whether the pointed node has migrated or not; unrestricted random walks are well known in the art.

The technical solution of the present invention is illustrated by a detailed example below:

example (b): for example, 3 numbered musical notation can be used for constructing a pitch mode and rhythm mode network, random walk generates a pitch and rhythm sequence, and the pitch and rhythm sequence is fused to generate a new music piece.

1) Constructing an original koji spectrum set;

in this embodiment, 3 well-known classical songs are selected as the music score set, including friendship, chang di chang, safety night, and jubilation. Each numbered musical notation is selected from the book "easy learning numbered notation" written by many three requirements (shanghai music college press, 2014) and is easily obtained from network resources.

2) Constructing a pitch and rhythm mode network;

according to the sequence of the sound heights of the first two bars in the music score 'friendship long-time earth', the middle sound 1 is selected as the reference pitch. The relative pitch of the first two measures is changed to (1-30002101) as a pitch pattern 1 node. And so on to form a pitch mode node set.

According to the music score "friendship long praise", the change of the note duration of the first two bars is (11.50.5111.50.51), which is taken as the rhythm pattern 1 node. Two music scores of the same beat type, namely friendship, everlasting period and jubilation, are selected, and a rhythm mode node set is formed according to the method and the analogy.

Connecting any node pair in the pitch and rhythm mode node set by using one edge to form two fully-connected networks, as shown in fig. 2 and fig. 3.

3) Setting a random walk initial condition;

nodes are randomly selected from the pitch pattern set, and the selected node is a node 16 which is set as a wandering start node and is also an initial current node to be wandered. The walk maximum step size is set to 4. At this time, the value of the step length of the actual wandering is an initial value 0.

And randomly selecting a node from the rhythm mode set, wherein the selected node is a node 5 which is set as a wandering initial node and is also an initial current node to be wandered. The walk maximum step size is set to 4. At this time, the value of the step length of the actual wandering is an initial value 0.

4) Judging whether to execute the wandering at present;

according to the judgment condition, the actual wandering step length in the two networks is 0 and is less than the maximum wandering step length 4. And thus all go to the next step of wandering.

5) Performing wandering and re-judging;

in the pitch pattern network, the nodes are randomly selected from the pitch node set, and the selected node is 5. The initial sequence of the walk is 16, at which point node 5 is added to the sequence of nodes, i.e., 16, 5. The currently wandering source node 16 is updated to 5 and the actual wandering step 0 is updated to 1. And after the current wandering is executed, returning to the step 4), and judging whether to execute new wandering. And executing a new round of wandering and judging according to the above steps until the judging result is false, and finishing wandering. A new round of wandering and judgment in the cadence pattern network is also processed.

6) Restoring a pitch mode sequence and a rhythm mode sequence, and fusing to generate a new song;

finally, the total node sequence that the pitch pattern network has randomly walked is 16, 5, 9, 13. Reducing the pitch sequence to a relative pitch sequence { -2-10-1-2-3-2-6; 242245, respectively; 11321010-2; -3-3-2-1-1-2-2}. Given a new reference pitch of mediant 3, it is translated back to a normal pitch sequence

The sequence of all nodes randomly walked by the rhythm pattern network is {5, 2, 5, 2 }. It is reduced to a sequence of duration values 111110.50.511; 11.50.5113, respectively; 111110.50.511, respectively; 11.50.5113}. Taking two sections as units, corresponding pitch sum in sequenceThe rhythm resynthesizes the notes to produce a new song, as shown in fig. 4, which is an example analysis of the new song based on the dual-mode network wandering fusion.

According to the new music generation method based on the dual-mode network wandering fusion, only random creation and fusion are respectively carried out on the pitch and the duration of notes in manual music composition, two basic construction units with nodes as modes are adopted, continuity and variability of the pitch and the rhythm are considered, and the construction complexity of a mode network is reduced.

Furthermore, the invention sets a reference pitch according to the pitch distributions of the first two measures before constructing a pitch pattern for different operculums, thereby limiting the pitch fluctuation in the subsequently generated pitch sequence to a relatively small extent. The rhythm types of music scores are different, and in the construction of the actual rhythm mode nodes, the music scores with the same rhythm type are usually selected from a music score set, so that the number difference between the pitch and the rhythm at the later stage is small, the fusion is well completed, and the problem that the note sequence generated by the random walk of a note network lacks rhythm sense can be solved.

It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.

Claims

1. A new song generation method based on dual-mode network wandering fusion is characterized by comprising the following steps: the method comprises the following steps:

2. The method for generating a new song based on dual-mode network wandering fusion as claimed in claim 1, wherein: the process of constructing the original music score set in the step S100 includes:

constructing an original koji spectrum set:

M＝{S(k)}|card(M)＝n；k＝1,2,…,n

3. The method according to claim 2, wherein the method comprises: the step S200 includes:

s201: defining an abstract method of pitch and rhythm change mode nodes;

s202: constructing a pitch and rhythm mode network;

4. The method according to claim 3, wherein the method comprises: the step S300 includes:

s301, setting a wandering node;

S302: setting a walking step length;

5. The method according to claim 4, wherein the method comprises: in step S400, determining whether each network currently executes the wandering process includes:

6. The method according to claim 5, wherein the method comprises: the step S500 includes:

randomly selecting one node from the nodes pointed by the current nodes to be walked as a node to be walked, wherein the node is defined as V_next(ii) a Defining the node sequence of the wandering as TV, and setting the initial sequence as S]And adding the wandering node V after each wandering_nextInto a node sequence TV, i.e. TV ═ TV, V_next](ii) a Updating the original node V which is currently walked to V_nextI.e. V ═ V_next(ii) a Will actually wander step length T_vIs updated to T_v+1, i.e. T_v＝T_v+1；