JP3074693B2 - Music evaluation device - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a music evaluation apparatus suitable for singing ability evaluation and musical instrument performance evaluation, and more specifically, music evaluation incorporating evaluation criteria of a music expert. Related to the device.

<Prior Art> In an electronic keyboard musical instrument, a model performance sound is generated, and thereafter, the performance data obtained by performing a keyboard operation by a player by imitating the model performance sound is taken in, and compared with reference data to display a score.
For example, there are JP-A-58-132797 and JP-A-58-132798.

In recent years, there has been known a karaoke system which evaluates singing information taken from a microphone and displays a score.

<Problems to be Solved by the Invention> These conventional music evaluation devices incorporated in an electronic musical instrument or karaoke are designed to compare performance data or singing information with reference data in order, and use the score as a score for the degree of coincidence. However, at the same time, there was a problem that the evaluation criteria were not very bland and different from the evaluation criteria of actual music experts.

In general, it is impossible for a human to calculate the degree of coincidence with reference data as precisely as a machine, but a music expert analyzes a large number of parameters based on experience and perception. Make an evaluation.

The present invention has been made in view of the above circumstances, and has as its object to provide a music evaluation device that can incorporate evaluation criteria of a music expert.

<Means for Solving the Problems> In order to achieve this object, a music evaluation device according to the present invention includes: an acoustic information capturing unit; an internal or external storage unit that stores at least accompaniment information, teacher data, and fuzzy weight data; Converting means for converting the sound information fetched by the sound information fetching means into music data of the same dimension as the teacher data; forming music evaluation data by applying the music data, teacher data and fuzzy weight data to a predetermined fuzzy rule Fuzzy inference means, and display means for performing display corresponding to the music evaluation data, wherein the fuzzy weight data is set for each music piece or each music genre.

<Operation> According to the above configuration, acoustic information of singing or playing a musical instrument is taken in along with the accompaniment, and the taken-in acoustic information is given to the fuzzy inference means together with the teacher data and the fuzzy weight data. Can be The fuzzy inference means forms music evaluation data based on the given information. Since the fuzzy inference means can prepare a plurality of evaluation criteria in a rule format, general evaluation criteria of a music expert are adopted. Furthermore, since the fuzzy weight data can be stored in the internal or external storage means corresponding to the music, an evaluation criterion corresponding to the music or the genre of the music can be adopted, and the evaluation according to the music can be performed.

<Embodiment> An embodiment of the present invention will be described below with reference to the drawings. In this embodiment, the music evaluation device of the present invention is applied to a karaoke system.

First, the outline will be described with reference to FIGS. 3 to 5.

In FIG. 3, reference numeral 1 denotes an optical disk, and accompaniment information and video information corresponding to the music are stored for each music. In addition, score section data, teacher data, and fuzzy weight data are stored for each song. The scoring section data is data for extracting a section that particularly affects the singing ability evaluation throughout the music, and may be four measures or eight measures. In some cases, the entire song may be evaluated. The teacher data is data serving as a reference for singing ability evaluation, and is reference data in a marking section for each of the five parameters as shown in FIG. In this embodiment, five parameters of a sound length, a pitch difference, a pitch, a level difference, and vibrato are evaluated. As apparent from FIG. 4, the pitch is a parameter representing the duration of a sound such as a quarter note, an eighth note, etc., and the pitch difference is a relative pitch frequency difference from the preceding sound. The pitch is a parameter representing an absolute pitch frequency, the level difference is a parameter representing a relative level difference from the previous sound, and the vibrato is a sound having a predetermined amplitude and a predetermined period. This is a parameter representing a high frequency change. The fuzzy weight data is data representing weighting coefficients of a plurality of fuzzy rules used in the music evaluation device 4. Reference numeral 2 in FIG. 3 denotes a pickup for extracting these data from the optical disc 1 which is driven to rotate. Among the extracted data, accompaniment information and video information are supplied to the AV system 5, where the scoring section data, the teacher data The fuzzy weight data is supplied to the music evaluation device 4. Reference numeral 3 in FIG. 3 denotes a microphone, which captures singing information, and supplies them to the music evaluation device 4 and the AV system 5. Music evaluation device 4
Forms music evaluation data by fuzzy inference based on given information. The AV system 5 includes an amplifier, a speaker, a display, and the like. The AV system 5 emits accompaniment information and singing information through the amplifier and the speaker, and displays video information on the display.

Here, the principle of the music evaluation device 4 of FIG. 3 will be described.
First, the singing information taken in from the microphone 3 is extracted for the scoring section data and converted into a plurality of parameters of the same dimension as the teacher data. Then, these parameters are compared with the teacher data. This comparison result is normalized based on the length of the scoring section data, and is applied to the following fuzzy rules for each parameter.

(1) If the pitch matches well, it is an advanced person. If it hardly matches, it is a beginner. (2) If the pitch difference matches well, it is an advanced person. If it hardly matches, it is a beginner. (3) The pitch is good. If they match, they are advanced. If they hardly match, they are beginners. (4) Level differences If they match well, they are advanced. If they hardly match, they are beginners. (5) Vibrato. If they match well, they are advanced. If there is almost no match, it is a beginner. The fitness of each of the above fuzzy rules is combined after weighting according to the fuzzy weight data. The weighting coefficient between the fuzzy rules of each parameter is determined for each music genre, for example, as shown in FIG. In this embodiment, enka has a high importance of vibrato, and rock has a high importance of the level difference. Instead of setting the weighting coefficient for each genre of the music as described above, a different weighting coefficient may be set for each music within the same genre.

Next, the principle of fuzzy inference performed in accordance with these fuzzy rules and fuzzy weight data will be described with reference to FIG. FIGS. 6 (a) and 6 (b) show the degree of coincidence between the singing information and the teacher data applied to the fuzzy rule "If they match well, then you are an expert."
(A) is a membership function of the antecedent part "good match", and the horizontal axis represents the degree of coincidence (the degree of coincidence is closer to 0 and higher as it is closer to 1), and the vertical axis is the grade.
(B) is a membership function of the consequent part "I am an advanced person", in which the horizontal axis represents the degree of goodness (goodness) and the vertical axis represents the grade. Here, if the degree of coincidence between the singing information and the teacher data is 0.4, a grade 0.75 is obtained from the membership function (a) of the antecedent part, and the 0.75 indicates that the membership function (b) of the consequent part is obtained. Sharp head. The remaining area (shaded area in FIG. 6 (b)) is weighted according to the fuzzy weight data to obtain FIG. 6 (c).
FIGS. 6 (d), (e), and (f) apply the degree of coincidence between the singing information and the teacher data to the fuzzy rule "If there is almost no coincidence, a beginner" in the same manner. Such inference is made for all fuzzy rules. FIG. 6 (c) thus obtained,
The area as shown in (f) is synthesized as shown in FIG. 6 (g), and the center of gravity is obtained. The abscissa of the center of gravity is the final inference result. It should be noted that such a fuzzy inference method may use any known method.

Next, a more detailed hardware configuration of a karaoke system to which the music evaluation device of the present invention is applied will be described with reference to FIG. In the figure, the same blocks as those in FIG. 3 are denoted by the same numbers. Reference numerals 1 and 2 denote playback devices including an optical disc and a pickup, which perform a playback operation under the control of the CPU 28 and send accompaniment information and video information to the AV system 5. Also, prior to the reproducing operation, the CPU 28 captures the scoring section data, the teacher data, and the fuzzy weight data. Reference numeral 3 denotes a microphone that converts sound information into an electric sound signal. Reference numeral 11 denotes an amplifier circuit for amplifying an input signal. Numerals 12 to 25 denote circuits for converting the acoustic signal sent from the amplifier circuit 11 into music evaluation parameters of pitch, pitch difference, pitch, vibrato and level difference, which are parameters of the same dimension as the teacher data. Reference numeral 12 denotes an analog / digital conversion circuit for converting an audio signal provided in an analog form into a digital audio signal.
The sampling frequency of 44.1 kHz and the number of quantization bits of 16 bits, which are the specifications of (compact disc), can be adopted. Reference numeral 13 denotes a pitch extraction circuit for extracting a pitch (frequency) from a digitized sound signal, and any method such as a method using autocorrelation and a method using Fourier analysis can be adopted. 14
Is a circuit for converting a pitch signal into a scale frequency signal.
This circuit includes a plurality of comparison circuits and a representative value output circuit, and converts a continuous pitch signal into a stepwise frequency signal corresponding to the equal temperament scale and outputs the signal. A circuit 15 detects a change in the scale frequency signal. A counter 16 is reset by a given change detection signal, and measures a time interval between the change detection signals. Reference numeral 17 denotes a latch circuit that captures the count value of the counter 16 at a change detection signal generation timing given from the change detection circuit 15. This latch circuit
The output of 17 is taken in by CPU 28 via interface 27 as a tone length parameter. 18 is change detection circuit 1
5 is a latch circuit which takes in a pitch signal given from the pitch extracting circuit 13 at a change detection signal generation timing given from 5. A subtraction circuit 19 subtracts the output of the latch circuit 18 supplied to the-input terminal from the output of the pitch extraction circuit 13 supplied to the + input terminal, and outputs the result. The output of this subtraction circuit is used as a pitch difference parameter in interface 27.
Is taken in by the CPU 28 via the. Pitch extraction circuit 1
The pitch signal which is the output of 3 is taken in by the CPU 28 via the interface 27 as a direct pitch parameter. A circuit 20 detects the maximum value of the pitch signal supplied from the pitch extraction circuit 13 and holds the maximum value, and is reset by a change detection signal supplied from the change detection circuit 15. That is, the maximum value of the pitch signal between the time when the previous change detection signal was generated and the time when the current change detection signal was generated is detected. A circuit 21 detects the minimum value of the pitch signal supplied from the pitch extraction circuit 13 and holds the minimum value, and is reset by a change detection signal supplied from the change detection circuit 15. That is, the minimum value of the pitch signal between the time when the previous change detection signal is generated and the time when the current change detection signal is generated is detected. A subtraction circuit 22 subtracts the output of the minimum value detection / hold circuit 21 applied to the − input terminal from the output of the maximum value detection / hold circuit 20 applied to the + input terminal, and outputs the result. Reference numeral 23 denotes a latch circuit that captures the output of the subtraction circuit 22 at the timing of generation of a change detection signal given from the change detection circuit 15. The output of the latch circuit 23 is captured by the CPU 28 via the interface 27 as a vibrato parameter. In this embodiment, only the change width of the pitch signal is used as the vibrato parameter, but the change speed of the pitch signal may be used if necessary. The rate of change can be easily obtained by a method of measuring a time interval in which the pitch signal crosses a predetermined value (for example, scale frequency) twice. 24 is a circuit for extracting an envelope from the digitized sound signal given from the analog / digital conversion circuit 12,
It is composed of a digital low-pass filter and the like. twenty five
Is a latch circuit that takes in the output of the envelope detection circuit 24 at the change detection signal generation timing given from the change detection circuit 15. A subtraction circuit 26 subtracts the output of the latch circuit 25 supplied to the − input terminal from the output of the envelope extraction circuit 24 supplied to the + input terminal, and outputs the result. This subtraction circuit 26
Output of the interface 27 as a level difference parameter
Is taken in by the CPU 28 via the. Reference numeral 27 denotes an interface composed of a buffer register and the like. 28 is RO
The playback devices 1, 2 according to the control program stored in M31
The CPU (central processing unit) executes a series of processes such as an operation command of the above and data acquisition via the interface 27. The main processing contents will be described later. Further, the CPU 28 executes an interrupt program when a change detection signal is given from the change detection circuit 15. Reference numeral 29 denotes a RAM (random access memory) for temporarily storing and storing data. 30
Is a ROM (read only memory) that stores data such as control programs and fuzzy rules in advance. Reference numeral 31 denotes a display device for displaying music evaluation data and the like. 32
Is an operation panel having operators such as switches and volume. 33 is a data and address bus.

Next, the operation of the karaoke system will be described with reference to the flowchart of FIG.

FIG. 2A shows a main flow. When the power of the karaoke system is turned on, the CPU 28 first determines whether or not the music selection switch on the operation panel 32 is pressed according to a control program in the ROM 30. Perform (step 41). At this time, if the power has just been turned on and no switch has been operated, the process proceeds to step 42 according to the determination of NO. In step 42, a parameter setting operator such as a volume setting operator is scanned to set various parameters. Steps 41 and 42 are repeated until the music selection switch is pressed. In this state, if the music selection switch is pressed, the process proceeds to step 43 according to the determination of YES in step 41. In step 43, the optical disc 1 is driven to rotate, and the pickup 2 is positioned at a predetermined track (track storing scoring section data, teacher data, and fuzzy weight data) in an area corresponding to music selection.
In step 44, the scoring section data, teacher data, and fuzzy weight data are fetched and stored in a predetermined area of the RAM 29. In step 45, the process waits for the reproduction switch on the operation panel 32 to be pressed. In this state, if the reproduction switch is pressed, YES in step 45
In accordance with the determination of, the process proceeds to step 46 to start reproduction. At this time, the singer sings according to the accompaniment and the video (including lyrics) to be reproduced. In step 47, the scoring section data fetched in step 44 is compared with a time code given according to the progress of the music, and it is determined whether or not the scoring section has been completed. If the scoring section has not been completed yet, the process proceeds to step 48 according to the determination of NO in step 47. In step 48, it is determined whether or not the music has ended. Since the scoring section is set at a predetermined part of the song, immediately after step 47 (without going through step 51)
If the process has proceeded to step 48, it is impossible that the music has been completed, so the process proceeds to step 49 according to the determination of NO in step 48. In step 49, it is determined whether or not the stop switch on the operation panel 32 has been pressed.
At this time, if the stop switch has not been pressed, the determination of steps 47 and 48 is repeated according to the determination of NO.
Next, when the process proceeds to step 49, if the stop switch has been pressed, the process proceeds to step 50 according to the determination of YES, and the reproducing operation ends. Steps 49 and 50 make it possible to forcibly stop the reproducing operation even during the reproduction of the music. If the scoring section has ended in step 47, the process moves to step 51 according to the determination of YES. Completion of the scoring section means that all of the evaluation parameters of the sound length, the high pitch difference, the pitch, the level difference, and the vibrato have been captured by the interrupt processing described later. In step 51, the flag F is determined. The flag F indicates whether or not music evaluation based on fuzzy inference has already been performed in the subsequent step 52. When the process first proceeds to step 51, F = 0 is set. Move to 52. In step 52, music evaluation based on the above-described fuzzy inference is performed, and the obtained evaluation data is stored in the RAM 29. In step 53, the flag F is set to 1. This is to prevent the same processing from being repeated many times. When the process proceeds to step 51 again after step 47, the process proceeds to step 48 according to the determination of YES. At this time, if the music has been completed, the processing shifts to step 54 to end the reproduction operation. In step 55, the already obtained evaluation data is displayed on the display device 31 in a point format. In step 56, the flag F is reset. In this example,
The score is displayed after the song ends,
This is because the singer does not want to sing in the middle while singing is displayed. After that, the process returns to step 41 to wait for the next music selection. The above processing is repeated.

FIG. 2 (b) shows an interrupt program. When the CPU 28 receives a change detection signal from the change detection circuit 15, the interrupt program is executed prior to the processing performed so far. In step 60, the scoring section data is compared with the time code to determine whether or not it is a scoring section. At this time, if it is not a scoring section, the process immediately returns according to the determination of NO, but if it is a scoring section, the process proceeds to step 61 according to the determination of YES. In step 61, the evaluation parameters of pitch, pitch difference, pitch, level difference, and vibrato are fetched and stored in a predetermined area of the RAM 29.

The example in which the music evaluation device of the present invention is applied to the karaoke system has been described above, but the present invention can be implemented in various other forms. Modifications are listed below.

The karaoke software is not limited to an optical disk, but may be a magnetic disk, a magnetic tape, a semiconductor memory, or the like.

Further, the karaoke software may be separate from the karaoke system and exchanged according to the music, or may be incorporated in the karaoke system.

The fuzzy rules to be applied are not limited to those of this embodiment, and various rules can be applied. Different fuzzy rules may be stored in the karaoke software for each song.

Evaluation parameters also include pitch, pitch difference, pitch, level difference,
The quality of the voice (tone color) is not limited to the five types of vibrato, but may be obtained by frequency analysis and used as an evaluation parameter.

In the embodiment, the processing performed by hardware may be replaced by software, and the processing performed by software may be replaced by hardware.

This music evaluation device is not limited to singing power evaluation of karaoke, but may be used for musical instrument performance evaluation and the like.

<Effects> As described above, the music evaluation device of the present invention can perform music evaluation that meets the evaluation criteria of an actual music expert by introducing fuzzy inference. In particular, since an evaluation criterion corresponding to a tune or a genre of a tune is used, high-quality evaluation according to a tune to be played can be performed.

[Brief description of the drawings]

FIG. 1 is a hardware configuration diagram of an embodiment in which the present invention is applied to a karaoke system. FIG. 2 is a flowchart for explaining the operation of the present invention. FIG. 3 is a diagram for explaining the outline of the present invention. FIG. 4 is a diagram for explaining evaluation parameters used in the embodiment of the present invention. FIG. 5 shows an example of fuzzy weight data. FIG. 6 is a diagram for explaining the principle of fuzzy inference. 1. Optical disc 2. Pickup 3. Microphone 4. Music evaluation device 5. AV system 12 to 26. As a whole, a conversion means for converting acoustic information into music data of the same dimension as teacher data is provided. Show. 28 CPU (central processing unit) 29 RAM (random access memory) 30 ROM (read only memory) 31 display device

Continuation of the front page (56) References JP-A-3-84599 (JP, A) JP-A-58-132797 (JP, A) JP-A-2-139763 (JP, A) JP-A-2-272637 (JP, A) JP-A-2-272636 (JP, A) JP-A-3-85580 (JP, A) Patent 2838602 (JP, B2) (58) Fields investigated (Int. Cl. ⁷ , DB name) G10L 11 / 00-13/08 G10L 19/00-21/06 G09B 15/00 G10K 15/04 302 G10L 15/08 G10L 15/10 JICST file (JOIS)

Claims (1)

(57) [Claims] 1. An audio information capturing means, an internal or external storage means storing at least accompaniment information, teacher data and fuzzy weight data, and a music piece having the same dimension as the teacher data as the audio information captured by the audio information capturing means. Conversion means for converting the music data, teacher data and fuzzy weight data into predetermined fuzzy rules to form music evaluation data; display means for performing display corresponding to the music evaluation data; The fuzzy weight data is set for each song or for each genre of the song.