JPH07121556A - Musical information retrieving device - Google Patents

Abstract

PURPOSE:To provide a musical information retrieving device for retrieving required musical information from a database by using a part of musical melody as a retrieving condition. CONSTITUTION:This musical information retrieving device has an input means 110 for inputting a part of the musical melody, a conversion means 120 for selecting frequency indicating a spectrum whose intensity is maximum in a frequency range corresponding to the range of a prescribed scale out of spectra obtained by Fourier-transforming the inputted part of the musical melody and encoding the inputted part of the melody to a scale corresponding to the frequency to generate a scale string, a compensating means 130 for comprensating a deviation between the absolute components of respective scales in the generated scale string, a retrieving means 140 for retrieving the melody of the musical piece by using the compensated scale string as a retrieving condition, and an output means 150 for outputting the retrieved musical melody.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a music information retrieving apparatus, and more particularly to a music information retrieving apparatus using a part of a melody of a music as a retrieval condition. In particular, the present invention relates to a music information search device that can be used when searching a database that stores information about music, such as when selecting music with a jukebox or a karaoke device.

[0002]

2. Description of the Related Art In an apparatus for retrieving information such as music according to the prior art, generally, a name assigned to a group of information to be searched such as a title of a music, or a symbol obtained by encoding such a name is used by a user. The input information is used to search the database. Here, the encoding of the name is performed manually.

FIG. 9 shows a music information retrieval apparatus according to the above-mentioned prior art, which includes an input unit 401 for inputting retrieval conditions from a user and a retrieval unit 4 for retrieving information based on the retrieval conditions.
02, a database 403 that stores the searched information,
The output unit 404 that outputs the search result, the speaker 405 that outputs the search result as a music melody, and the display 406 that displays the search result. Input section 40
The title or symbol input in 1 is sent to the search unit 402, and the search unit 402 searches the database 403 for information corresponding to the title or symbol. The search result music information is extracted from the database 403 to the output unit 404. The output unit 404 converts the music data in the search result music information into an acoustic signal and appropriately amplifies it,
The auditory information is provided to the user via the speaker 405. On the other hand, the character information or the image information in the search result music information is displayed on the display 406 by the output unit 404 and is provided to the user as visual information. The speaker 405 and the display 406 are arranged according to the type of information included in the information output by the music information retrieval device, that is, the presence or absence of auditory information and / or visual information.

[0004]

However, according to the above-mentioned conventional device, the user can specify as the search condition only the title of the music, the name corresponding to the title, or the symbol. Unable to retrieve music information about unknown music. On the other hand, it is empirically known that, for a song in which the user has a melody, if the user can search the music information using the stored melody, the music information can be searched more effectively. .

However, when a melody is used as a search condition, it is not efficient to search using all the melody of the music, and at the same time, the user stores all the melody of the music and gives them as the search condition. Is not effective because it imposes a heavy burden on the user. In addition, the tone of the music is not necessarily uniquely determined (for example, it may be in C major or F major), and it is also difficult for the user to reproduce the melody in an accurate tone. Therefore, it is difficult to search for a song from the melody stored by the user.

In view of the above problems, the present invention is an information retrieval apparatus capable of retrieving desired information such as music information of a musical composition from a database by using a part of a melody of a musical composition given by a user in various tones as a retrieval condition. The purpose is to provide.

[0007]

FIG. 1 is a block diagram showing the principle of the present invention. The music information retrieval device of the present invention includes an input unit 110 for inputting a part of a melody of a musical composition and an input unit 11.
In the spectrum obtained by Fourier transforming a part of the melody of the music input by 0, the frequency showing the spectrum having the maximum intensity within the frequency range corresponding to the range of the predetermined scale is selected, and the input means 110 is selected. The conversion means 120 for generating a scale sequence by encoding a part of the melody of the music input by the above into the scale corresponding to the frequency, and the conversion means 1.
Correction means 130 for correcting the deviation of the absolute value component of each scale in the scale sequence generated by 20; search means 140 for searching the melody of the musical composition using the scale sequence corrected by the correction means as the search condition; Output means 150 for outputting the melody of the music searched for by.

Further, the correction means 130 of the music information retrieval apparatus of the present invention sets the difference between the first scale and the predetermined reference scale from each scale of the scale sequence so that the first scale of the scale sequence matches the predetermined scale. The difference is corrected to correct the deviation of the absolute value component of each scale of the scale sequence.

Further, the correction means 130 of the music information retrieval apparatus of the present invention corrects the deviation of the absolute value component of each scale of the scale sequence by making a difference between two adjacent scales in the scale sequence.

[0010]

The input means 110 of the music information retrieval apparatus of the present invention
Can input the voice itself from the user with only a part of the melody of the music as the search condition, and the user can input the search condition even when the search condition is ambiguous. at the same time,
Since only a part of the melody of the song is used as the search condition instead of the entire melody of the song, the size of the search data is limited, and it is possible to suppress an increase in the search processing time that increases according to the size of the search data. .

Further, the conversion means 120 of the music information retrieval apparatus of the present invention selects the frequency having the maximum intensity spectrum within a predetermined frequency range from the spectrum obtained by Fourier transforming a part of the melody of the inputted music. Since it is encoded into a musical scale, undesired noise is less affected during encoding. More importantly, since the melody of the music is encoded, the search for the melody of the music can be easily replaced by replacing the conventionally known character string search.

In addition, the correction means 130 of the music information retrieval apparatus of the present invention can extract the fluctuation component of the scale in the scale by correcting the deviation of the absolute value component of the scale in the scale. For example, even if a part of the frequencies of the melody of the inputted music has a deviation from the frequency of the ideal signal, the influence of this deviation can be eliminated. Therefore, when a part of the melody of a musical composition is input, the device of the present invention can convert it into a scale sequence that is not affected by the difference in tone, so that an accurate search result can be obtained even if there is a pitch difference in pitch. It becomes possible to obtain.

Further, as described above, in the music information retrieval apparatus of the present invention, since a part of the melody of the music which is the retrieval condition is encoded into a character string or the like, the character string should be used as the retrieval key of the database. And search means 140
Can be searched by a character string search.

[0014]

Embodiments of the present invention will now be described in detail with reference to the drawings. FIG. 2 is a block diagram of an information retrieval system according to the first embodiment of the present invention. A microphone 201 for inputting a part of a melody that is a voice from a user and a part of a melody of an inputted music are shown. A Fourier transform unit 207 that performs a Fourier transform to determine a scale according to the maximum value of the spectrum and generates a scale sequence, and a Fourier transform unit 207.
A transposing unit 208 that corrects the deviation of the absolute value component of the scale so that the scale sequence generated in 1. starts from the standard scale, and a scale sequence of a part of the melody of the song as a search key A database 203 from which information is searched,
A search unit 202 that searches the database 203 using the scale sequence in which the deviation of the absolute value component of the scale is corrected by the transposing unit 208 as a search condition, and music information that includes the melody of the song from the database 203 according to the search result of the search unit 202. Of the sound information such as a melody is converted into an audio signal, and video information such as characters and images is converted into a video signal and output, and a speaker 205 that makes the audio signal audible.
And a display 206 for visualizing a video signal.

The search operation of the music information search system according to the first embodiment of the present invention will be described in detail below. When the user inputs a part of the melody of a piece of music for which information is desired to be searched for via the microphone 201, a part of the input melody is sequentially Fourier-transformed by the Fourier transform unit 207 to calculate the spectrum, and the obtained spectrum is calculated. The spectrum is converted to the scale corresponding to the frequency at which the maximum value is obtained. Here, the maximum value of the spectrum is not included in a normal musical piece because the scale converted corresponding to the frequency of the spectrum is selected from the spectrum of frequencies in the range included in the normal scale range. The effects of unwanted frequencies are eliminated. When adding the converted scale to the scale sequence, it is stored in the scale sequence every time the converted scale changes.
As a result, the length of the note is ignored, and the notes until different notes appear are treated as one note. Eventually, the scale sequence stores the scale each time the scale changes corresponding to a part of the melody, so that it is possible to perform a search that is not affected by variations in melody speed or rhythm.

Next, in the transposing unit 208, the scale stored in the scale sequence is modulated so that the scale of the tone stored at the beginning of the scale sequence, that is, the scale of the first tone, which is a part of the melody, matches a predetermined reference scale. . Specifically, the difference (hereinafter referred to as a bias component) between the scale of the first tone of a part of the melody and a predetermined reference scale is subtracted from all the scales stored in the scale sequence. As a result, a scale sequence in which the bias component is corrected is obtained. This enables a search that does not depend on the absolute value component of the scale of the first note that is a part of the melody.

The scale sequence corrected by the transposing unit 208 is input to the search unit 202, and the one matching the search key of the music information stored in the database 203 is searched. Here, since a scale sequence having scales as data, that is, a character string is used as a search key, for example, a conventional character string search technique such as a forward match search or a search for outputting search results in descending order of similarity by similarity search. Can be used to create the database 203. By the above search, the music having the melody that best matches a part of the inputted melody is searched.

The output unit 204 extracts music information regarding the corresponding music piece from the database 203 based on the search result of the search unit 202. The output unit 204 converts sound information such as a melody in the music information into an acoustic signal, amplifies it appropriately, and then outputs it to the speaker 205 to make it an audible sound. On the other hand, video information such as character information and image information in the music information is converted into a video signal and output to the display 206 and displayed on the screen of the display 206.

In the structure of the information search system according to the first embodiment of the present invention, both the speaker 205 and the display 206 are independently installed for each search system. The necessary devices may be installed according to the output information. Also,
Speaker 205 and display 206 may be connected to allow switching between multiple information retrieval systems.

FIG. 3 shows a spectrum obtained by Fourier transforming the inputted melody in the Fourier transform unit 207 of the information retrieval system according to the first embodiment of the present invention. In the spectrum 1 of (a) of the figure, the frequency at which the spectrum shows the maximum value is included in the frequency range of the normal tone, and the scale corresponding to this frequency is, for example, “D”.
It indicates that the sound is. Here, the scale is represented by the alphabets from "A" to "G". The Fourier transform unit 207 encodes the input acoustic signal into “D”. On the other hand, in the spectrum 2 of FIG. 6B, the frequency at which the spectrum has the maximum value is outside the frequency range of the normal sound, so the frequency at which the spectrum has the maximum value is selected within the frequency range of the normal sound. Indicates that the scale corresponding to that frequency is, for example, the "E" tone. That is, the Fourier transform unit 207 encodes the input melody into "E".

FIG. 4 shows the Fourier transform unit 2 in the modulation unit 208 of the information retrieval system according to the first embodiment of the present invention.
The conversion of the scale sequence output from No. 07 is shown. In the figure, the output of the Fourier transform unit 207 is "GDEDCBA".
In the case of “G”, the predetermined reference scale is “C”, the scale is lowered by 5 degrees so that the “G” of the first tone matches the scale “C”, and the conversion result of the scale sequence in the transposing unit 208. Is "C
It is shown that it becomes "GAGFEDC".

FIG. 5 is a flow chart for briefly explaining the operation flow of the information search system according to the first embodiment of the present invention described above.

The microphone 201 inputs a part of the melody, performs processing such as A / D conversion, and outputs it to the Fourier transform unit 207 (step 100).

The Fourier transform unit 207 sequentially performs a Fourier transform on a part of the melody, obtains its spectrum (step 110), and the scale corresponding to the frequency at which the spectrum has the maximum value is outside the normal scale range. In the case of, the process proceeds to step 130 (step 120), the maximum spectrum corresponding to the scale outside the normal scale range is deleted, and the process returns to step 120 (step 130), where the spectrum corresponds to the frequency having the maximum value. If the scale is within the range of the normal scale, a part of the Fourier-transformed melody is encoded into a scale within the range of the normal scale corresponding to the frequency at which the spectrum has the maximum value and output (step 14).
0).

The transposing unit 208 obtains the difference between the scale of the first tone and the scale of the predetermined reference tone in the sequence of scales output from the Fourier transform unit 207 (step 150), and obtains the difference between the obtained scales. Are subtracted from the scales stored in the scale sequence to obtain a sequence of transposed scales and output to the search unit 202 (step 160).

The search unit 202 retrieves a search key consisting of a part of the scale of the melody from the database 203 (step 170), compares it with the sequence of scales obtained from the modulation unit 208 and the search key obtained from the database 203, If they do not match, the process returns to step 170 (step 18).
0).

The output unit 204 obtains a search key as a search result from the search unit 202, obtains music information corresponding to the search key from the database 203, extracts an acoustic signal from the music information, and appropriately amplifies it. Output to the speaker 205 (step 190).

The speaker 205 obtains an acoustic signal and converts it into an audible sound (step 200).

FIG. 6 is a block diagram of an information retrieval system according to the second embodiment of the present invention, in which a microphone 201 for inputting a part of the melody of a music piece and a part of the melody of the input music piece are coded. Fourier transform unit 207 that generates a scale sequence by converting the encoded data into the encoded data according to the maximum value of the spectrum, and the difference between adjacent scales stored in the scale sequence generated by the Fourier transform unit 207. A difference unit 210 that calculates (hereinafter referred to as a difference), generates a scale sequence including the differences, and corrects the deviation of the absolute value component of the scale.
And, using the search sequence as the scale sequence of some differences in the melody,
The database 203 for searching the stored music information such as music, the search unit 202 for searching the database 203 using the scale sequence whose bias component is corrected by the difference unit 210 as a search condition, and the search result of the search unit 202 According to the output unit 204, the music information including the melody of the music is acquired from the database 203, the sound information such as the melody is converted into an acoustic signal, and the video information such as characters and images is converted into a video signal and output. , A speaker 205 for making an acoustic signal audible
And a display 206 for visualizing a video signal.

As can be seen from the above, in the second embodiment of the present invention, the modulation section 20 in the first embodiment of the present invention is used.
8 is replaced by the difference section 210. less than,
The search operation in the information search system according to the second embodiment of the present invention will be described in detail. When the user inputs a part of the melody of a piece of music for which information is desired to be searched for via the microphone 201, a part of the input melody is sequentially Fourier-transformed by the Fourier transform unit 207 to calculate the spectrum, and the obtained spectrum is calculated. The spectrum is converted to the scale corresponding to the frequency at which the maximum value is obtained. Here, the maximum value of the spectrum is not included in a normal musical piece because the scale converted corresponding to the frequency of the spectrum is selected from the spectrum of frequencies in the range included in the normal scale range. The effects of unwanted frequencies are eliminated. When adding the converted scale to the scale sequence, it is stored in the scale sequence every time the converted scale changes. As a result, the length of the note is ignored, and the notes until different notes appear are treated as one note. Eventually, the scale sequence stores the scale each time the scale changes corresponding to a part of the melody, so that it is possible to perform a search that is not affected by variations in melody speed or rhythm.

Next, the difference section 21 of the second embodiment of the present invention.
In order to correct the deviation of the absolute value component of the scale in the scale sequence obtained by the Fourier transform unit 207, 0 extracts the scale difference which is the difference between the scales in the scale sequence. More specifically, in the scale difference unit 210, the difference between the scales stored adjacently in the scale sequence, for example, when the scales are “G” and “D”, the scale “G” and the scale “D” are Convert a scale sequence into a sequence of scale differences by obtaining a numerical value that indicates how many semitones apart. The scale difference between adjacent scales corresponds to the changing component of the adjacent scale, and the influence of the absolute value component of the scale is removed, so the deviation of the absolute value component of the scale can be calculated by finding the scale difference of the scale. Will be corrected. As a result, it becomes possible to execute a search that does not depend on the absolute value component of a part of the scale of the melody.

The scale sequence of the difference obtained by the difference unit 210 is input to the search unit 202, and a match with the search key of the music information stored in the database 203 is searched.
Here, since a scale string having the scale difference of data as a data, that is, a character string is used as a search key, for example, a conventional character string such as a prefix search or a search that outputs search results in descending order of similarity by similarity search is used. Database 20 using column search technology
You can do 3. Incidentally, in response to conversion of the scale into the scale difference in the scale difference unit 210, the search key of the database 203 according to the second embodiment of the present invention is
It is a scale sequence that stores the difference of some scales of the melody. Further, the search unit 202 searches for a match between a part of the scale of the melody and the search key in the database 203. By the above search, the music having the melody that best matches a part of the inputted melody is searched.

The output unit 204 extracts the music information regarding the corresponding music piece from the database 203 based on the search result of the search unit 202. The output unit 204 converts sound information such as a melody in the music information into an acoustic signal, amplifies it appropriately, and then outputs it to the speaker 205 to make it an audible sound. On the other hand, video information such as character information and image information in the music information is converted into a video signal and output to the display 206 and displayed on the screen of the display 206. still,
In the configuration of the information search system according to the second embodiment of the present invention, the speaker 205 and the display 20 are provided.
Both 6 and 6 are installed independently for each search system, but necessary devices may be installed according to the information output as a search result. Also, the speaker 205
And display 206 may be connected to allow switching between multiple information retrieval systems.

FIG. 7 is a diagram for explaining the difference extraction of the difference unit 210 of the music information search system according to the second embodiment of the present invention, in which the output of the Fourier transform unit 207 is "G".
In the case of "DEDCBAG", the corresponding difference is expressed by how many times the difference between the front and rear sounds is a semitone, so that "(7) (2) (-2) (-2) (-1)" (-
2) (−2) ”difference can be obtained. Here,
A positive number difference indicates that the following note is higher than the previous note, while a negative number difference indicates that the following note is lower than the previous note. Such a search using the difference is particularly effective in finding a match of a part of the melody in the middle of the music.

FIG. 8 is a flow chart briefly showing the operation flow of the information search system according to the second embodiment of the present invention. Note that steps 100 to 140 and steps 190 to 200 are common to the flowchart shown in FIG. 5 showing the flow of operation of the information search system according to the first embodiment of the present invention. In the following, not shown in the flow chart for the first embodiment of the invention described in FIG.
The steps specific to the second embodiment of the present invention will be described.

Step 250) The scale difference unit 210 obtains the scale sequence of the scale difference by obtaining the difference between adjacent scales of the scale sequence output from the Fourier transform unit 207.

(Step 270) The retrieval unit 202 retrieves the retrieval key consisting of the difference in the scale of a part of the melody from the database 203.

Step 280) The search unit 202 uses the scale difference sequence obtained from the scale difference unit 210 and the database 20.
It is compared with the search key obtained from No. 3, and if they do not match, the process returns to step 270.

[0039]

As described above, according to the music information retrieval apparatus of the present invention, a part of the melody stored by a person is input as a condition for retrieval, and the music or the related information of the music is retrieved. Since the search can be performed, there is an effect that the success probability of the intended music information search is increased.

[Brief description of drawings]

FIG. 1 is a principle configuration diagram of the present invention.

FIG. 2 is a configuration diagram of an information search system according to the first embodiment of the present invention.

FIG. 3 is a diagram showing a Fourier transformed spectrum of an acoustic signal according to the first embodiment of the present invention.

FIG. 4 is a diagram for explaining modulation according to the first embodiment of the present invention.

FIG. 5 is a flowchart illustrating an operation of the information search system according to the first embodiment of the present invention.

FIG. 6 is a configuration diagram of an information retrieval system according to a second embodiment of the present invention.

FIG. 7 is a diagram for explaining difference extraction according to the second embodiment of the present invention.

FIG. 8 is a flowchart illustrating an operation of the information search system according to the second exemplary embodiment of the present invention.

FIG. 9 is an explanatory diagram of a conventional technique.

[Explanation of symbols]

 100 music information retrieval device 110 input means 120 conversion means 130 correction means 140 retrieval means 150 output means 190 database 201 microphones 202, 402 retrieval unit 203, 403 database 204, 404 output unit 205, 405 speaker 206, 406 display 207 Fourier transformation unit 208 Modulation section 210 Difference section 401 Input section

Claims (3)

[Claims] 1. A music information retrieval apparatus for retrieving a melody and music information of a music piece from a database, input means for inputting a part of the melody of the music piece, and a part of the melody of the music piece input by the input means. A part of the melody of the music inputted by the input means by selecting a frequency showing a spectrum having the maximum intensity in a frequency range corresponding to a predetermined scale range among spectra obtained by Fourier transforming A conversion means for generating a scale sequence by encoding a scale corresponding to the frequency, a correction means for correcting the deviation of the absolute value component of each scale in the scale sequence generated by the conversion means, and the correction means. The search means for searching the melody of the music by using the scale sequence corrected by the search condition and the output of the melody of the music searched by the search means. Music information retrieval apparatus and means. 2. The correction means subtracts the difference between the leading scale and the predetermined reference scale from each scale of the scale sequence so that the leading scale of the scale sequence matches a predetermined scale, 2. The music information retrieval apparatus according to claim 1, wherein a deviation of the absolute value component of each scale of the scale sequence is calculated and the deviation is corrected. 3. The correction means corrects the deviation of the absolute value component of each scale of the scale sequence by making a difference between two adjacent scales in the scale sequence. Music information retrieval device.