WO2013019057A2

WO2013019057A2 - Apparatus for automatically selecting music according to a user's preference

Info

Publication number: WO2013019057A2
Application number: PCT/KR2012/006087
Authority: WO
Inventors: 김희찬
Original assignee: 주식회사 제이디사운드
Priority date: 2011-08-02
Filing date: 2012-07-31
Publication date: 2013-02-07
Also published as: KR101142244B1; WO2013019057A3

Abstract

The apparatus of the present invention includes: a spectrum analyzer for spectrum-analyzing music files stored in a memory unit and classifying the spectrum-analyzed music files based on a preference or feelings according to set conditions; a list classifier for generating one or more playing lists for the preference or feelings classified by the spectrum analyzer and storing the playing lists; a user input unit for receiving music file information that the user has specified or a user's preference information in order to determine the user's preference or feelings; and a controller for generating pattern information on specific music file information input from the user input unit using the spectrum analyzer, selecting a list determined to be the most similar to the generated pattern information from among the playing lists of the list classifier, and outputting information on the selected list to the user to receive an input on whether to play same, or extracting one or more music files determined to be similar by the spectrum analyzer according to the generated pattern information from among the music files stored in the memory unit to generate a list, and outputting the generated list to the user and receiving an input on whether to play same.

Description

[Revision 31.07.2012 by Rule 26] 26 Automatic selection device according to user's preference

The present invention relates to an automatic selection apparatus for generating a list of songs having similar pattern information through user's taste information. More particularly, the present invention analyzes a music file based on taste or feeling through spectrum analyzer to generate pattern information. By creating a list by extracting songs similar to the music file selected by the user, the user maximizes convenience and satisfaction when listening to music, and creates a ring type by properly mixing similarity rankings when creating a list. The present invention relates to an automatic selection apparatus according to a user's taste capable of minimizing a width of a pattern change such as a taste or a feeling between songs.

Recently, as the types of multimedia contents have increased and interest in multimedia has increased, the issue of how to select and listen to music has become an issue. To this end, a description method for classifying multimedia contents is made in the MPEG-7 standard.

In the case of a terminal such as a portable music player, an MP3 player or a personal computer (PC), the number of users who listen to music using the terminals is increasing rapidly. It has become important to choose and listen to the music you want. According to this trend, various technologies for selecting a user's favorite music according to their taste and providing a list have been proposed.

Currently used music selection service is provided in two ways. One method is to create a playlist by automatically selecting music that meets the user's preferences. Using the data on the music stored in the storage unit of the terminal, the user inputs a taste standard desired by the user to the input unit of the terminal to automatically select music that meets the criteria.

The terminal automatically selects music that meets the user's taste standard among the stored music. In this case, the user's taste criteria input to the input unit may be a song name, a composer, and a record name of a song. If there is no such information in a corresponding song, speed, mood, or genre may be used.

Accordingly, the music is automatically selected as a list, and the created list is displayed for the user to view so that the user can listen to the music. According to the above method, when the user clearly inputs a taste of music desired by the user to the input unit of the terminal, the user may automatically select music that matches the taste.

Next, when the user selects one or more music samples, the user acquires user's taste information and automatically selects music according to the user's taste to create a playlist.

In this case, the terminal generates a plurality of music samples to analyze the user's taste. The music sample may be used as a sample by arbitrarily selecting a plurality of stored music and may be used as a sample classified by genre or speed.

When the user listens to the music sample and determines that the music being listened to is his or her desired taste, the user selects the desired music by using the input device of the terminal while the song is being output.

Accordingly, the terminal analyzes the user's taste information through the music selected by the user, and then automatically selects music that meets the analyzed taste criteria, creates a list of the selected songs, and displays them for viewing by the user. In the above-described method, the user can automatically select the music that suits his or her taste only by selecting a sample, without clearly inputting a taste of the music desired by the user to the input unit of the terminal.

However, the above two automatic music selection devices simply select music similar to the user's taste information and provide it to the user as a playlist, and output it, so that the pattern analysis accuracy is not guaranteed. The pattern deviation between the most similar songs and the songs with the least similarity is severe, reducing the user's satisfaction when the list is repeatedly played.

Therefore, as the overall pattern continuity between the selected music files is improved during the generation of the playlist, an improvement of the automatic selection apparatus that can maximize the user's convenience and satisfaction is required.

The present invention has been made in order to solve the above problems, and through the taste information or music file information input from the user to extract a particular number of music files that are determined to be similar to the taste or feeling of all the songs stored in the memory unit and repeated playback The user can maximize the user's convenience and satisfaction by automatically selecting a music file suitable for the user's taste by creating a ring-type playlist for the user, and repeating the arrangement of song arrangements appropriately when creating a ring-type playlist. Minimizes the pattern variation of the entire song during playback, and cross-fading, beat-leaf, and sound effects are mixed and output through the crossfader, beatbox, and surround effect between songs that are played continuously. Use to increase connectivity and continuity The purpose is to provide an automatic selection device according to the taste.

The present invention has the following features to achieve the above object.

According to an aspect of the present invention, there is provided a memory unit for storing a plurality of music files, a spectrum analyzer for performing pattern classification based on a taste or a feeling based on a setting condition by analyzing the music files stored in the memory unit, and the spectrum. A list sorter for generating one or more playlists according to tastes or feelings by the analyzer and storing the playlists; a user input unit for receiving music file information specified by the user or inputting taste information of the user to grasp the user's tastes or feelings; When specific music file information is input from the user input unit, pattern information is generated through a spectrum analyzer to select a list which is determined to be the most similar among the playlists of the list classifier, and outputs the selected list information to the user for playback. Input or award And a controller for extracting one or more music files determined to be similar through a spectrum analyzer among the music files stored in the memory unit according to the generated pattern information, generating a list, outputting them to a user, and inputting whether to play them. .

Here, the control unit converts the list output to the user into a play list of a ring type capable of repeat play, but the arrangement order of the ring type play list generates a similarity rank according to the user's pattern information, In order to reduce the pattern gap between the first music file and the final music file, when the entire music file is composed of 2n songs, the ring type playlist order is 1, 3, ..., 2n-1, 2n, 2n-2 , ...., arranged in the order of 4, 2, when composed of 2n-1 songs 1, 3, ..., 2n-1, 2n-2, 2n-4, ...., 4 In addition, the ring list generator is arranged in the order of 2.

In addition, the automatic selection device further includes a crossfader for crossfading between neighboring music files when the playback is performed by the ring list generator of the controller, wherein the crossfader includes an input unit for continuously inputting music files to be played; A time stretcher which receives the music file from the input unit and expands the number of sampling data per second of the music file at a specific ratio, a fish buffer which receives the sample data from the time stretcher, and temporarily stores the sample data, and from the fish buffer A mixer receiving sampling data, an output unit for outputting sampling data received from the mixer, an end detector for detecting an end signal of a music file transmitted to the time stretcher unit through the input unit, the input unit and the time stretcher unit Installed between the shutdown detector When a signal is detected, any one of the music inputs continuously by controlling the input unit, the time stretcher, the PC buffer, the end detector, and the switching unit to bypass the music file to be input to the mixer for a specific time. At the end of the file, the music file and the music file continuously inputted thereon are cross-faded through the mixer for a specific time. When the corresponding sound source data ends after a specific time has elapsed, the music file bypassed by the switching unit is timed. It includes a controller for delivering to the treasurer to output the music file.

In addition, the automatic selection apparatus includes information about two songs and pattern patterns between songs and songs among a plurality of pre-stored bit leaves during cross fading to increase connectivity between two consecutive music files when performing cross fading through the cross fader. Selects a bit leaf that is determined to be similar and mixes it in the cross fading section, and outputs the bitbox, and it is determined that the pattern information is most similar to the two songs between the song and the song among the pre-stored background effect sounds during the cross fading section. The apparatus further includes a surround effector which selects a background effect sound and mixes the result in a cross fading period.

In addition, the spectrum analyzer analyzes the characteristics of the beat component, the code component, the musical instrument component, and the vocal component through the decomposed frequency information when generating the pattern information for pattern classification, and collects each characteristic information according to the weight to determine the similarity of the pattern information. This is done.

According to the present invention, the taste information or music file information input from the user is pattern-analyzed by the spectrum analyzer to automatically select a certain number of similar music files, thereby providing convenience that the playlist user does not need to designate separately. By properly arranging the song arrangement order when generating the playlist, there is an effect of minimizing the width of the pattern change such as the taste or the feeling between the entire songs during repeated playback.

In addition, the crossfader, beatbox, and surround effector increase the natural connectivity and continuity between two songs played in succession.

1 is a block diagram showing the internal configuration of an automatic selection apparatus according to the present invention.

2 is a block diagram showing a detailed configuration of a crossfader according to the present invention.

3 is a diagram illustrating a ring type playlist generation process according to the present invention.

10: memory 20: spectrum analyzer

30: list sorter 40: user input unit

50: crossfader 60: beatbox

70: surround effect 80: control unit

90: ringlist generator

Hereinafter will be described in detail with the accompanying drawings for the automatic selection device according to the present invention.

1 is a block diagram showing the internal configuration of the automatic selection apparatus according to the present invention, Figure 2 is a block diagram showing a detailed configuration of the crossfader according to the present invention.

Referring to the drawings, the automatic selection device according to the present invention is a memory unit 10, spectrum analyzer 20, list classifier 30, the user input unit 40, crossfader 50, bit box 60, surround It consists of an effector 70 and a control unit 80.

Here, the memory unit 10 is provided for the user to store music files for playback, and the memory unit 10 transmits the music files to the beat box 60 and the surround effect 70 as well as the music file, which will be described later. The beat leaf or sound effect data mixed in the cross fading section during playback may also be stored together.

On the other hand, the spectrum analyzer 20 is provided to perform a pattern classification based on a setting condition based on the taste or feeling by analyzing the music file stored in the memory unit 10, such a spectrum analyzer 20 performs the pattern classification When the pattern information is generated, the similarity of the pattern information is determined by analyzing characteristics of the beat component, the chord component, the musical instrument component, and the vocal component through the decomposed frequency information.

The spectrum analyzer 20 includes a BPM calculator 21, a chord extractor 22, and an instrument information extractor to analyze characteristic information for each bit component, code component, musical instrument component, and vocal component. Extractor (23), and Voice Extractor (24).

When the frequency decomposition information is generated through the spectrum analyzer 20, the bit number calculator 21 extracts the number of bits per minute as a frequency amplitude change pattern, and may be used for genre analysis through the speed of the song or the bit pattern of the song. The code extractor 22 is provided to generate information on which section has which code by extracting the scale and chord information of the music file through the frequency decomposition information.

In addition, the instrument information extractor 23, after storing the specific frequency pattern information for each instrument in advance, when the frequency decomposition information is generated by the spectrum analyzer 20 determines the frequency pattern information of which instrument, and finally generates an instrument component, The voice extractor 24 is configured to analyze the characteristics of the tone of the singer who calls the music file and classify it into a form of caustic, true, husky, aesthetic.

Accordingly, the characteristic information analyzed for each component may be configured to determine the similarity as a whole by assigning a weight to each component according to a setting condition according to a user's input. For example, the user sets the speed of a song to a weight of 70%. If the chord, musical instrument, and vocal component are set to 10%, the spectrum analyzer 20 can determine the overall similarity even if the chord component, the musical instrument component, and the vocal component are not similar to each other.

The weight may be set by being input through the user input unit 40 to be described later, and the setting condition may be stored in the controller 80 or the spectrum analyzer 20 to be used when determining the similarity.

Meanwhile, the list classifier 30 is provided by the spectrum analyzer 20 to generate and store one or more playlists according to tastes or feelings. For example, the list classifier 30 may be classified into slow songs, normal songs, and fast songs. You can create individual playlists, and in addition to being fast, you can also combine chords, musical instruments, and vocal characteristics to create playlists based on specific feelings.

Accordingly, the playlists classified according to the present invention are further analyzed after the user analyzes specific music file information or taste information and searches for and selects the most similar playlist, and separately selects the entire music file stored in the memory unit 10 for the specific music file or taste of the user. Whenever there is an information input, it can be analyzed to prevent time consuming generation of a list.

In addition, the user input unit 40 is provided to receive the music file information specified by the user or to receive the user's taste information in order to determine the user's taste or feeling, in the case of the music file information, the user memory unit 10 It is inputted by selecting any one of all music files stored in the user. In the case of the taste information, the user inputs data in a text form such as "sad ballad" or "lightly danced music". Is specified.

At this time, the music file information generates the taste pattern information of the user through the spectrum analyzer 20, and in the case of the taste information, among the plurality of taste information example items stored in advance by recognizing data input by the user. The item determined to be the most similar may be selected to provide a playlist corresponding to the item to the user through the list classifier 30.

In addition, the user input unit 40 receives the user's input information through a touch screen method or a keypad method, and in addition to determining the similarity, based on a specific component among the fast, chord, musical instrument, and vocal components, or separately weights. The criterion may be set according to a user's selection, such as to perform a similarity determination by specifying the result. Alternatively, the user may be input whether or not to output a playlist provided to the user.

Meanwhile, when specific music file information is input from the user input unit 40, the controller 80 generates pattern information through the spectrum analyzer 20 to determine a list that is most similar to the playlist of the list classifier 30. Select to output the selected list information to the user to receive playback or extract one or more music files determined to be similar through the spectrum analyzer 200 among the music files stored in the memory unit 10 according to the generated pattern information After the list is generated, it is outputted to the user and provided to receive input.

As described above, when taste information in the form of text data is input from the user input unit 40, the list sorter 30 selects an item which is determined to be the most similar among a plurality of taste information example items stored in advance. Of course, provided to the user through.

In addition, the control unit 80 is provided with a ring list generator 90 for generating a ring-type playlist before providing the playlist generated by the list classifier 30 to the user, the ring list generator 90 is a user In order to reduce the pattern gap between the first music file and the last music file in order of similarity, the entire music file is composed of 2n songs. When the ring type playlist is arranged in order of 1, 3, ..., 2n-1, 2n, 2n-2, ...., 4, 2, and consists of 2n-1 songs 1, 3, ..., 2n-1, 2n-2, 2n-4, ...., 4, 2 will be arranged in order.

Of course, the arrangement order of the playlists through the ring list generator 90 may be arranged in a manner other than two songs unless two songs having a large ranking difference in the similarity ranking are arranged adjacent to each other.

Meanwhile, the crossfader 50 is provided to perform cross fading between neighboring music files when playing through a playlist. The crossfader 50 according to the present invention includes an input unit 51 for continuously inputting music files to be reproduced. A time stretcher unit 52 receives the music file from the input unit and expands the number of sampling data per second of the music file at a specific ratio, and receives a sampling data from the time stretcher 52 to temporarily store the sample data. And a mixer 54 receiving sampling data from the fish buffer 53, an output unit 55 for outputting sampling data received from the mixer 54, and an input unit. End detector 56 for detecting the end signal of the music file delivered to the time stretcher 52, and between the input unit and the time stretcher 52 to the end detector 56 When the end signal is detected, the control unit 57 continuously bypasses the music file to be input for a specific time, and controls the input unit, the time stretcher unit, the PC buffer, the end detector, and the switching unit. At the end of one of the music files, the music file and the music file continuously inputted thereon are cross-faded through a mixer for a specific time, and when the corresponding sound source data ends after a specific time has elapsed, the switching unit bypasses the switching. The controller 58 is configured to transmit the music file to the time stretcher so that the music file is output.

When the music file is input to the input unit 51, the controller 58 transmits the music file to the time stretcher 52 through the switching unit 57, and the time stretcher 52 samples the music files at a predetermined ratio. To increase. The predetermined ratio is preferably about 2 to 3%. Accordingly, the playing time of the music file is increased by the extended sampling number, and the music file and the cross fading section sequentially input during the corresponding time are generated.

At this time, the end detector 56 detects the end signal of the input music file through the end bit detector 56a, the silent detector 56b, and the code pattern detector 56c, which is a digital sound source such as MP3. In the case where the end bit is embedded at the end of the sound source data, the end bit detection unit 56a is configured to detect this, and in the case of the sound source data without the end bit, the silent portion generated at the end of the song is detected by the silence detection unit 56b. Will be done.

In addition, the code pattern detecting unit 56c may have a silent portion in the middle of a song, depending on the characteristics of the sound source data, so that the end point detection reliability of the sound source data may be degraded. When the silent section is detected and it is determined that the chord of the music file, that is, the code pattern, corresponds to a code pattern that is frequently used at the end point, the end point is recognized.

After detecting the end point, the music files sequentially input afterwards are immediately bypassed through the switching unit 57 to the mixer 54, and the previous music files and the subsequent music files are mixed and cross-crossed by the sampling extension section of the previous music files. Fading is configured to be performed.

In addition, the first volume control unit 59a and the second volume control unit 59b are respectively installed on the bypass path of the rear end of the fish-buffer 53 and the switching unit 57, so that the previous song gradually increases when cross fading is performed. By lowering the volume and gradually adjusting the music file, the music file can be configured to move naturally.

In addition, the music file mixed through the mixer 54 is converted into an analog signal through the D / A converter 55a and transferred to the output unit 55, the music file is output.

In addition, the input unit 51 may be provided with an A / D converter 51a for converting an input music file into an analog form in a digital form.

The cross fader 50 according to the present invention can be simplified as cross fading is possible with only one input unit without a plurality of input units.

In addition, the beat box 60 may be placed between songs and songs among a plurality of pre-stored bit leaves during the cross fading period to increase connectivity between two successive music files when performing cross fading through the cross fader 50. The bit leaf which is determined to be the most similar to the song and the pattern information is selected to be mixed and output in the cross fading execution section.

In addition, the surround effector 70 selects a background effect sound which is determined to have the most similar pattern information between the song and the song among the pre-stored background effect sounds during the cross fading period to be mixed and output in the cross fading period. The crossfader 50, the beat box 60, and the surround effector 70 may induce natural sequential playback by maximizing continuity and connectivity between two music files that are continuously played.

The crosspater 50, beatbox 60, and surround effector 70 are mixed with the music files of the playlist by a mixer 91 provided before the audio output through the playlist, such as cross fading, beatleaf, effect sounds, and the like. It is of course configured to output.

Referring to the drawing, the ring-type playlist generation process first analyzes all the songs stored in the memory unit 10 through the spectrum analyzer 20 through the user's taste pattern information according to the similarity of music files determined to be similar. Arranged in sequential order, the arranged list order is generated by the ring list generator 90 as a playlist of ring type in which similarities are appropriately mixed and arranged.

In FIG. 2, assuming that there are 14 songs from song A to song O in the memory unit 10 as a process of generating a ring list based on BPM, if the user selects a music file information or a taste information slow, BPM is selected. Creates a list order of a certain number of slow songs in a small order.

Looking at the generated sequential list, nine songs are arranged in order of song G, song L, song F, .... song H, and the ring list generator 90 mixes and arranges them appropriately. A, ..., song F arranged in order.

Accordingly, since the similarity gap is small between the first song G and the last song S during repeated playback, the user may not notice a large pattern change.

The mixed batch method may be formed in various forms in addition to the batch method of the ring list generator 90 described above.

As described above, the present invention has been described with reference to one embodiment shown in the drawings, but this is merely an example, and those skilled in the art may realize various modifications and other equivalent embodiments therefrom. Will understand.

Therefore, the true technical protection scope of the present invention will be defined by the technical spirit of the appended claims.

Claims

A memory unit for storing a plurality of music files,

A spectrum analyzer for spectrum analysis of the music file stored in the memory unit to perform pattern classification according to a preference or a feeling based on a preference condition;

A list classifier for generating one or more playlists according to tastes or feelings by the spectrum analyzer and storing them;

A user input unit for receiving music file information specified by the user in order to grasp the user's taste or feeling;

When specific music file information is input from the user input unit, pattern information is generated through a spectrum analyzer to select a list which is determined to be the most similar among the playlists of the list classifier, and outputs the selected list information to the user and inputs whether or not to play it. Receiving or extracting one or more music files that are determined to be similar through the spectrum analyzer of the music files stored in the memory unit in accordance with the generated pattern information to generate a list and output them to the user; Automatic selection device according to the user's preference, characterized in that configured.
The method of claim 1,

The control unit

The list output to the user is converted into a playlist of a ring type that can be repeatedly played, and the order of arranging the playlist of the ring type generates a similarity rank according to the user's pattern information, and then repeats the first music file of the similarity rank. In order to reduce the pattern gap between the final music files, when the entire music file is composed of 2n songs, the ring type playlist order is 1, 3, ..., 2n-1, 2n, 2n-2, ... ., Arranged in the order of 4, 2, when composed of 2n-1 songs 1, 3, ..., 2n-1, 2n-2, 2n-4, ...., 4, 2 Automatic selection device according to the user's preference, characterized in that it comprises a ring list generator arranged in.
The method of claim 2,

The automatic selection device

And a crossfader for cross fading between neighboring music files when the reproduction is performed by the ring list generator of the control unit.
The method of claim 3,

The crossfader is

An input unit for continuously receiving a music file to be played, a time stretcher unit receiving the music file from the input unit to expand the number of sampling data per second at a specific rate, and receiving and temporarily storing the sampling data from the time stretcher unit Detects an end signal of a music file transmitted to the time stretcher through the input buffer, a mixer receiving sampling data from the fish buffer, an output unit for outputting sampling data received from the mixer, and the input unit And a switching unit installed between the input detector and the time stretcher to bypass the music file input to the mixer for a specific time when a termination signal is detected from the ending detector, the input unit, the time stretcher, and the fish. Continuously controlled by controlling the buffer, termination detector and switching part At the end of any one of the music files being output, the music file and the music file continuously inputted thereon are cross-faded through a mixer for a specific time. An automatic selection device according to the user's taste, characterized in that it comprises a controller for transferring the music file passed to the time stretcher to output the music file.
The method of claim 3,

The automatic selection device

In order to increase the connectivity between two successive music files when performing cross fading through the cross fader, a bit determined to have the most similar pattern information between two songs among songs and songs among a plurality of beat leaves previously stored during a cross fading period. Select a leaf to be mixed and output in the cross fading section, and select a cross between the track and the background effect sound pre-stored during the cross fading section. The automatic selection device according to the user's taste, characterized in that it further comprises a surround effector to be mixed and output in the fading section.
The method of claim 1,

The spectrum analyzer analyzes the characteristics of the beat component, the chord component, the musical instrument component, and the vocal component through the decomposed frequency information when generating the pattern information for pattern classification, and collects each characteristic information according to the weight to determine the similarity of the pattern information. Automatic selection device according to the user's preference, characterized in that made.