Musical piece analysis device, musical piece analysis method, and program

Abstract

Provided is a musical piece analysis device comprising a processor that executes: processing for detecting a first beat count, which is the count of a beat position at which in a musical piece, the level thereof first exceeds a threshold, with respect to a temporary bar position; processing for detecting a second beat count, which is the count of a beat position immediately after at least one break section in the musical piece with respect to the temporary bar position; and processing for specifying a bar position such that the first beat count is 1 regardless of the temporary bar position when the first beat count and the second beat count match, and specifying the temporary bar position as the bar position when the first beat count and the second beat count do not match.

Description

Music analysis device, music analysis method, and program 【0001】 This invention relates to a music analysis device, a music analysis method, and a program. 【0002】 Techniques for detecting beat positions and other information from the audio signals of a musical piece are known. For example, Patent Document 1 describes a technique for analyzing an audio signal to detect beat positions and the probability of existence of each instrument sound. Patent Document 2 describes a technique for detecting the provisional beat position and the sounding position of a snare drum in musical data, and if the sounding position of the snare drum is not the provisional beat position, setting the beat position based on the sounding position of the snare drum. 【0003】 Japanese Patent Publication No. 2010-134231, International Publication No. 2019 / 058543 【0004】 The technologies described above analyze the entire audio signal of a song to identify beat positions and, based on those beat positions, bar positions. For example, by using machine learning to analyze the audio signal, the accuracy of identifying beat positions can be improved. However, regarding bar positions, there have been cases where incorrect bar positions were identified, such as when the third beat in a 4 / 4 time signature was mistakenly identified as the first beat. 【0005】 Therefore, the present invention aims to provide a music analysis device, a music analysis method, and a program that can more accurately identify bar positions in audio data of a music piece in which beat positions have been identified. 【0006】According to one aspect of the present invention, a music analysis device is provided that includes a processor that performs the following steps: a process of detecting a first beat count, which is a count of the provisional bar position at the beat position in the music where the level first exceeds a threshold; a process of detecting a second beat count, which is a count of the provisional bar position at the beat position immediately following at least one break section in the music; and a process of identifying a bar position such that the first beat count becomes 1 regardless of the provisional bar position if the first beat count and the second beat count match, and identifying the provisional bar position as the bar position if the first beat count and the second beat count do not match. According to another aspect of the present invention, a method for analyzing a musical piece is provided in which a processor performs the following steps: detecting a first beat count which is a count relative to a provisional bar position of the beat position in the musical piece where the level first exceeds a threshold; detecting a second beat count which is a count relative to the provisional bar position of the beat position immediately following at least one break section in the musical piece; and, if the first beat count and the second beat count match, identifying a bar position such that the first beat count becomes 1 regardless of the provisional bar position, and if the first beat count and the second beat count do not match, identifying the provisional bar position as the bar position. In yet another aspect of the present invention, a program is provided for a computer processor to perform the following steps: a process of detecting a first beat count, which is a count of the provisional bar position at the beat position in the musical piece where the level first exceeds a threshold; a process of detecting a second beat count, which is a count of the provisional bar position at the beat position immediately following at least one break section in the musical piece; and a process of identifying a bar position such that the first beat count becomes 1 regardless of the provisional bar position if the first beat count and the second beat count match, and identifying the provisional bar position as the bar position if the first beat count and the second beat count do not match. 【0007】This figure shows an example of the overall configuration of a system according to an embodiment of the present invention. This is a block diagram showing the schematic functional configuration of the music analysis device in the example of Figure 1. This is a flowchart showing an example of the processing of the music analysis device. This is a diagram conceptually showing an example of a song in an embodiment of the present invention. This is a magnified view of part A shown in Figure 4. This is a magnified view of part B shown in Figure 4. This is a magnified view of part C shown in Figure 4. This is a magnified view of part D shown in Figure 4. This is a magnified view of part E shown in Figure 4. This is a magnified view of part F shown in Figure 4. This is a diagram showing another example of the part where the song ends. This is a diagram showing yet another example of the part where the song ends. 【0008】 Figure 1 shows an example of the overall configuration of a system according to an embodiment of the present invention. The system 10 according to this embodiment includes a PC (Personal Computer) 100, a DJ controller 200, and a speaker 300. The PC 100 is a device that stores, processes, and plays back audio data, and is not limited to a PC; it may also be a terminal device such as a tablet or smartphone. The PC 100 includes a display 101 that displays information to the user, and an input device such as a touch panel or mouse that acquires user operation input. The DJ controller 200 is connected to the PC 100 via a communication means such as USB (Universal Serial Bus), and acquires user operation input related to music playback via channel faders, crossfaders, performance pads, jog dials, and various knobs and buttons. Audio data is played back using, for example, the speaker 300. 【0009】In this embodiment, the PC 100 functions as a music analysis device in the system 10 described above. For example, the PC 100 performs processing on the stored audio data in response to user input during playback of the audio data. Alternatively, the PC 100 may perform processing on the audio data before playback and save the processed audio data. In this case, the DJ controller 200 and speakers 300 do not need to be connected to the PC 100 at the time the processing is performed. In this embodiment, the PC 100 functions as a music analysis device, but in other embodiments, DJ equipment such as a mixer or an all-in-one DJ system (digital audio player with communication and mixing functions) may function as a music analysis device. Furthermore, a server connected to the PC and DJ equipment via a network may function as a music analysis device. 【0010】 Figure 2 is a block diagram showing the schematic functional configuration of the music analysis device in the example shown in Figure 1. The PC 100, which functions as a music analysis device, is a computer equipped with a processor such as a CPU (Central Processing Unit) or a DSP (Digital Signal Processor). The processor performs level increase beat count detection processing 120, break end beat count detection processing 130, and measure position identification processing 140 according to the program. The processor may further perform music end beat count detection processing 160 and beat position / temporary measure position identification processing 170. The program is read from the storage of the PC 100 or from a tangible and non-temporary medium such as a removable recording medium, or downloaded from a server via a network and deployed into the memory of the PC 100. 【0011】In this specification, the beat count indicates the position of each beat within a measure. In the case of 4 / 4 time, the beat counts are 1, 2, 3, and 4 from the first beat of the measure, and the beat following the beat with beat count 4, i.e., the first beat of the next measure, has a beat count of 1. In this embodiment, the music is, for example, in 4 / 4 time and the beat counts are one of {1, 2, 3, 4}, but it is not limited to this example. Also, the beat positions and provisional measure positions have already been identified by a device other than the PC 100, or by another process performed by the PC 100, and are associated with the music audio data 110 as metadata, except when the beat position / provisional measure position identification process 170 described later is executed. In the level rise beat count detection process 120, the break end beat count detection process 130, and the music end beat count detection process 160, the first to third beat counts are detected based on the music audio data 110 and the already identified beat positions and provisional measure positions. 【0012】 The level-increase beat count detection process 120 analyzes the music audio data 110 and detects a first beat count, which is a count relative to the provisional measure position of the beat position where the level first exceeds the threshold in the music. For example, in a music piece where the level is low at the beginning and increases later, beats and measures may not be clearly defined while the level is low, but become clear after the level increases. In such cases, the beat position where the level first exceeds the threshold in the music may be the first beat position of the actual measure. In the case of a music piece that starts at a high level from the beginning, the first beat count may be detected at the first beat position of the music. 【0013】The break end beat count detection process 130 analyzes the music audio data 110 to detect a second beat count, which is the count of the beat position immediately following at least one break section in the music relative to the provisional measure position. A break section can be identified, for example, as a section in which the level is significantly lower than the sections before and after it. The beat position at which the level increases again after the break section ends is the beat position immediately following the break section. This beat position, like the beat position where the level first exceeds the threshold in the music, may also be the first beat position of the actual measure. If there are multiple break sections in the music, the counts of the beat positions immediately following each of the multiple break sections are detected, and the mode among them is taken as the second beat count. 【0014】 The measure position identification process 140 determines whether to correct the already identified provisional measure position based on the first beat count detected by the level rise beat count detection process 120 and the second beat count detected by the break end beat count detection process 130. Specifically, if the first beat count and the second count match, the measure position identification process 140 identifies the measure position such that the first beat count becomes 1, regardless of the provisional measure position. In this case, if the first beat count is 1, the provisional measure position becomes the measure position as is; otherwise, the provisional measure position is shifted forward by (first beat count - 1) to become the measure position. If the first beat count and the second beat count do not match, the measure position identification process 140 identifies the provisional measure position as the measure position without correction. The measure position identification process 140 outputs the identified measure position data 150. 【0015】 As will be described later, if the music end beat count detection process 160 is executed, a condition based on the beat count of the beat position where the music ends may be added to the determination of the measure position identification process 140. However, even in this case, the provisional measure position is corrected only if at least the first beat count and the second count match, and the provisional measure position is not corrected if the first beat count and the second beat count do not match. 【0016】As already mentioned, the beat position where the level first exceeds the threshold in a piece of music, or the beat position immediately following a break section, may be the actual first beat position of the measure. However, since techniques such as anacrusis and syncopation may be used, the beat position where the level first exceeds the threshold in a piece of music is not always the actual first beat position of the measure. Therefore, in this embodiment, if the first beat count and the second beat count match as described above, and their beat counts are not 1, that is, if the beat position where the level first exceeds the threshold in a piece of music and the beat position immediately following a break section both have the same beat count, but their beat positions do not coincide with the provisional measure position, the provisional measure position is corrected. This allows for more accurate identification of the measure position. 【0017】 The music end beat count detection process 160 detects a third beat count, which is a count of the beat position where the music ends relative to the provisional measure position. The beat position where the music ends is, for example, the beat position just before the level falls below a threshold at the end of the music. Even when the music fades out, the beat position just before the level falls below a threshold can be detected, but since this beat position is not meaningful in determining the measure position, the music end beat count detection process 160 outputs information indicating that the end of the music is a fade-out, along with the third beat count, or instead of the third beat count. Alternatively, if the end of the music is a fade-out, which has been detected in advance and associated with the music audio data 110 as metadata, the music end beat count detection process 160 does not need to be executed. 【0018】When the above music end beat count detection process 160 is executed, the measure position identification process 140 determines whether the first beat count detected by the level increase beat count detection process 120, the second beat count detected by the break end beat count detection process 130, and the third beat count detected by the music end beat count detection process 160 match, or whether the first beat count, the second beat count, and the next beat count after the third beat count match. If they match, regardless of the temporary measure position, the measure position is identified such that the first beat count becomes 1. Otherwise, the measure position identification process 140 identifies the temporary measure position as the measure position. Note that when the end of the music is a fade-out as described above, the third beat count is not detected or not used for identifying the measure position. 【0019】 Except when the end of the music is a fade-out, it is common for the ending beat of the music to be the first beat or the last beat of a measure. Therefore, in order to further improve the accuracy of identifying the measure position, the third beat count detected by the music end beat count detection process 160 may be used for determining the measure position. 【0020】 The beat position / temporary measure position identification process 170 is a process of analyzing the music audio data 110 to identify the beat position and the temporary measure position. The beat position can be identified, for example, by detecting the pronunciation position of the bass drum or the position with high acoustic energy from the waveform of the audio data. The temporary measure position can be identified, for example, from the pattern of the strength and weakness of the pronunciation at each beat position. Note that the methods for identifying the beat position and the temporary measure position are not limited to the above examples, and various known methods can be used. Machine learning may also be used. As already described, in this embodiment, the beat position and the temporary measure position of the music may already be identified and associated with the music audio data 110 as metadata. Therefore, the beat position / temporary measure position identification process 170 does not necessarily have to be executed by the processor of the PC 100. 【0021】Figure 3 is a flowchart showing an example of the processing of the music analysis device. First, the beat position and the provisional measure position of the music are acquired (step S101). As described above, the beat position / provisional measure position identification process 170 may be executed to analyze the music audio data 110 to identify the beat position and the provisional measure position, or the beat position and the provisional measure position that have already been identified and associated with the music audio data 110 as metadata may be read. Hereinafter, step S102, step S103, and steps S104 and S105 are executed in any order. In step S102, the level-up beat count detection process 120 detects the first beat count, that is, the first level-up beat count n of the music Ｓ In step S103, the music end beat count detection process 160 detects the third beat count, that is, the music end beat count n of the music Ｅ In step S104, the break end beat count detection process 130 detects the beat count n' at the end of the break Ｂ When there are a plurality of break sections in the music, the count of the beat position immediately after each of the plurality of break sections is detected as the beat count n' Ｂ In step S105, the most frequent value n of the beat count n' Ｂ is determined. When there is only one break section, the beat count n' of that section Ｂ becomes the most frequent value n Ｂ becomes the most frequent value n Ｂ When the end of the music is not a fade-out (NO in step S106), the measure position is specified using the beat counts n 【0022】 detected in the processing so far. Specifically, when n Ｓ , n Ｅ , n Ｂ are equal (YES in step S107), the beat of the beat count n Ｓ =(n Ｅ +1)=n Ｂ or n Ｓ =n Ｅ =n Ｂ is set as the measure position (step S108). That is, the measure position is specified so that the beat count n Ｓ becomes 1. Note that n Ｓ becomes 1. Note that n Ｅ+1 is the beat count n Ｅ means the next count, and when the beat count n Ｅ is the maximum count (4 for a quadruple time), n Ｅ +1 becomes 1. n Ｓ =(n Ｅ +1)=n Ｂ but even if n Ｓ =n Ｅ =n Ｂ is not the case (NO in step S107), the temporary measure position is set as the measure position (step S109). When the end of the music is a fade-out (YES in step S106), the beat count n Ｅ at the end of the music is not used, and the beat count n Ｓ , n Ｂ is used to specify the measure position. Specifically, when n Ｓ =n Ｂ is the case (YES in step S110), the beat of the beat count n Ｓ is set as the measure position (step S108), and if not (NO in step S110), the temporary measure position is set as the measure position (step S109). 【0023】 Fig. 4 is a diagram conceptually showing an example of a music in an embodiment of the present invention. In the music MP, after the level rises after the start of the music, there are five break intervals BR1, BR2, BR3, BR4, BR5. In each break interval, the level is significantly lower than the intervals before and after it. 【0024】 Figs. 5A to 5F are diagrams respectively showing the enlarged parts A to F shown in Fig. 4. Fig. 5A shows the beginning part of the music. In the illustrated example, since the music starts at a large level from the beginning, there is a beat of the level rise beat count n Ｓ at the first beat position of the music. Figs. 5B to 5E show the end parts of the second to fifth break intervals. The beat count n Ｂ ' - 2, n Ｂ ' - 3, n Ｂ ' - 4, n ＢFigure 5F shows the point where the song ends. In the illustrated example, the song ends on the last beat of the measure at the temporary measure position TB. In this case, the song end beat count n Ｅ The beat is the original last beat of the measure, and in the judgment of step S107 shown in Figure 3, n Ｓ = (n Ｅ +1) = n Ｂ It could become that. 【0025】 Figure 6 shows another example of the end of a piece of music. In the illustrated example, the piece ends on the first beat of the measure at the provisional measure position TB. In this case, the beat count for the end of the piece is n. Ｅ The beat is the original first beat of the measure, and in the determination of step S107 shown in Figure 3, n Ｓ = n Ｅ = n Ｂ It could become that. 【0026】 Figure 7 shows yet another example of the ending of a song. In the illustrated example, the song ends with a fade-out (F.O.). In this case, the song end beat count n can be used to determine the measure position. Ｅ It is not detected. 【0027】 According to the embodiments of the present invention described above, the process involves detecting a first beat count, which is a count relative to the provisional bar position of the beat position where the level first exceeds a threshold in the music; detecting a second beat count, which is a count relative to the provisional bar position of the beat position immediately following at least one break section in the music; and, if the first beat count and the second beat count match, identifying the bar position so that the first beat count becomes 1 regardless of the provisional bar position, and if the first beat count and the second beat count do not match, identifying the provisional bar position as the bar position. By performing these processes, it is possible to identify cases where the beat position of the first level increase in the music and the beat position of the level increase after the end of the break section both have the same beat count, but these beat positions do not coincide with the provisional bar position, and to correct the provisional bar position to more accurately identify the bar position. 【0028】Furthermore, a process is executed to detect a third beat count, which is a count relative to the temporary measure position at the beat position where the song ends. If the first beat count, the second beat count, and the third beat count match, or if the beat count following the first beat count, the second beat count, and the third beat count match, the measure position is determined so that the first beat count becomes 1 regardless of the temporary measure position. Otherwise, the temporary measure position may be determined as the measure position. If the song ends with a fade-out, the third beat count does not need to be used to determine the measure position. The accuracy of determining the measure position can be further improved by also using the beat position where the song ends in the determination. 【0029】 The aforementioned at least one break section may include multiple break sections, and the process for detecting the second beat count may include detecting the mode of the count of the beat position immediately following each of the multiple break sections relative to the provisional measure position as the second beat count. If there are multiple break sections in a musical piece, the measure position can be appropriately identified by statistically processing them. 【0030】 10...System, 101...Display, 110...Music audio data, 120...Level increase beat count detection process, 130...Break end beat count detection process, 140...Measure position identification process, 150...Measure position data, 160...Music end beat count detection process, 170...Beat position / temporary measure position identification process, 200...DJ controller, 300...Speaker.

Claims

1. A music analysis device comprising a processor that performs the following processes: detecting a first beat count, which is a count of the provisional bar position at the beat position in the music where the level first exceeds a threshold; detecting a second beat count, which is a count of the provisional bar position at the beat position immediately following at least one break section in the music; and, if the first beat count and the second beat count match, identifying a bar position such that the first beat count becomes 1 regardless of the provisional bar position, and if the first beat count and the second beat count do not match, identifying the provisional bar position as the bar position.

2. The music analysis device according to claim 1, wherein the processor further performs a process to detect a third beat count, which is a count of the beat position where the music ends relative to the provisional measure position, and if the first beat count, the second beat count, and the third beat count match, or if the beat count following the first beat count, the second beat count, and the third beat count match, the measure position is identified such that the first beat count becomes 1 regardless of the provisional measure position, and otherwise the provisional measure position is identified as the measure position.

3. The music analysis device according to claim 2, wherein if the end of the music is a fade-out, the third beat count is not used to identify the measure position.

4. The music analysis apparatus according to claim 1, wherein the at least one break section includes a plurality of break sections, and the process for detecting the second beat count includes detecting the mode of the count of the beat position immediately following each of the plurality of break sections relative to the provisional measure position as the second beat count.

5. A music analysis method in which a processor performs the following steps: detecting a first beat count which is a count relative to the provisional bar position of the beat position in the music where the level first exceeds a threshold; detecting a second beat count which is a count relative to the provisional bar position of the beat position immediately following at least one break section in the music; and if the first beat count and the second beat count match, identifying a bar position such that the first beat count becomes 1 regardless of the provisional bar position, and if the first beat count and the second beat count do not match, identifying the provisional bar position as the bar position.

6. A program to cause a computer processor to execute the following processes: detecting a first beat count, which is a count of the provisional bar position at the beat position in the music where the level first exceeds a threshold; detecting a second beat count, which is a count of the provisional bar position at the beat position immediately following at least one break section in the music; and, if the first beat count and the second beat count match, identifying a bar position such that the first beat count becomes 1 regardless of the provisional bar position, and if the first beat count and the second beat count do not match, identifying the provisional bar position as the bar position.

Images