WO2006030712A1 - 楽曲推薦装置及び方法 - Google Patents
楽曲推薦装置及び方法 Download PDFInfo
- Publication number
- WO2006030712A1 WO2006030712A1 PCT/JP2005/016642 JP2005016642W WO2006030712A1 WO 2006030712 A1 WO2006030712 A1 WO 2006030712A1 JP 2005016642 W JP2005016642 W JP 2005016642W WO 2006030712 A1 WO2006030712 A1 WO 2006030712A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- music
- range
- user
- search keyword
- update
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/60—Information retrieval; Database structures therefor; File system structures therefor of audio data
- G06F16/68—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/686—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using information manually generated, e.g. tags, keywords, comments, title or artist information, time, location or usage information, user ratings
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/60—Information retrieval; Database structures therefor; File system structures therefor of audio data
- G06F16/63—Querying
- G06F16/635—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10H—ELECTROPHONIC MUSICAL INSTRUMENTS; INSTRUMENTS IN WHICH THE TONES ARE GENERATED BY ELECTROMECHANICAL MEANS OR ELECTRONIC GENERATORS, OR IN WHICH THE TONES ARE SYNTHESISED FROM A DATA STORE
- G10H2240/00—Data organisation or data communication aspects, specifically adapted for electrophonic musical tools or instruments
- G10H2240/075—Musical metadata derived from musical analysis or for use in electrophonic musical instruments
- G10H2240/085—Mood, i.e. generation, detection or selection of a particular emotional content or atmosphere in a musical piece
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10H—ELECTROPHONIC MUSICAL INSTRUMENTS; INSTRUMENTS IN WHICH THE TONES ARE GENERATED BY ELECTROMECHANICAL MEANS OR ELECTRONIC GENERATORS, OR IN WHICH THE TONES ARE SYNTHESISED FROM A DATA STORE
- G10H2240/00—Data organisation or data communication aspects, specifically adapted for electrophonic musical tools or instruments
- G10H2240/121—Musical libraries, i.e. musical databases indexed by musical parameters, wavetables, indexing schemes using musical parameters, musical rule bases or knowledge bases, e.g. for automatic composing methods
- G10H2240/131—Library retrieval, i.e. searching a database or selecting a specific musical piece, segment, pattern, rule or parameter set
- G10H2240/141—Library retrieval matching, i.e. any of the steps of matching an inputted segment or phrase with musical database contents, e.g. query by humming, singing or playing; the steps may include, e.g. musical analysis of the input, musical feature extraction, query formulation, or details of the retrieval process
Definitions
- the present invention relates to a technical field of a music recommendation device and method used in, for example, powerful audio and stereo that can provide music according to the user's sensibility from among a large number of music.
- Patent Document 1 According to the information selection / playback apparatus described in Patent Document 1 (hereinafter referred to as "conventional technology"), a search keyword (for example, spectacular music, scat Up to and including the correlation values between the feature words (for example, brightness, darkness, enjoyment, spirit, goodness, etc.) , And a feature word list in which the presence / absence of features related to each feature word is described.
- search keyword for example, spectacular music, scat Up to and including the correlation values between the feature words (for example, brightness, darkness, enjoyment, spirit, goodness, etc.)
- a feature word list in which the presence / absence of features related to each feature word is described.
- the characteristics of the user's sensitivity are quantitatively extracted depending on the power of the selected music, for example, whether or not the user skipped, and the correlation value of the sensitivity table is updated.
- the Kansei table is updated to reflect the user's Kansei, so that it is possible to select content corresponding to the Kansei of the individual.
- Patent Document 1 Japanese Patent Application Laid-Open No. 2003-132085
- the present invention has been made in view of the above-described problems, and an object of the present invention is to provide a music recommendation device and method capable of recommending appropriate music for each user.
- the music recommendation device of the present invention is quantified for each of n (where n is a natural number) feature items predetermined to represent the characteristics of each of a plurality of music pieces.
- a range setting that sets a first range that defines a range that includes the n pieces of music feature values assigned to each of a plurality of search keys in an n-dimensional coordinate system having music feature values as n coordinate axes.
- input means capable of inputting a desired search keyword out of the plurality of search keywords, and when the desired search keyword is input via the input means, the plurality of pieces of music In the n-dimensional coordinate system, (i) the n music feature quantities belonging to the first range corresponding to the desired search keyword or (ii) the second range including the first range are included.
- a first range is set that defines a range including the n pieces of music feature amounts assigned to each of a plurality of search keywords.
- music is a concept including all music and music information that can be viewed by a user regardless of the genre, such as rock, pop, enka, jazz, or classical music.
- the place where such music is viewed is not limited to a moving body such as a car, a room such as a home or a company, or the outdoors. Also compatible with these viewing locations
- the situation at the time of viewing can also take various forms such as moving or stopping a moving body such as a car, working, studying, or walking.
- the music recommendation device of the present invention is also part or all of audio equipment, in-vehicle audio equipment, car navigation equipment, portable music equipment, etc., or an external function expansion that adds functions to them. It can take various forms such as a device.
- these music may be, for example, music included in commercially available music software recorded in advance on a CD or DVD, or the user via a network such as the Internet. May be music that has also been downloaded from a predetermined server or distributed from a predetermined sano. If the music is downloaded or distributed in this way and is acquired by the user, such music is appropriately stored on the user side by HD (Hard Disk), CD-ROM, CD-RW, DVD— It may be recorded and saved on a recording medium such as ROM or DVD RAM.
- HD Hard Disk
- CD-ROM Compact Disk
- CD-RW Compact Disc
- feature item is a parameter prepared in advance in order to categorize these multiple musical pieces.
- “classify sensitively” means, for example, “pleasant”, “bright”, “good”, “dark”, “quiet”, “good”, etc.
- the feature items in the present invention enable these sensibility classifications, for example, the property of music such as “rhythm”, “beat”, or “pitch” can be replaced as physical numerical values. Item power also becomes. For example, taking “rhythm” as an example, songs with slow rhythms are generally “quiet” songs, which are often “quiet” songs.
- all of the quantifiable items having some correlation with the sensitive classification items can be “characteristic items” in the present invention.
- the "first range” refers to each of a plurality of search keywords in an n-dimensional coordinate system with n music feature quantities quantified for each of n types of feature items as n coordinate axes. This is a range that defines a range that includes the n music feature values assigned to. Since n is a natural number, the coordinate system according to the present invention can take various forms such as a coordinate axis, a two-dimensional coordinate plane, or a multidimensional coordinate space.
- “n pieces of music feature values are used as n coordinate axes” means that n feature items are used as coordinate axis items.
- the coordinate value in the coordinate system may be the music feature amount itself or the music feature value. It may be a value generated or calculated from the collected amount, or a value obtained by normalizing the music feature amount. In the case of normalization, the difference between coordinate values related to one coordinate axis may be normalized so that the feature weights of the same degree are shared among a plurality of coordinate axes (that is, music feature quantities). .
- the “search keyword” is a phrase used to select a desired music piece on the user side. For example, “fun”, “bright”, “dark”, “energetic” It refers to phrases such as “good” or “good”. That is, the search keyword is a concept also called a “sensitive word” for searching for music pieces classified as “sensitive” as described above. In the present invention, each search keyword is assigned a music feature amount in a feature item in the same manner as a music piece.
- the first range is a range that defines a range including n pieces of music feature values assigned to each of these search keywords.
- the “covering range” described here refers to an appropriate range including a coordinate point that truly corresponds to the music feature amount assigned to each search keyword in the n-dimensional coordinate system.
- the search keyword indicates It is preferable that the range be considered as not deviating from the sensitivity characteristics. In the following description, it will be expressed that the sensuous characteristic power indicated by the search keyword is not deviated, and “matches the search keyword” as appropriate.
- the coordinate system is represented by a two-dimensional coordinate plane, but the first range set for one search keyword is the search keyword. It may be a circular range corresponding to the music feature amount assigned to the.
- an elliptical range may be used.
- it may be a quadrangular range obtained by superimposing the range of coordinate points allowed on each coordinate axis.
- it may be an appropriate range as a result of appropriately determining the first range as the initial value.
- the input means The desired one is entered.
- a search keyword such as “fun” is input.
- An input means that enables such input is, for example, a keyboard.
- Various forms such as a keyboard, a mouse, a scroll button, various buttons, or a touch panel device can be adopted, but as long as a user can input a search keyword in order to search for a song that the user desires to view, the form is arbitrary.
- the recommendation unit When a search keyword is input, the recommendation unit includes a first range corresponding to the search keyword input in the n-dimensional coordinate system or a second range including the first range from among a plurality of pieces of music. One or a plurality of music pieces having n music feature quantities belonging to the range are searched and recommended to the user.
- music recommendation reflecting the user's sensitivity is realized as follows.
- a song belonging to the first range described above or a wider second range including the first range is recommended as a “fun” song.
- the “second range” is a range including the first range.
- Music that falls within the second range and outside the first range is music that is originally considered not to match the input search keyword (for example, "fun").
- “does not match” means that it is highly likely that the feature is not the emotional feature indicated by the related search keyword and is highly likely to be seen from the music feature amount of the set feature item. It ’s not good. Therefore, depending on the user, there is a possibility that the music that exists outside the first range is felt as music that matches the input search keyword.
- This second range may be any one as long as it includes the first range, that is, a range including a music feature amount assigned to the search keyword (preferably a range that is considered to match the search keyword).
- the second range may be determined in advance by an empirical, experimental, or simulation technique.
- the recommendation means uses as the range recommended to the user may be freely set. For example, music belonging to the second range may be recommended at a rate of once for the first range multiple times. Or vice versa. Alternatively, music belonging to the second range may be recommended only when the user's selection operation prompts the user to recommend music belonging to the second range.
- one or a plurality of songs recommended in this way are viewed by the user, for example, by the user sequentially selecting from the recommended list or automatically starting to play in an appropriate order. Is done.
- a song is “viewed”, for example, the music name, artist name, or performance time of the music to be played is visually displayed on a powerful information display panel such as a liquid crystal display panel. It is a concept that includes cases where
- matches sensibility means that the sensibility element (for example, “fun”) entered as a search keyword matches the sensibility element that the user actually receives from this song. Point to.
- Such determination by the determination means is performed based on, for example, whether or not the user has listened to the music piece to the end. At this time, if the song has been listened to (played back), it can be determined that the user has determined that the song matches the input search keyword. On the other hand, if the performance is stopped or skipped immediately after the performance is started or during performance, it can be determined that they do not match.
- a dedicated button or the like may be prepared in advance for transmitting a force that matches a user's sensibility to a device, which is related to the reproduction status of the music. In this case, the button may be configured in common with the input means of the present invention.
- the user feels that it matches the search keyword that he / she actually entered, and simply played the song according to his / her mood. The case exists. If this situation is not taken into account, the user's sensibility is studied incorrectly. Therefore, the skipping of music by the user may be excluded from the discrimination target by the discrimination means.
- the range setting unit updates the first range corresponding to the input search keyword based on the determination result by the determination unit. More specifically, for example, when a song that falls within the second range and outside the first range is viewed to the end, the first range corresponding to the search keyword is large. Update to be. Further, even if it is within the first range, when a music piece belonging to the boundary is rejected by the user, the first range corresponding to the search keyword is updated so as to become smaller. Note that the update mode is not limited to the example illustrated here, and the first range is set so that the recommended music is closer to the user's sensibility based on whether or not the recommended music matches the user's sensibility. As long as it is updated, it may be updated in any way.
- the first range set for each search keyword can be updated to reflect the sensitivity of the user.
- this update can be performed bidirectionally, in other words, in a direction to further widen and narrow the first range, it is possible to recommend an appropriate piece of music for each user.
- an extraction unit that extracts the n music feature amounts, and the extracted music feature amounts are used for each of the plurality of music pieces.
- first storage means for storing the information in association with each other.
- the extraction unit extracts the music feature amount for each of the plurality of pieces of music, and the extracted music feature amount is associated with each of the plurality of pieces of music and is stored in the first storage unit.
- the music table in which music information and extracted music feature values are stored in association with each other may be created, and this music table may be stored.
- the music information may include musical tone data for playing a musical piece, information on a storage location of the musical tone data, a musical title, or an artist name.
- the first storage means means that if the music recommendation device according to the present invention is configured to recommend music from a predetermined recording medium such as a large-capacity HDD in advance. Such a recording medium for storing music may be used.
- the extracted music feature values may be stored in a format that can be referred to later in a form attached to the musical tone data of the music stored in the recording medium.
- the music setting device further includes second storage means for storing the set first range in association with each of the plurality of search keywords, and the range setting means. Updates the first range stored in the second storage means.
- the first range set by the range setting unit is stored in the second storage unit in a form associated with each search keyword, the latest update result is reflected for each user. It is also very convenient because it can always be referred to.
- a search table in which a search keyword name, information on the music feature amount to be assigned, information defining the first range, and the like are stored may be created and stored.
- the second storage unit may be configured as the same hardware as the first storage unit described above, or may be configured as different hardware. Also, some of them may be configured as common hardware.
- the determination unit determines whether or not the recommended music matches the sensibility of the user based on the individual determination of the user. To do.
- the determination means can determine whether or not the music recommended by the recommendation means matches the user's sensitivity based on the individual determination of the user. Sensitivity can be reflected with high credibility.
- Such individual determination by the user may be supplied to the determination means via, for example, a dedicated button or key representing “match” or “non-match”. Alternatively, it may be supplied to the discriminating means using a known voice recognition technology.
- the range setting means is a distance of reference coordinate power specified by a music feature amount assigned to the desired search keyword in the n-dimensional coordinate system.
- a set of points where R is R (R: positive real number) is the boundary of the first range. Set as a border.
- the boundary of the first range is set by the range setting means as a set of coordinate points that are equidistant from the reference coordinate point in the coordinate system.
- the “reference coordinate point” refers to a coordinate point uniquely determined by the music feature amount assigned to each search keyword in the n-dimensional coordinate system. Therefore, this reference coordinate point represents the emotional feature represented by the search keyword with the highest probability in the coordinate system.
- “equal distance from the reference coordinate point” is, for example, in the case of a two-dimensional coordinate system, the outer periphery of a circle centered on the reference coordinate point, and the first range in that case is That is, it points within the circle.
- the first range refers to the inside of the sphere.
- the value setting value of “R” that defines the first range changes by reflecting the user's sensibility by the update operation as described above of the range setting means. Note that the initial value of R is set to such a value when an optimal value can be predicted, estimated, or analogized in advance by a method such as experimental, empirical, or simulation. Have you been?
- the load required for the process of setting the first range is relatively light, and the reference coordinate point corresponding to one search keyword is set at the center of the first range. Therefore, it is possible to set the first range to a range that considers all feature items in a well-balanced manner.
- the recommendation means includes a range obtained by enlarging the second range at a predetermined ratio in the n-dimensional coordinate system. Set to.
- the recommendation unit sets the second range as a range obtained by expanding the first range at a predetermined rate, it is possible to reduce the processing load required for music recommendation. is there. In this case, even if the first range has a relatively complicated shape, the ratio and balance of various feature items in one search keyword represented by the first range are maintained.
- the second range can also be set easily, and music can be recommended effectively.
- the recommending means uses the second set of points whose distance from the reference coordinate point force is a value obtained by adding R and a predetermined margin r (r: a positive real number smaller than R). Set as the boundary of the range.
- the recommendation means sets a point corresponding to a distance obtained by adding a margin r to R defining the first range as a boundary of the second range. It is possible to reduce the required processing load.
- the value of the margin r is set to such a value when an optimal value can be predicted, estimated, or estimated in advance by a method such as experimental, empirical, or simulation. May be.
- the value of margin r is set to a value of about 10% of R that defines the first range, the sensibility characteristic indicated by one search keyword can be used so that the music does not deviate significantly. While it is possible to make a recommendation, it is relatively preferable to reflect the user's sensitivity without omission.
- the range setting means includes the reference described above among the plurality of songs in the n-dimensional coordinate system.
- the distance from the coordinate point is within a first update range that is greater than R and less than or equal to the sum of R and margin r, or (ii) less than R and from R to the margin
- the first range is updated based on the determination result by the determination means.
- the range setting means determines the result of determination by the determination means only when a song having a distance of “R ⁇ r” from the reference coordinate point is recommended. Since the first range is updated based on this, it is possible to effectively reduce the update frequency of the first range and perform efficient music recommendation. For example, a song corresponding to the vicinity of a reference coordinate point is an answer that many users feel that it matches a single search keyword, and the viewing results of such a song are within the first range. The need for renewal is considered extremely low. In addition, since the vicinity of the center of the first range is subject to update, the basic music recommendation algorithm in the music recommendation device of the present invention may be affected, which is not practical. .
- the range setting means recommends music existing in the first update range, and the recommended music matches the sensitivity of the user by the discrimination means. If it is determined that the first range is present, the first range may be updated in the direction of increasing.
- the range setting means recommends a song that exists within the second update range, and determines that the recommended song does not match the sensitivity of the user by the discrimination means. The first range may be updated in the direction of decreasing.
- the first range becomes large, and the music belonging to the second update range matches the user's sensibility. If this is not done, the first range will be small, and it will be possible to reflect the user's sensibility very effectively.
- the range setting means includes (i) before and after the update is performed. R in the above, R and R, respectively (ii)
- the first range can be constantly and effectively updated based on the above mathematical formula.
- the value of the update coefficient ⁇ may be set to any value within the above range, but is determined to be optimal or optimal in advance, predicted, estimated, or estimated by a technique such as experimental, empirical, or simulation. If a value that can be inferred is known, it may be set to such a value. For example, set the update coefficient to about "0.05" In this case, it is preferable that both the convergence speed of the first range and the credibility of the first range can be secured relatively.
- the characteristic items include a chord change rate, the number of beats per minute, the maximum beat amplitude, the average tone signal strength, and the maximum tone signal amplitude. Including at least one item.
- the feature item may be any item as long as the emotional feature of the music can be expressed quantitatively or numerically. For example, as described here, the chord change If the feature item includes at least one of rate, beats per minute, maximum beat amplitude, average tone signal intensity, and maximum tone signal amplitude, the feature of the song is relatively well represented. It becomes possible.
- chord change rate For example, if the chord change rate is relatively large, the possibility of a relatively lively song is high. If the chord change rate is relatively small, the possibility of a relatively quiet song is high. Yes. For example, if the number of beats per minute is relatively large, it is likely to be a relatively energetic or enjoyable song. If the number of beats per minute is relatively small, it is relatively relaxed. There is a high possibility that it is a powerful song. For example, if the maximum amplitude of the beat is relatively large, it is likely that the song is relatively intense. If the maximum amplitude of the beat is relatively small, the song may be relatively moderately powerful. High nature.
- the music is likely to be relatively healthy. If the average tone signal intensity is relatively small, the song may be relatively quiet. High nature. Or, for example, if the maximum amplitude of the music signal is relatively large, the music signal is likely to be a relatively passionate song. If the maximum amplitude of the music signal is relatively small, the music is relatively dark. Is likely.
- the music recommendation method of the present invention is quantified for each of n (where n is a natural number) feature items predetermined to represent the characteristics of a plurality of music pieces.
- a range setting that sets a first range that defines a range that includes the n pieces of music feature values assigned to each of a plurality of search keys in an n-dimensional coordinate system having music feature values as n coordinate axes.
- a desired search key ⁇ ⁇ ⁇ among the plurality of search keywords And when the desired search keyword is input in the input step, (i) the desired search keyword is selected from the plurality of songs in the n-dimensional coordinate system.
- the effects of the music recommendation device of the present invention described above can be easily realized by the steps corresponding to the steps in the music recommendation device of the present invention described above. .
- the music recommendation device of the present invention since the range setting means, the input means, the recommendation means, and the discrimination means are provided, it is possible to recommend appropriate music for each user. It becomes possible. According to the music recommendation method of the present invention, the range setting step, the input step, the recommendation step, and the discrimination step are provided, so that the effect exhibited by the music recommendation device can be easily realized.
- FIG. 1 is a block diagram of an in-vehicle electronic device according to an embodiment of the present invention.
- FIG. 2 is a schematic diagram of a music table stored in a memory in the in-vehicle electronic device of FIG.
- FIG. 3 is a schematic diagram of a search table stored in a memory in the in-vehicle electronic device of FIG.
- FIG. 4 is a schematic diagram of a search coordinate plane related to music recommendation processing.
- FIG. 5 is a schematic diagram of a search coordinate plane on which music Mi related to music recommendation processing is arranged.
- FIG. 6 is a flowchart of music recommendation processing.
- FIG. 7 is another schematic diagram of a search coordinate plane related to music recommendation processing.
- [0065] 10 Electronics for vehicle installation, 20 ... Bus line, 100 ... Navigation processing system, 200 ... Communication system, 210 "-MCU, 220 ... extraction unit, 230" CDZDVD drive, 240 ... Wireless communication device, 250 ... Display unit, 251 ... Display control unit, 252 ... Graphic controller, 253 ⁇ Noffer memory, 260... Input device, 270 “Memory, 280 ... Speech recognition processing unit, 290 ... Compression / decompression processing unit, 300 ... Audio processing system, 310 ... Hard disk device, 320 ... Sound output unit, 321 -DZA section, 322 ... variable amplifier, 323a, 323b ... speaker.
- FIG. 1 is a block diagram of the in-vehicle electronic device 10.
- an in-vehicle electronic device 10 is configured as an in-vehicle electronic device that integrates a vehicle-mounted audio unit and a navigation unit and further captures music data and a navigation map through a communication network.
- a large number of music pieces for example, data compression music: MPEG-1 Audio Layer-3Z commonly known as MP3 are stored in advance in a hard disk device (HDD) by processing on the user side.
- a large number of compressed music data downloaded from a music website on a communication network (especially the Internet) is stored on a node disk by a process on the user side.
- the in-vehicle electronic device 10 includes a navigation processing system 100, a common processing system 200, and an audio processing system 300.
- the navigation processing system 100 also includes a self-contained positioning device and a GPS receiver.
- this type of in-vehicle electronic device 10 is provided with an input / output (I / O) circuit, an external interface (IZF) unit, and the like, which are not shown.
- the common processing system 200 includes an MCU (Micro Controller Unit) 210, an extraction unit 220, a CD / DVD drive 230, a wireless communication device 240, a display unit 250, an input device 260, a memory 270, a voice recognition A recognition processing unit 280 and a compression / decompression processing unit 290.
- MCU Micro Controller Unit
- the audio processing system 300 is composed of a hard disk device (HDD) 310 and an audio output unit 320.
- HDD hard disk device
- the MCU 210 is a system controller that controls the entire apparatus, and executes a music recommendation process described later.
- the MCU 210 includes a CPU 211, a ROM 212 that is a nonvolatile solid-state storage element, and a working RAM 213, and exchanges data with each unit connected to the bus line 20. Processing control by this exchange of data is executed by a boot program and a control program stored in the ROM 212.
- the RAM 213 performs a working process for temporarily storing processing data of the CPU 211 by a user operation from the input device 260.
- the CPU 211 functions as an example of each of the “range setting unit”, “recommendation unit”, and “discrimination unit” according to the present invention.
- the extraction unit 220 is an example of the “extraction unit” according to the present invention.
- the extraction unit 220 displays the characteristics of the played song as a feature item song. It can be extracted as a feature quantity. Characteristic items include, for example, at least part or all of items representing the rate of chord change, the number of beats per minute, the maximum beat amplitude, the average tone signal intensity, and the maximum tone signal amplitude. It refers to at least one index that can quantitatively express the emotional characteristics.
- the feature items are assumed to be two types of feature item forces, “first feature item” and “second feature item”, in order to simplify the explanation.
- the music feature amount extracted by the extraction unit 220 is stored as a music table in the memory 270 which is a rewritable recording medium. The music table will be described later.
- the CDZDVD drive 230 is a common drive that reads music data, video data, text data, map data, and the like from the CD 231 or the DVD 232. Note that the CDZDVD drive 230 may be provided individually for each medium, or only one of them may be provided. However, do not install a large-capacity storage device such as the CDZDVD drive 230 when capturing music data, etc. by communication via the wireless communication device 240. It is also possible to do.
- the hard disk device (HDD) 310 stores a large number of songs (compressed music data) as described above. That is, the user stores a large number of compressed music data obtained by downloading via the wireless communication device 240, for example.
- the CDZDVD drive 230 stores a large number of music data read from the CD231 or DVD232.
- the node disk device 310 stores music data, video data, text data, map data, and the like previously read by the CDZDVD drive 230 by user processing, and can be read at any time after this storage.
- the audio data and video stored in the hard disk device 310 can be read and audio output and video output can be performed.
- audio data, video data, map data, etc. downloaded by the wireless communication device 240 in advance by the user side processing are stored in the node disk device 310 and read out at an arbitrary time thereafter. Output is possible.
- the wireless communication device 240 takes (receives) music data on a communication network into the hard disk device 310, and has the same configuration as a general-purpose mobile phone.
- PDC Personal Digital and ellular Telecommunication System
- PH3 ⁇ 4 Personal Handyphone System
- TDMA Time Division Multiplexing unit
- TDD Time Division Multiplexing unit
- CDMA configuration high frequency radio transceiver, encoding / decoding unit, time division multiplexing unit, control unit, This is a configuration known as a voice input / output unit or the like.
- the display unit 250 displays various processing data on the screen under the control of the MCU 210.
- the display unit 250 controls each unit of the display control unit 251 based on control data transferred from the CPU 211 through the internal graphic controller 252 power bus line 20.
- a buffer memory 253 using V-RAM or the like temporarily stores displayable image information.
- the display control unit 251 performs display control.
- the image data output from the display 254 force graphic controller 252 constituted by a liquid crystal display (LCD), an EL (Electro-Luminescence) display, or a cathode ray tube (CRT) is displayed on the screen.
- This The display 254 is disposed, for example, in the vicinity of the front panel in the vehicle.
- an internal DZA converter 321 converts a digital audio signal into an audio analog signal under the control of the MCU 210, and a variable amplifier (AMP) 322 variably amplifies the input operation by a user. Output to speakers 323a and 323b.
- AMP variable amplifier
- the input device 260 is an example of the "input means" according to the present invention, which includes various keys, button switches, remote controllers, and the like for inputting various commands and data.
- the voice recognition processing unit 280 is composed of a digital signal processor (DSP) or the like, and transfers various commands and data identifying the voice signal of the voice input operation (for example, music playback instruction input) to the MCU 210. To do.
- music playback on / off, music selection input instructions, and music playback skip instructions on or after music selection list can be performed by voice recognition (eg, linear prediction spectrum analysis). It has been.
- the compression / decompression processing unit 290 is configured by a digital signal processor (DSP) or the like, compresses music data when stored in the audio disk device 310, and decompresses (demodulates, for example, the above-described data). MPEG-1 method).
- DSP digital signal processor
- the speech recognition processing unit 280 and the compression / decompression processing unit 290 are not configured with a digital signal processor (DSP) or the like, but are installed with dedicated utility software, and this software (program) is loaded into the MCU 210. May be executed to perform the same processing.
- DSP digital signal processor
- FIG. 2 is a schematic diagram of the music table 271
- FIG. 3 is a schematic diagram of the search table 272.
- the memory 270 may be configured to function as an example of each of the “first storage unit” and the “second storage unit” according to the present invention.
- the music table 271 is a music feature amount in the first feature item and the second feature item by the extraction unit 220 among the plurality of songs stored in the hard disk device 310.
- the music feature amount in the first feature item is "0.1"
- the second feature item The music feature amount in is given as “0.5”.
- these music feature values are music feature values as a result of normalization processing performed by the MCU 210 on each music feature value extracted by the extraction unit 220. That is, in the present embodiment, real values from 0 to 1 are assigned to the music feature quantities of the feature items.
- the search table 272 includes a plurality of preset search keywords Ki.
- This is an updatable LUT in which the music feature quantity of the feature item is assigned to (i l, 2, ⁇ , n) in the same way as the music table 271.
- this music feature amount is not extracted by the extraction unit 220 but is stored in advance in the ROM 212 in the MCU 210 as a fixed value. That is, the MCU 210 reads these music feature quantities from the ROM 212 and stores them in the memory 2.
- a search table 272 is generated on 70.
- Search keywords are words that indicate the sensual features of a song, such as “fun”, “energetic”, “lively”, “good”, and the song features of each feature item assigned to these search keywords.
- the amount is a music feature amount most applicable to the emotional feature represented by these search keywords.
- the music feature amount assigned to these search keywords is given an optimum value in advance by a technique such as experimental, empirical, or simulation.
- FIG. 4 is a schematic diagram of the search coordinate plane.
- the search coordinate plane is a coordinate plane having two types of feature items as the vertical axis and the horizontal axis, and the music feature amount in each corresponds to the coordinate value.
- the search coordinate plane is an example of an “n-dimensional coordinate system” according to the present invention.
- each search keyword can be arranged at a coordinate point corresponding to the music feature amount assigned to each search keyword.
- the coordinate point corresponding to each search keyword is an example of the “reference coordinate point” according to the present invention.
- the coordinate values on the horizontal axis and the vertical axis are 0 respectively.
- the coordinate points corresponding to 5 and 0.6 are the reference coordinate points.
- This reference coordinate point is the coordinate point that most appropriately indicates the sensibility feature indicated by each search keyword in the search coordinate plane, but the area around this reference coordinate point is equally sensitive. Fit for features It is clear visually that this is considered.
- a circle centered on the reference coordinate point in each search keyword Ki is defined, and a range corresponding to the circle is a range that matches the search keyword Ki. That is, the circle set for each search keyword Ki is a “match range” as an example of the “first range” according to the present invention. “Radius R” stored in the search table 272 represents the radius of this matching range.
- the radius R of the matching range is set to be updatable on the search table 272, and its initial value is stored in the ROM 212 when the vehicle-mounted electronic device 10 is shipped from the factory, for example.
- the MCU 210 When the MCU 210 generates the search table 272, the MCU 210 reads the initial value from the ROM 212 and generates the initial search table 272.
- the search table 272 in FIG. 3 indicates that “0.2” is given as an initial value.
- FIG. 5 is a schematic diagram of a search coordinate plane on which the music piece Mi is arranged.
- the music piece Mi is arranged at the coordinate point corresponding to the music feature amount extracted by the extraction unit 220.
- the music Mn belongs to the matching range (radius Rn is the initial value) corresponding to the search keyword Kn.
- the music M3 belongs to the matching range corresponding to the search keyword K1
- the music M2 belongs to the matching range corresponding to the search keyword K3.
- the matching range corresponding to the search keyword K2 does not include any music.
- the music Ml is not included in the matching range corresponding to any search keyword.
- the music pieces that belong to the matching range are treated as music pieces that match the search keyword.
- the search keyword Kn is a keyword “fun music”, it means that the music Mn is a fun music.
- FIG. 6 is a flowchart of the music recommendation process realized by the MCU 210 executing the music recommendation program stored in the ROM 212. Note that Figure 6 It is assumed that the search keyword Kn is input via the input device 260!
- the CPU 211 refers to the music table 27 1 and the search table 272 stored in the memory 270 and selects “recommended range” from the music stored in the hard disk device 310.
- the music data is supplied to the audio output unit 320.
- This music is processed by the audio output unit 320 and emitted from the speakers 323a and 323b (step S10). At this point, the user views this music piece.
- a range in which the radius is further expanded by the margin r is a “recommended range” as an example of the “second range” according to the present invention.
- the value of the margin r is set to a value corresponding to 10% of the radius R of the matching range.
- the radius R of the mate range is the initial value “0.2”, so in FIG. 7, the margin r is set to “0.02”, and the recommended range is centered on the reference coordinate point.
- the corresponding range is within the circle with a radius of 0.22.
- the recommended range includes a matching range and is larger than the matching range
- songs that are considered not to match the search keyword Kn are included because of the concept of the search coordinate plane.
- the music Mn5 is included in the power recommendation range, which is a music that does not match the search keyword Kn, the CPU 211 recognizes this music Mn5 as a selection target.
- the CPU 211 sequentially selects and plays music pieces belonging to the recommended range. Note that the CPU 211 does not always need to select and play music from the recommended range when one search keyword is input from the user. For example, normally, music may be selected and played from music belonging to the matching range, and music belonging to the recommended range may be selected and played at an appropriate timing.
- step S11 The user evaluates the reproduced music by his / her own judgment, and feeds it back to the CPU 211 via the input device 260 or the voice recognition processing unit 280 (step S11).
- This feedback For example, when the user determines that the song does not match the search keyword Kn, the user selects and inputs a button switch indicating “non-match”. Or, if you decide that this song does not match the search keyword Kn, you can say “Skip”. If there is no positive feedback from the user, the CPU 211 determines that this song matches the user's sensitivity. Therefore, in step S11, “feedback” does not necessarily indicate an active response of the user side, but the process related to the music recommendation process continues. The feedback result by the user is temporarily stored in the RAM 213 regardless of whether it is active or inactive.
- the CPU 211 determines whether or not the music being played is a music within the first update range (step S12). If the music is not within the first update range (step S12: NO), the CPU 211 further determines whether or not the music being played is a music within the second update range (step S15).
- the first update range refers to the range defined by “R ⁇ S ⁇ R + r”, where S is the distance of the reference coordinate point force.
- the second update range refers to the range defined by the distance force “R — r ⁇ S ⁇ R” from the reference coordinate point.
- the music belonging to the first update range is music Mn5
- the music belonging to the second update range is music Mn4.
- the matching range corresponding to the search keyword input reflecting the user's sensitivity is updated.
- step S12 when the music being played is within the first update range (step S12: YES), the CPU 211 obtains the user feedback result from the RAM 213, and this music matches the user's sensitivity. It is determined whether or not the force is present (step S13). At this time, if the user feedback result is not stored in the RAM 213, the CPU 211 stops the process until the result is stored. The time when the CPU 211 determines that the music being played matches the user's sensibility may be set to any time.
- step S13 If the music being played matches the user's sensibility (step S13: YES), the CPU 211 updates the matching range corresponding to the search keyword Kn in the enlargement direction (step S14). In other words, the radius R of the matching range is updated so as to increase.
- the CPU 211 updates the value of the radius R according to the following formula (1).
- the updated match range more specifically, the updated radius R value is stored in the search table 272 as the latest radius R for the search key Kn.
- Rn is the value of radius R at the time of the n ⁇ first update
- Rn + 1 is the value of radius R after the nth update
- ⁇ is an update coefficient, which is defined in a range of “0 ⁇ ⁇ 1,” and is set to “0. 05” in the present embodiment.
- D represents the distance to the music being played in the reference coordinate point force on the search coordinate plane (see Fig. 7). For example, if the music being played is the music Mn5 in FIG. 7, and if this music matches the user's sensitivity, the matching range for the search keyword Kn is updated according to the above equation.
- step S13 NO
- the CPU 211 returns to step S10 without updating the matching range, and the next music Start playback.
- the reason why the update is not performed is that the song Mn5 is originally a song that does not match the search keyword Kn, so if the user determines that it does not match, the match range corresponding to this search keyword Kn is correct. Yes, there is no need for renewal.
- step S12 when the music being played does not belong to the first update range (step S12: NO), the process related to step S15 is performed. If the song being played is within the second update range (step S15: YES), the CPU 211 obtains the user feedback result from the RAM 213, and whether or not this song does not match the user's sensitivity. Is determined (step S16). This At this point, if the user feedback result is stored in the RAM 213, the CPU 211 stops the process until the result is stored as in step S13.
- step S16 If the music being played does not match the user's sensibility (step S16: YES), the CPU 211 updates the matching range corresponding to the search keyword Kn in the reduction direction (step S17). Also at this time, the matching range is updated based on the above equation (1). For example, if the music being played is the music Mn4 in FIG. 7, and if this music does not match the user's sensitivity, the matching range for the search keyword Kn is updated according to the above formula.
- step S16 If the music being played matches the user's sensibility (step S16: NO), the CPU 211 does not update the matching range corresponding to the search keyword Kn and proceeds to step S10. Go back and start playing the next song.
- step S15 If the music being played is not within the second update range (step S15: NO), the matching range is not updated, and the process proceeds to step S10.
- the music being played is the music Mnl, Mn2, or Mn3
- the distance from the reference coordinate point is smaller than “R ⁇ r”, so that the user feedback indicates The mate range is not updated.
- the matching range is updated, or the power of the music being played does not belong to the first and second update ranges, and the music being played is within the first update range. If the music that does not match the user's sensibility, or the song being played is within the second update range and matches the user's sensibility, the process related to the currently playing song ends. Then, the CPU 211 returns the process to step S10 and starts playing the music having the next playing order.
- the in-vehicle electronic device 10 is configured to be able to update the matching range for the search keyword as described above. Therefore, different sensibility can be reflected for each user, and appropriate music can be recommended for each user.
- the matching range is determined based on the initial value
- the music Mn6 and Mn7 are outside the recommended range, and the device is used when the search keyword Kn is input.
- the music is not recommended by the side.
- One user can search If the keyword Kn has a wide sensitivity, that is, if the user feels that most songs match the search keyword ⁇ , the matching range update process based on the above equation (1) During several updates, these songs may fall within the recommended range. Conversely, if another user has a narrow sensitivity to the search keyword ⁇ , that is, if the user feels that most of the songs do not match the search keyword ⁇ , it will be updated several times. Sometimes the music Mnl, ⁇ 2 and ⁇ 3 may leak the recommended range power. Therefore, even if the initial state is the same, the sensitivity of each user is sufficiently reflected as time passes.
- the coordinate system according to the present invention can also adopt a three-dimensional coordinate space, a multidimensional coordinate space, or a one-dimensional coordinate axis.
- the matching range is set as a circle centered on the reference coordinate point.
- the shape of the force matching range can be determined to match the emotional characteristics indicated by the search keyword. Any shape can be used as long as it is within the range. For example, it may be oval.
- the music recommendation apparatus and method according to the present invention can be used for, for example, car audio and stereo that can provide music according to the user's sensitivity from among a large number of music.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Library & Information Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
Description
Claims
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2006535849A JPWO2006030712A1 (ja) | 2004-09-17 | 2005-09-09 | 楽曲推薦装置及び方法 |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2004-271287 | 2004-09-17 | ||
JP2004271287 | 2004-09-17 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2006030712A1 true WO2006030712A1 (ja) | 2006-03-23 |
Family
ID=36059965
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2005/016642 WO2006030712A1 (ja) | 2004-09-17 | 2005-09-09 | 楽曲推薦装置及び方法 |
Country Status (2)
Country | Link |
---|---|
JP (1) | JPWO2006030712A1 (ja) |
WO (1) | WO2006030712A1 (ja) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2012103832A (ja) * | 2010-11-09 | 2012-05-31 | Sony Corp | 情報処理装置および方法、情報処理システム、並びに、プログラム |
CN105608105A (zh) * | 2015-10-30 | 2016-05-25 | 浙江大学 | 一种基于收听上下文的音乐推荐方法 |
WO2017166132A1 (zh) * | 2016-03-30 | 2017-10-05 | 马岩 | 网络信息推送方法及*** |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH11184883A (ja) * | 1997-12-24 | 1999-07-09 | Canon Inc | 楽曲検索システム、楽曲検索方法及び記録媒体 |
JP2002183152A (ja) * | 2000-12-18 | 2002-06-28 | Jinyama Shunichi | 音楽検索装置及び方法並びに音楽検索用ソフトウェアを記録した記録媒体 |
JP2002278547A (ja) * | 2001-03-22 | 2002-09-27 | Matsushita Electric Ind Co Ltd | 楽曲検索方法、楽曲検索用データ登録方法、楽曲検索装置及び楽曲検索用データ登録装置 |
JP2004199437A (ja) * | 2002-12-19 | 2004-07-15 | Link Cube Kk | 検索装置 |
-
2005
- 2005-09-09 JP JP2006535849A patent/JPWO2006030712A1/ja active Pending
- 2005-09-09 WO PCT/JP2005/016642 patent/WO2006030712A1/ja active Application Filing
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH11184883A (ja) * | 1997-12-24 | 1999-07-09 | Canon Inc | 楽曲検索システム、楽曲検索方法及び記録媒体 |
JP2002183152A (ja) * | 2000-12-18 | 2002-06-28 | Jinyama Shunichi | 音楽検索装置及び方法並びに音楽検索用ソフトウェアを記録した記録媒体 |
JP2002278547A (ja) * | 2001-03-22 | 2002-09-27 | Matsushita Electric Ind Co Ltd | 楽曲検索方法、楽曲検索用データ登録方法、楽曲検索装置及び楽曲検索用データ登録装置 |
JP2004199437A (ja) * | 2002-12-19 | 2004-07-15 | Link Cube Kk | 検索装置 |
Non-Patent Citations (1)
Title |
---|
SATO S. ET AL: "Kanjo ni Motozuku Ongaku Sakuhin Kensaku System no Jitsugen ni Mukete no Kento (A Discussion about the Realization of Impression-based Retrieval System for Music Collection)", INFORMATION PROCESSING SOCIETY OF JAPAN KENKYU HOKOKU, NIPPON, INFORMATION PROCESSING SOCIETY OF JAPAN, vol. 2001, no. 16, 23 February 2001 (2001-02-23), pages 51 - 56, XP002998826 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2012103832A (ja) * | 2010-11-09 | 2012-05-31 | Sony Corp | 情報処理装置および方法、情報処理システム、並びに、プログラム |
CN105608105A (zh) * | 2015-10-30 | 2016-05-25 | 浙江大学 | 一种基于收听上下文的音乐推荐方法 |
WO2017166132A1 (zh) * | 2016-03-30 | 2017-10-05 | 马岩 | 网络信息推送方法及*** |
Also Published As
Publication number | Publication date |
---|---|
JPWO2006030712A1 (ja) | 2008-07-31 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP4037081B2 (ja) | 情報選択装置及び方法、情報選択再生装置並びに情報選択のためのコンピュータプログラム | |
US20060224260A1 (en) | Scan shuffle for building playlists | |
US7613736B2 (en) | Sharing music essence in a recommendation system | |
US9495449B2 (en) | Music steering with automatically detected musical attributes | |
KR100952186B1 (ko) | 악곡들을 식별하는 방법 | |
US20070157797A1 (en) | Taste profile production apparatus, taste profile production method and profile production program | |
EP1420388A1 (en) | Music searching method, music searching device, and music searching program | |
JP5594532B2 (ja) | 情報処理装置および方法、情報処理システム、並びに、プログラム | |
CN110033789A (zh) | 生物测定个性化音频处理*** | |
EP1965322A1 (en) | Information processing apparatus, information processing method, and information processing program | |
JP4322691B2 (ja) | 選曲装置 | |
JP5269829B2 (ja) | カラオケシステム、カラオケシステムの制御方法、及びカラオケシステムの制御プログラム並びにその情報記録媒体 | |
WO2006030712A1 (ja) | 楽曲推薦装置及び方法 | |
JP2010044756A (ja) | 情報処理装置、情報処理方法、及びプログラム | |
JP2005115164A (ja) | 楽曲検索装置 | |
JP2005346347A (ja) | 情報検索装置、情報検索方法、情報検索プログラムおよび記録媒体 | |
JP3714865B2 (ja) | カラオケ自動選曲演奏システム | |
JPWO2006077769A1 (ja) | 提示データ選択装置及び提示データ選択方法等 | |
JP2005285285A (ja) | コンテンツ読出装置及び楽曲再生装置 | |
JP4279274B2 (ja) | 携帯端末装置および楽曲情報表示方法 | |
US20100058186A1 (en) | Reproducing apparatus and reproducing method | |
JP2007316830A (ja) | 情報処理装置及びコンテンツ検索プログラム | |
JP2006323438A (ja) | 楽曲検索システム | |
JP2017084313A (ja) | プレイリスト生成方法、プレイリスト生成装置、及びプログラム、並びに再生リスト生成方法 | |
JPH11296181A (ja) | 音楽再生装置 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BW BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EC EE EG ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KM KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NA NG NI NO NZ OM PG PH PL PT RO RU SC SD SE SG SK SL SM SY TJ TM TN TR TT TZ UA UG US UZ VC VN YU ZA ZM ZW |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): BW GH GM KE LS MW MZ NA SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LT LU LV MC NL PL PT RO SE SI SK TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
DPE1 | Request for preliminary examination filed after expiration of 19th month from priority date (pct application filed from 20040101) | ||
WWE | Wipo information: entry into national phase |
Ref document number: 2006535849 Country of ref document: JP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |