WO2017057532A1

WO2017057532A1 - Instrument type identification device and instrument sound identification method

Info

Publication number: WO2017057532A1
Application number: PCT/JP2016/078754
Authority: WO
Inventors: 慶太有元
Original assignee: ヤマハ株式会社
Priority date: 2015-09-30
Filing date: 2016-09-29
Publication date: 2017-04-06
Also published as: JP6601109B2; JP2017068125A

Abstract

The present invention realizes an instrument type identification device or the like capable of accurately identifying an instrument even when acoustic signals from multiple instruments are inputted. The instrument type identification device is characterized by including: an indicator value acquisition means for acquiring indicator value data comprising an indicator value representing the possibility of corresponding to an instrument type, for each acoustic signal instrument type and acoustic signal characteristic, on the basis of the acoustic signals acquired through a plurality of channels; an inter-channel characteristic information detection means for detecting the characteristics of an acoustic signal between channels as characteristic information, on the basis of an acoustic signal between multiple channels; and a score information generation means for generating, as score information, a value according to the probability of corresponding to the instrument type, for each instrument type in each channel, on the basis of the indicator value data and the characteristic information.

Description

Musical instrument identification device and instrument sound identification method

The present invention relates to a musical instrument identification device and a musical instrument sound identification method.

For example, Japanese Patent Application Laid-Open No. 2013-15601 discloses a sound source identification device that identifies a sound source from sounds emitted from a sound source such as an acoustic instrument, a natural sound, a person, or a living thing. Specifically, the sound source identification device registers the sound feature data emitted from the sound source in a database, and identifies the sound source based on the correlation with the sound feature database emitted from the sound source to be identified. .

However, since the sound source identification device is configured to specify a sound source independently for each target acoustic signal, for example, when acoustic signals from a plurality of musical instruments are input, musical instruments of similar tones The specific accuracy of the sound source may be insufficient.

In view of the above, for example, the present invention realizes a musical instrument identification device and a musical instrument sound identification method capable of identifying a musical instrument with higher accuracy even when acoustic signals from a plurality of musical instruments are input. Objective.

The instrument identification device of the present invention is based on the acoustic signal acquired for each of a plurality of channels, and the index value representing the characteristics of the acoustic signal and the possibility of corresponding to the instrument for each instrument of the acoustic signal Index value acquisition means for acquiring index value data composed of: interchannel feature information detection means for detecting, as feature information, characteristics of the acoustic signal between the channels based on acoustic signals between the plurality of channels; Score information generating means for generating, as score information, a value corresponding to the accuracy corresponding to the instrument for each instrument of each channel based on the index value data and the feature information. Features.

The instrument identification method of the present invention is based on the acoustic signal acquired for each of a plurality of channels, and the index value representing the characteristics of the acoustic signal and the possibility of corresponding to the instrument for each instrument of the acoustic signal. Index value data configured by: obtaining the characteristic information of the acoustic signal between the channels based on the acoustic signal between the plurality of channels as feature information, and based on the index value data and the feature information A value corresponding to the accuracy corresponding to the musical instrument is generated as score information for each musical instrument of each channel.

It is a figure which shows an example of the outline | summary of an acoustic signal processing system. It is a figure which shows an example of the outline | summary of a structure of a mixer. It is a figure which shows an example of a functional structure of the control part of the mixer in 1st Embodiment. It is a figure which shows an example of a structure of an index value acquisition part. It is a figure which shows an example of index value data. It is a figure for demonstrating an example of a process of a threshold value determination / exclusion part. It is a figure for demonstrating an example of the feature detection of the feature information detection part between channels. It is a figure for demonstrating another example of the feature detection of the feature information detection part between channels. It is a figure which shows an example of score information. It is a figure which shows an example of the flow of a process after acquiring the acoustic signal in 1st Embodiment until determining the musical instruments corresponding to each channel. It is a figure which shows an example of a functional structure of the control part of the mixer in 2nd Embodiment. It is a figure which shows an example of combination score information. It is a figure which shows an example of the combination score information extracted by the combination score information extraction part. It is a figure which shows an example of the flow of a process after acquiring the acoustic signal in 2nd Embodiment until determining the musical instruments corresponding to each channel. It is a figure which shows an example of a functional structure of the control part of the mixer in 3rd Embodiment. It is a figure which shows an example of the correlation value data between each channel. It is a figure for demonstrating another example of the process of top determination. It is a figure for demonstrating another example of the process of top determination.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. In addition, about drawing, the same code | symbol is attached | subjected to the same or equivalent element, and the overlapping description is abbreviate | omitted.

[First Embodiment]
FIG. 1 is a diagram illustrating an example of an outline of an acoustic signal processing system according to the present embodiment. As shown in FIG. 1, the acoustic signal processing system 100 includes, for example, a keyboard 101, a drum 102, a guitar 103, a microphone 104, a top microphone 105, a mixer 106, an amplifier 107, and a speaker 108.

The keyboard 101 is, for example, a synthesizer or an electronic piano, and outputs an acoustic signal according to the performance of the performer. For example, the microphone 104 collects a singer's voice and outputs the collected sound as an acoustic signal. The drum 102 includes, for example, a drum set and microphones that collect sounds generated by hitting a percussion instrument (for example, a bass drum or a snare drum) included in the drum set. The microphone is provided for each percussion instrument, and outputs the collected sound as an acoustic signal. The guitar 103 includes, for example, an acoustic guitar 103 and a microphone, and the sound of the acoustic guitar 103 is collected by the microphone and output as an acoustic signal. The guitar 103 may be an electric acoustic guitar or an electric guitar. In that case, there is no need to provide a microphone. The top microphone 105 is a microphone installed above a plurality of musical instruments, for example, a drum set, and collects a sound from the entire drum set and outputs it as an acoustic signal. The top microphone 105 may be composed of a plurality of microphones, for example, installed on the left and right. The top microphone 105 inevitably picks up sounds from musical instruments other than the drum set, though the volume is low.

The mixer 106 has a plurality of input terminals, and electrically adds, processes, and outputs acoustic signals from the keyboard 101, the drum 102, the guitar 103, the microphone 104, and the like input to the input terminals. Specifically, the mixer 106 includes, for example, a level control unit that controls the level of the volume and the like, a pan control unit that changes the sound balance and adjusts the sound localization, and the like. The acoustic signals are mixed by the mixing unit and output to the amplifier 107. The mixer 106 in the present embodiment has a musical instrument identification function and the like in addition to the configuration of the general mixer 106 as described above. Details of the musical instrument identification function and the like will be described later. Further, since the configuration of the general mixer 106 is well known, a detailed description thereof will be omitted.

The amplifier 107 amplifies the acoustic signal output from the output terminal of the mixer 106 and outputs it to the speaker 108. The speaker 108 emits sound according to the amplified acoustic signal.

Next, an example of the configuration of the mixer 106 in the present embodiment will be described. FIG. 2 is a diagram for explaining the outline of the configuration of the mixer 106 in the present embodiment. As illustrated in FIG. 2, the mixer 106 includes, for example, a control unit 201, a storage unit 202, an operation unit 203, a display unit 204, and an input / output unit 205. The control unit 201, the storage unit 202, the operation unit 203, the display unit 204, and the input / output unit 205 are connected to each other via an internal bus 206.

The control unit 201 is, for example, a CPU or MPU, and operates according to a program stored in the storage unit 202. The storage unit 202 is an information recording medium that includes an information recording medium such as a ROM, a RAM, and a hard disk, and holds a program executed by the control unit 201.

The storage unit 202 also operates as a work memory for the control unit 201. The program may be provided by being downloaded through a network (not shown), or provided by various information recording media that can be read by a computer such as a CD-ROM or DVD-ROM. May be.

The operation unit 203 outputs the content of the instruction operation to the control unit 201 in accordance with the user's instruction operation such as a slide-type volume, button, or knob. The display unit 204 is, for example, a liquid crystal display, an organic EL display, or the like, and displays information according to an instruction from the control unit 201.

The input / output unit 205 has a plurality of input terminals and output terminals. Acoustic signals are input to each input terminal from each instrument such as the keyboard 101, drum 102, guitar 103, microphone 104, and top microphone 105. Further, an acoustic signal obtained by electrically adding and processing the input acoustic signal is output from the output terminal. The configuration of the mixer 106 is an example and is not limited to this. For example, some functions such as level control may be processed in an analog manner.

Next, the functional configuration of the control unit 201 of the mixer 106 in the present embodiment will be described with reference to FIG.

The acoustic signal acquisition unit 301 acquires an acoustic signal for each channel. Here, each channel corresponds to each input terminal of each acoustic signal from each instrument such as the keyboard 101, the drum 102, and the guitar 103.

The onset / offset detector 302 extracts the onset and offset from the input acoustic signal. Onset corresponds to the rise of the acoustic signal, and offset corresponds to the output of the acoustic signal being equal to or less than a predetermined value (for example, approximately 0).

The feature amount extraction unit 303 extracts a feature amount of an acoustic signal (hereinafter simply referred to as “sounding section”) between onset and offset. The extraction is performed for each channel and each sounding section.

The index value acquisition unit 304 acquires an index value representing an index indicating which musical instrument is identified for each sound generation section based on the feature amount. The index value is acquired for each channel and each instrument.

Specifically, for example, the index value acquisition unit 304 is configured using three SVMs (Support Vector Vector Machine) as shown in FIG. The three SVMs identify which musical instrument candidate acoustic signal corresponds to the input sound generation section based on the feature amount (for example, feature vector). Specifically, SVM0 is a harmonic sound representing guitar, male, female, female (male Vo, female Vo), or a percussive sound such as snare or cymbal. To identify. The SVM 1 identifies whether the input onset is a percussion instrument such as a kick, snare, or cymbal. The SVM 2 identifies whether the input sound generation section is a bass (Bass), guitar (Guitar), male vocal (maleVo), female vocal (femaleVo), or the like.

Note that the types and number of musical instruments identified by each SVM are merely examples, and the present embodiment is not limited to these. SVM is one of machine learning algorithms, and learns the feature value of the acoustic signal of each instrument beforehand, and based on this, which instrument is the input feature value? Although it is a technique to classify, since it is well known, the description is omitted for details. In this embodiment, the case where SVM is used as an example of classification by a machine learning algorithm will be described. However, other algorithms (for example, simple regression analysis) may be used. Furthermore, although the case where three SVMs are used has been described above, the present embodiment is not limited to this, and may be configured with two SVMs, for example.

FIG. 5 shows an example of index value data composed of the index values acquired as described above. Here, in FIG. 5, onset represents information for identifying the sound generation interval of each input acoustic signal. Onset Time represents the start time (hereinafter referred to as onset time) of the sound generation interval, and Offset Time represents the end time (hereinafter referred to as offset time) of the sound generation interval. amplitude represents the amplitude of the pronunciation interval. Further, Non-Percussive (corresponding to the above harmonic sound) and Percussive represent the output from SVM0, Conga to Kick represent the output from SVM1, and Bass to Wind represent the output from SVM2. In addition, it represents that possibility that it corresponds to the said musical instruments is so high that the numerical value of an index value is large. In FIG. 5, SVM0 is designed to take a numerical value from 0 to 1.

The threshold determination / exclusion unit 305 determines whether or not a sound generation section included in one channel is equal to or less than a predetermined volume threshold. Then, the threshold determination / exclusion unit 305 excludes the information related to the sound generation section determined to be equal to or lower than the predetermined volume from the index value data. Specifically, for example, in the case shown in FIG. 6, since the maximum value of the sound generation section (other source) other than that indicated as Vo. Active is equal to or less than the volume threshold indicated by the dotted line, the pronunciation represented by the other source Exclude sections.

The inter-channel feature information detection unit 306 compares the sound generation sections of the respective channels and detects predetermined feature information. Here, the predetermined feature information is, for example, information indicating a feature that the start timing of the sound generation interval is different from that of other channels. Further, the predetermined feature information may be information indicating a characteristic that the signal level of the sound generation section of the other channel is very small (below a predetermined threshold) as compared with the signal level of the sound generation section of the channel. Specifically, for example, when the sound generation interval of each channel is shown in FIG. 7, the sound generation interval of the channel indicated by 701 in FIG. 7 differs from the other channels in the start timing (onset time) of the sound generation interval. In addition, the signal level of the other channel is very small compared to the signal level of the sounding section.

Further, the predetermined feature information includes, for example, the onset time and the offset time almost simultaneously (within a predetermined threshold range) across a plurality of channels, and the index value for the sound generation section of the sound generation channels is the highest. The same musical instrument may be used. Specifically, for example, as shown in the part indicated by 801 in FIG. 8, the feature information has the same onset time and offset time for the three channels at the same time, and the Snare index value for the three channels indicates the sound generation interval. This is characteristic information indicating the highest value. Note that the portion represented by 802 indicates the largest index value among the three channels.

The score information generation unit 307 generates score information based on the index value data and the detected feature information. Specifically, for example, as shown in FIG. 4, after multiplying the output of SVM1 and the output of SVM2 by the output identified as the harmonic sound and the output identified as the percussion instrument sound, respectively, among the outputs of SVM0, Add to generate score information. In this case, each output from the SVM 1 is in the range of 0 to 1, and the output from the SVM 0 of the SVM 1 and SVM 2 is weighted and multiplied. Note that the index value data corresponds to the index value data from which the onset is excluded when the onset below the threshold is included as described above.

Here, the score information generation unit 307 generates score information while adjusting the degree of contribution to the score information of each pronunciation section according to the detected feature information. Specifically, for example, the predetermined feature information is information indicating that the start timing of the sound generation interval is different from that of the other channel, or the signal level of the sound generation interval of the other channel is the signal level of the sound generation interval of the channel. In the case of information indicating a feature that is very small compared to, the adjustment is made so as to increase the contribution of the pronunciation section. In addition, the predetermined feature information is, for example, when the onset time and the offset time are almost the same over a plurality of channels and the musical instruments having the largest index value for the sound generation section are the same. Make adjustments to reduce the contribution of the pronunciation interval in the channel. In addition, for example, when the start times of the sound generation intervals are almost the same, the contribution may be adjusted so as to reduce the contribution other than the sound generation interval of the earliest onset time. In the above description, it is assumed that the higher the contribution, the higher the score information.

The score information generated in this way is shown in FIG. As shown in FIG. 9, in the score information, a score indicating that each instrument is for each channel is indicated by a numerical value. In other words, the larger the value, the more likely that the channel is the instrument. That is, the score information is a value corresponding to the accuracy corresponding to the musical instrument.

The reliability level acquisition unit 308 acquires the reliability level of each channel. The reliability is acquired based on, for example, the consistency (variance) of each index value in the index value data, the number of sounding sections, the average sound volume of all sounding sections, and the like. Specifically, for example, when the number of sounding intervals included in the index value data is small, the reliability is lowered, for example, the reliability is increased as the average sound volume of all sounding intervals is larger than the average sound volume of other channels. is there. Note that the above assumes that the microphone gains of all channels are uniform. Further, the acquisition of the reliability is an example, and the present embodiment is not limited to the above.

The instrument determination unit 309 determines instruments for each channel based on the reliability and the score. Specifically, for example, first, among the channels for which musical instruments have not been determined, the musical instrument having the highest reliability and the highest score is selected and determined to correspond to the musical instrument of the channel. Next, the instrument having the highest score in the channel with the second highest reliability is determined as the instrument of the channel. Similarly, the instruments for each channel are determined.

Note that the instrument determining unit 309 may be configured to determine whether or not the user violates a predetermined restriction and to determine according to the determination result. Specifically, for example, when it is determined that the restriction is violated, the score of the determined instrument is set to 0, and the channel is selected in the same manner as described above. Note that the constraint is a constraint input by the user, for example, there is only one drum 102, up to two guitars 103, and no female vocals.

The image information generation unit 310 generates image information representing each musical instrument corresponding to each channel and displays it on the display unit 204. Note that when the musical instrument corresponding to the channel cannot be determined, a message indicating that the musical instrument cannot be determined for the channel may be displayed.

Next, an example of a processing flow from acquisition of an acoustic signal according to the present embodiment to determination of musical instruments corresponding to each channel will be described with reference to FIG. As shown in FIG. 10, first, the acoustic signal acquisition unit 301 acquires an acoustic signal for each input channel (S101). The onset / offset detection unit 302 extracts the onset and offset from the input acoustic signal (S102). The feature amount extraction unit 303 extracts the feature amount of the pronunciation section (S103). The index value acquisition unit 304 acquires an index value indicating which musical instrument is estimated for each sound generation section based on the feature amount (S104). The threshold determination / exclusion unit 305 determines whether or not a sound generation section included in one channel is equal to or less than a predetermined volume threshold (S105). The threshold determination / exclusion unit 305 excludes the information related to the sound generation section determined to be equal to or lower than the predetermined volume from the index value data (S106). The inter-channel feature information detection unit 306 compares the sound generation sections of the respective channels and detects predetermined feature information (S107). The score information generation unit 307 generates score information based on the index value data and the detected feature information (S108).

The reliability acquisition unit 308 acquires the reliability of each channel (S109). The instrument determination unit 309 first determines whether there is a channel for which the corresponding instrument has not been determined (S110). If it is determined that there is an undetermined channel, the instrument with the highest reliability and the highest score is selected from the undetermined channels (S111). Then, the process returns to S110. On the other hand, if it is determined that there is no undetermined channel, the process ends. The above process is an example, and the present embodiment is not limited to the above flow.

According to the present embodiment, it is possible to realize a musical instrument identification device or the like that can identify musical instruments more accurately even when acoustic signals from a plurality of musical instruments are input.

The present invention is not limited to the above-described embodiment. For example, the configuration substantially the same as the configuration shown in the above-described embodiment, the configuration having the same operational effects, or the same object can be achieved. It can be replaced with a possible configuration.

[Second Embodiment]
Next, a second embodiment of the present invention will be described. In the present embodiment, as shown in FIG. 11, the points having the combination score information acquisition unit 311 and the combination score information extraction unit 312 and the processing of the reliability acquisition unit 308 are mainly the first implementation. Different from form. Note that the description of the same points as in the first embodiment will be omitted below.

The combination score information acquisition unit 311 covers combinations of instruments for each channel, and acquires the total score based on the score information acquired by the score information acquisition unit. Specifically, for example, as shown in FIG. 12, the total score is obtained for every combination of musical instruments. In FIG. 12, the combination of musical instruments represented by combi1 indicates that channel 1 is kick, channel 2 is snare, etc., and the total score is 33.52. Note that the combination score information acquisition unit 311 may be configured to exclude combinations that do not satisfy the constraints given by the user. In addition, combi1 etc. represent each combination of musical instruments.

The combination score information extraction unit 312 extracts a predetermined number of combination score information in descending order of the total score. For example, FIG. 13 shows a case where five pieces of combination score information are extracted from a higher total score.

The reliability acquisition unit 308 acquires the reliability of each channel based on the extracted combination score information. Specifically, for example, the reliability obtaining unit 308 obtains the reliability of each channel based on whether or not the same instrument is stably selected in a higher order. More specifically, for example, Kick and Bass are all selected for channel 1 (ch1) and channel 5 (ch5), while Hi-Hat is selected for channel 3 (ch3) from combi1 to combi4. Therefore, the stability of the

channels

1 and 5 is configured such that a higher reliability than that of the channel 3 is obtained.

Next, the instrument determining unit 309 determines instruments corresponding to each channel in the order of the acquired reliability. Specifically, for example, in the case shown in FIG. 13, since the reliability of

channels

1 and 5 is the same, it is determined that

channels

1 and 5 correspond to Kick and Bass, respectively. When the reliability is the same, the order of determination may be any. Next, for channel 3, Hi-Hat is selected from combi1 to combi4, and since channel 3 is called Hi-Hat because it has higher stability and higher reliability than other undecided channels. decide. In the same manner, instruments corresponding to each channel are determined.

In the present embodiment, the determination of musical instruments may be suspended for channels for which the same musical instrument is not stably selected, that is, for channels whose reliability is a predetermined threshold value or less. Specifically, for example, in the case shown in FIG. 13,

channels

4 and 7 are put on hold because the selected musical instruments are unstable. In this case, after the user confirms and corrects the musical instruments, the combination score information is acquired only for the reserved channels, and the musical instruments corresponding to the reserved channels may be determined in the same manner as described above. .

Next, an example of a processing flow from acquisition of an acoustic signal in this embodiment to determination of musical instruments corresponding to each channel will be described with reference to FIG.

First, S201 to S208 are the same as S101 to S108 of the first embodiment, and thus description thereof is omitted. Next, the combination score information acquisition unit 311 covers combinations of musical instruments for each channel, and acquires the total score based on the score information acquired by the score information acquisition unit (S209). The combination score information extraction unit 312 extracts a predetermined number of combination score information in descending order of the total score (S210). The instrument determining unit 309 first determines whether there is a channel for which the corresponding instrument has not been determined (S211). If it is determined that there is an undetermined channel, the instrument having the highest reliability among the undetermined channels is determined (S212). Then, the process returns to S211. On the other hand, if it is determined that there is no undetermined channel, the process ends.

According to the present embodiment, as in the first embodiment, for example, it is possible to identify a musical instrument organization with higher accuracy than when identifying musical instruments for each channel. For example, it is possible to more easily determine from which device an acoustic signal is input. In addition, according to the present embodiment, it is possible to identify a musical instrument organization with higher accuracy than in the first embodiment.

The present invention is not limited to the above-described embodiment. For example, the configuration substantially the same as the configuration shown in the above-described embodiment, the configuration having the same operational effects, or the same object can be achieved. It can be replaced with a possible configuration. For example, the acquisition of reliability may be used in combination with the acquisition of reliability in the first embodiment.

[Third Embodiment]
Next, a third embodiment of the present invention will be described. As shown in FIG. 15, the present embodiment is different from the first embodiment in that a correlation value acquisition unit 313, a correlation value addition unit 314, and a top microphone determination unit 315 are included. Note that the description of the same points as in the first embodiment will be omitted below.

Correlation value acquisition unit 313 acquires a correlation value based on the correlation of acoustic signals between channels. Specifically, for example, correlation value data between channels as shown in FIG. 16 is acquired. The correlation value adding unit 314 adds the correlation values for each channel to obtain a total value.

The top microphone determination unit 315 determines the channel of the top microphone 105 based on the total value. Specifically, for example, normally, two top microphones 105 are arranged on the left and right, so that the two channels with the maximum total value are determined to be the top microphones 105. Specifically, in the case shown in FIG. 16, since the sum of

channels

3 and 4 is the maximum, it is determined that

channels

3 and 4 are top microphones. In addition, it may be configured such that the sum of the times when the onset and offset times overlap between channels is obtained, and the top microphone 105 is determined based on the time.

If a pair of channels having the same onset and offset times are detected, the channel of the top microphone 105 is determined based on the time difference from the onset time of other channels and the volume difference. You may comprise. Specifically, for example, when drums such as Kick and Snare are determined according to the first embodiment, the volume of each channel is subtracted from the volume of one of the drums. The state in that case is shown in FIG. Then, two channels are determined as the top microphone 105 in order from the channel representing the lowest volume (negative value and the largest absolute value) compared to the volume of the channel of the drum.

Further, for example, as shown in FIG. 18, when drums such as Kick and Snare are determined according to the first embodiment, the onset time of each channel is set to one of the drums. Subtract from the onset time. Further, the top microphone 105 may be configured to determine two channels that represent the onset time (positive value and the largest absolute value) that is delayed compared to the onset time of the channel of the drum type. Good. FIGS. 17 and 18 show an example in which each channel is not determined. For example, according to the first embodiment, at least one of the drums is determined. Assumption.

The image information generation unit 310 generates image information representing each instrument corresponding to each channel and displays it on the display unit. Here, each musical instrument includes a top microphone 105.

According to the present embodiment, for example, the channel corresponding to the top microphone can be determined with higher accuracy than in the first and second embodiments.

The present invention is not limited to the above-described embodiment, and is substantially the same configuration as the configuration shown in the above-described embodiment, a configuration that exhibits the same operational effects, or a configuration that can achieve the same purpose. Can be replaced.

For example, in the above description, the case where there are mainly two top microphones 105 has been described as an example, but the number of top microphones 105 is not limited to two, and may be one or three or more. In the above description, the case of combining the first embodiment and the configuration for determining the channel corresponding to the top microphone 105 has been described as an example. However, the configuration may be combined with the second embodiment, and the top microphone 105 may be combined. You may comprise so that only the structure which determines a corresponding channel may be implement | achieved independently.

Further, the third embodiment may be configured in combination with the first or second embodiment. Here, for example, with respect to the snare, all the sound generation sections exist at almost the same timing as the top microphone, the volume (amp) is larger than the other channels, and Onset time is faster. Thus, the contribution degree of the sound generation section corresponding to the snare may be reduced. In this case, for example, the contribution ratio may be a volume ratio with the sound generation section having the maximum volume in the onset at the same timing in other channels.

Further, in the first to third embodiments, the case where the instrument identification device is realized as the mixer 106 has been mainly described. However, the instrument identification device may be formed separately from the mixer 106 or other acoustic device. It may be realized within.

In the above description, the configuration for determining musical instruments based on the reliability has been described. However, the configuration is such that the musical instruments are determined based on the score information and the combination score information without using the reliability. Also good.

Further, in the above description, the configuration for performing processing such as acquisition of an index value and acquisition of feature information based on the characteristics of the sounding section of the acoustic signal has been described. However, acquisition of the index value and feature information based on the characteristics of the acoustic signal are described. Any other configuration may be used as long as it performs processing such as acquisition.

Claims

Based on the acoustic signal obtained for each of a plurality of channels, index value data constituted by an index value representing the characteristic of the acoustic signal and the possibility of corresponding to the instrument is obtained for each instrument of the acoustic signal. Index value acquisition means to
Inter-channel feature information detecting means for detecting, as feature information, characteristics of the acoustic signal between the channels based on acoustic signals between the plurality of channels;
Score information generating means for generating, as score information, a value corresponding to the accuracy corresponding to the musical instrument for each musical instrument of each channel based on the index value data and the feature information;
A musical instrument identification device comprising:
2. The musical instrument identification apparatus according to claim 1, wherein the index value data is configured based on a sound generation section in which a volume of an acoustic signal included in each channel is larger than a predetermined threshold.
The index value acquisition means provides a first SVM that provides an index value indicating whether the acoustic signal is a harmonic sound or a percussion instrument sound, and an index value that indicates which musical instrument the acoustic signal is a harmonic sound of. The musical instrument identification device according to claim 1, further comprising: a second SVM, and a third SVM that provides an index value indicating which percussion instrument sounds the acoustic signal is.
The index value acquisition means is based on a sum of values obtained by weighting the index value from the second SVM and the index value from the third SVM based on the index value from the first SVM. 4. The musical instrument identification device according to claim 3, wherein index value data is acquired.
The characteristic information of the acoustic signal between the channels is characteristic information based on a start time or a signal level of a sound generation period of the acoustic signal between the channels. Instrument identification device.
Reliability acquisition means for acquiring a predetermined reliability of each channel based on the index value;
Instrument determining means for determining instruments corresponding to each channel based on the reliability and the score information;
The musical instrument identification device according to claim 1, comprising:
The instrument identification device according to claim 6, wherein the reliability is acquired based on a variance of each index value in the index value data, the number of sounding sections, or an average volume of all sounding sections. .
Based on the score information, combination score information acquisition means for acquiring, for each combination of each instrument, combination score information representing an index corresponding to each instrument combination of each channel;
Combination score information extracting means for extracting a predetermined number of combination score information in descending order of the index of the combination score information;
A reliability acquisition means for acquiring a predetermined reliability of each channel based on the extracted combination score information;
Musical instrument determination means for determining musical instruments corresponding to each channel based on the combination score information and the reliability;
The musical instrument identification device according to claim 1, comprising:
The instrument identification device according to claim 8, wherein the reliability is based on whether or not the same instrument is more stably selected in the extracted combination information.
Correlation value acquisition means for acquiring a correlation value based on the correlation of the acoustic signal between the channels based on the acoustic signal acquired for each channel;
Top microphone identifying means for identifying a channel corresponding to a top microphone arranged to collect sound from a plurality of musical instruments based on the correlation value;
10. The musical instrument identification device according to claim 1, further comprising:
The said top microphone identification means identifies the channel corresponding to the said top microphone based on the relationship with the start or the end time of the sound generation area of the acoustic signal acquired for every said channel. Musical instrument identification device.
Based on the acoustic signal obtained for each of a plurality of channels, index value data constituted by an index value representing the characteristic of the acoustic signal and the possibility of corresponding to the instrument is obtained for each instrument of the acoustic signal. And
Based on the acoustic signals between the plurality of channels, the characteristics of the acoustic signals between the channels are detected as feature information,
Based on the index value data and the feature information, for each instrument of each channel, a value corresponding to the accuracy corresponding to the instrument is generated as score information.
A method for identifying musical instruments, comprising: