JP3707135B2 - Karaoke scoring device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a karaoke scoring device that evaluates the singing ability of a singer in a karaoke system.
[0002]
[Prior art]
This type of conventional karaoke scoring device is a model vocal signal extracted as a singing voice from a recording medium such as a laser disc (LD) or a compact disc (CD), and an actual singing by a karaoke singer. It compares the pitch (pitch), volume, and their changes with the microphone vocal signal, and scores the singer's singing ability based on the degree of coincidence. It is digital for sample vocal signal extraction processing, pitch extraction processing, etc. -A signal processor (DSP) was used.
[0003]
Using the DSP, the conventional karaoke scoring device extracts a sample vocal signal from digital musical sound signals passing through the DSP according to a predetermined program or data, and each of the extracted sample vocal signal and microphone vocal signal. Or extracting the pitch.
[0004]
[Problems to be solved by the invention]
However, in the karaoke scoring device according to the above-described conventional technique, the characteristics of the vocal signal are extracted by vocal signal extraction, pitch extraction, level detection, and the like, and these processes are performed on the sample vocal signal and the microphone vocal signal, respectively. Therefore, there is a problem that the processing amount of the entire algorithm increases and the configuration becomes complicated.
[0005]
In addition, the conventional karaoke scoring apparatus has a problem that the pitch cannot be extracted accurately when the accuracy of vocal signal extraction is not sufficient.
[0006]
Therefore, an object of the present invention is to provide a karaoke scoring device that has a simple configuration and can perform scoring similar to singing ability evaluation performed by a human sense.
[0007]
[Means for Solving the Problems]
The first invention for achieving the above object is the first wavelet transforming means for wavelet transforming a model tone signal based on the singing and accompaniment of a model person, and the same accompaniment as the singing and accompaniment of the scoring target person The second wavelet transforming means for wavelet transforming the musical signal to be scored based on the above and the outputs of the first and second wavelet transforming means with the same frequency band By comparing each other, it is repeatedly calculated over time, Each degree of matching obtained Is divided by the number of comparisons to normalize the values so as to be within a predetermined range, and each normalized degree of coincidence and a value within the predetermined range are taken. Increase or decrease scoring value based on random number The increase / decrease value is obtained, and the increase / decrease value is accumulated to an initial value set in advance each time the degree of coincidence is obtained, and the accumulated value is output as a scoring value. And scoring means.
[0008]
The second invention is a first wavelet transforming means for wavelet transforming a model vocal signal based on a singer of a model person, and a second wavelet for wavelet transforming a vocal signal to be scored based on a singer of a score target person. The matching frequency between the transforming means and the outputs of the first and second wavelet transforming means is the same frequency band. By comparing each other, it is repeatedly calculated over time, Each degree of matching obtained Is divided by the number of comparisons to normalize the values so as to be within a predetermined range, and each normalized degree of coincidence and a value within the predetermined range are taken. Increase or decrease scoring value based on random number The increase / decrease value is obtained, and the increase / decrease value is accumulated to an initial value set in advance each time the degree of coincidence is obtained, and the accumulated value is output as a scoring value. And scoring means.
[0009]
When the first converting means performs wavelet transform on the model vocal signal or model musical sound signal and the second converting means performs wavelet transform on the scoring target vocal signal or scoring target musical sound signal, the scoring means is the first and second converting means. The degree of coincidence is obtained for each frequency band that is the same between the outputs of No., and the scoring value for the singing of the scoring target person is obtained based on the obtained degree of coincidence.
[0010]
The wavelet transform output is divided into multiple frequency band signals with wider bands in the higher frequency range, and power signals in each band can be easily obtained, eliminating the need for pitch extraction algorithms and simplifying the configuration. It becomes. In addition, since the method of band division by (orthogonal) wavelet transform matches the human auditory characteristics, human scoring is possible.
[0011]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
[0012]
(First embodiment)
First, the first embodiment according to the present invention will be described.
[0013]
FIG. 1 is a block diagram of a karaoke system showing an application example of a karaoke scoring device according to an embodiment of the present invention.
[0014]
This karaoke system includes a karaoke scoring device 1A according to the first embodiment of the present invention and a karaoke processing unit 2.
[0015]
Here, in this karaoke system, the karaoke scoring device 1A is a digital sample musical tone signal comprising a model vocal signal Vr based on a model person's singing recorded in stereo on a compact disc (CD) 3 and an accompanying musical tone signal A ( Vr + A) can be input, the left channel signal L of the model tone signal (Vr + A) reproduced from this CD3 is input to the left channel digital input terminal 4a, and the right channel signal R is the right channel digital input terminal. 4b is input.
[0016]
The karaoke scoring device 1A includes a first mixing circuit 5a and a voice cancel unit 6 connected to both input terminals 4a and 4b, and an analog scoring target vocal based on the singing of a singer who is a scoring person from the microphone 7. Connected to an analog input terminal 8 for inputting a signal Vm, an A / D converter 9 for converting an analog scoring target vocal signal Vm from the analog input terminal 8 into a digital signal, a voice cancel unit 6 and an A / D converter 9 The second mixing circuit 5b, the first wavelet transform unit 10a connected to the first mixing circuit 5a, the second wavelet transform unit 10b connected to the second mixing circuit 5b, and the first And the second wavelet transform units 10a and 10b, and the scoring value is obtained based on the outputs of the first and second wavelet transform units 10a and 10b. Mel scoring unit 11, is connected to the scoring unit 11, and a such display unit 12 of a CRT display for displaying a scoring value by scoring unit 11.
[0017]
The karaoke processing unit 2 includes an echo processing unit 20 that gives an echo to the output of the A / D converter 9, a karaoke signal processing unit 21 that inputs the output of the voice canceling unit 6, an output of the echo processing unit 20, and a karaoke signal. A third mixing circuit 22 that mixes the output of the processing unit 21 and a D / A converter 23 that converts the output of the third mixing circuit 22 into an analog signal and outputs the analog signal to a sound system (not shown). are doing.
[0018]
In the karaoke scoring apparatus 1A, the first mixing circuit 5a mixes the left channel signal L and the right channel signal R of the model tone signal (Vr + A) reproduced from the CD 3, and outputs the mixed signal to the first wavelet transform unit 10a. Is.
[0019]
FIG. 2 is a block diagram showing a configuration of the voice cancel unit 6 of the karaoke scoring apparatus 1A.
[0020]
The voice cancel unit 6 inputs the left channel signal L and the right channel signal R from the input terminals 4a and 4b, removes the model vocal signal Vr from the model music signal (Vr + A), and outputs the musical signal A with accompaniment only. A high-pass filter (HPF) 60a, a low-pass filter (LPF) 61a and a multiplier 62a for inputting the left channel signal L reproduced from CD3, and a right channel signal R reproduced from CD3. Input high-pass filter (HPF) 60b, low-pass filter (LPF) 61b and multiplier 62b, adder 63 for adding the outputs of both multipliers 62a and 62b, and the output of HPF 60a for the left channel signal An adder 64a for the left channel signal that adds the output of the LPF 61a and the output of the adder 63, and a right channel signal And a the output and the output of LPF61b of HPF60b adder signal 64b for the right channel for adding an output of the adder 63.
[0021]
Each high-pass filter (HPF) 60a, 60b passes a high-frequency component of, for example, 7 kHz or more among the left channel signal L and the right channel signal R reproduced from CD3. Further, the low-pass filters (LPF) 61a and 61b pass low-frequency components of, for example, 100 Hz or less among the left channel signal L and the right channel signal R reproduced from the CD3. The frequency characteristics of these filters may be set to values that can cut the vocal signal (human voice) as effectively as possible.
[0022]
The multiplier 62a multiplies the left channel signal L reproduced from CD3 by a coefficient of +1, and the multiplier 62b multiplies the right channel signal R reproduced from CD3 by a coefficient of -1.
[0023]
In the voice cancel unit 6 having such a configuration, the left channel signal L multiplied by the coefficient +1 by the left channel signal multiplier 62a and the coefficient -1 by the right channel signal multiplier 62b. The right channel signal R is added by the adder 63 and supplied to the adders 64a and 64b, so that the sample vocal signal Vr at the center is canceled and only the accompaniment signal A in the voice band is compensated. Accompaniment signal A is output to the subsequent stage.
[0024]
The second mixing circuit 5b mixes the digital accompaniment signal A from the voice cancel unit 6 and the digital scoring target vocal signal Vm from the A / D converter 9, and outputs the result as a scoring target musical sound signal (Vm + A). 2 to the wavelet transform unit 10b.
[0025]
FIG. 3 is a block diagram showing the configuration of the first and second wavelet transform units 10a and 10b. Since the first and second wavelet transform units 10a and 10b are configured in the same manner, one of them will be described as a representative.
[0026]
“Wavelet transform” is a mathematical transformation of signal functions using basic functions (analyzing wavelets) localized in time and frequency as (1) expansion and contraction and (2) shifted functions as basis functions. In particular, when all the basis functions of the stretched and shifted function groups are orthogonal, this is called orthogonal wavelet transform. In order to realize this wavelet transform in hardware, the shift in (2) can be realized by using a filter, and the expansion and contraction in (1) can be realized by recursively using a filter of the same algorithm. It becomes possible.
[0027]
The wavelet transform unit 10 (10a, 10b) is for realizing the wavelet transform. For example, in order to perform orthogonal wavelet transform, the frequency band of the input signal is divided into a high frequency component and a low frequency component. Each subdivided subband filter 100a, 100b,... 100g, which outputs a high-frequency signal and a low-frequency signal, respectively, is further down-sampled by half. This is realized by recursively connecting this two-divided subband filter 100 to the side. This two-divided subband filter 100 includes a high-pass downsampling filter 101 and a low-pass downsampling filter 102 (I. Daubechies) that are completely orthogonal to each other. Note that these filters 101 and 102 may be linear filters (QMF) that satisfy a substantially orthogonal condition. FIG. 3 shows a configuration in which the frequency band of the input signal is divided into eight, and two-stage subband filters 100a to 100g are connected in seven stages. The high-pass downsampling filters 101a to 101g in the first stage to the seventh stage pass signals W. ₁ Thru W ₇ Is output to the scoring unit 11, and the pass signal E is output from the low-pass downsampling filter 102 in the seventh stage. ₇ Is output to the scoring unit 11. In this orthogonal wavelet transform, since the same configuration (algorithm) can be used recursively, the configuration (algorithm) is very simple.
[0028]
FIG. 4 is a diagram showing the frequency characteristics of the output signal of the wavelet transform unit 10.
[0029]
As shown in the figure, the signals W output from the high-pass downsampling filters 101a to 101g in the first to seventh stages. ₁ Thru W ₇ Are mainly frequency components of 2 kHz or more, 1 kHz to 2 kHz, 500 Hz to 1 kHz, 250 Hz to 500 Hz, 125 Hz to 250 Hz, 62 Hz to 125 Hz, 31 Hz to 62 Hz, respectively, from the 7th stage low-pass downsampling filter 102g. Output signal E ₇ Outputs mainly frequency components below 31 Hz. Accordingly, the frequency characteristics of the output signals of the first and second wavelet transform units 10a and 10b are band-divided with the characteristics that the bandwidth becomes wider as the frequency range is changed from the low frequency range by performing the orthogonal wavelet transform. .
[0030]
As will be described later, in this embodiment, scoring is performed by detecting the degree of coincidence between the same bands for each of the example side and the microphone side, but the bandwidth increases as the frequency band increases. Corresponds to the pitch-frequency characteristic. That is, comparing the same bands subjected to orthogonal wavelet transform is equivalent to comparing pitches rather than simply comparing frequencies, and is consistent with human auditory characteristics.
[0031]
FIG. 5 is a block diagram showing a configuration of the scoring unit 11.
[0032]
The scoring unit 11 obtains the degree of coincidence between the outputs of the first and second wavelet transform units 10a and 10b for each of the same frequency bands (bands), and sings the scored person based on the obtained degree of coincidence. The scoring value for. That is, the scoring unit 11 receives the signal W from the first wavelet transform unit 10a. ₁ Thru W ₇ , E ₇ The absolute value detectors (ABS) 110a to 110h for detecting the absolute values of the signal W and the signal W from the second wavelet transform unit 10b, respectively. ₁ Thru W ₇ , E ₇ Absolute value detectors (ABS) 111a to 111h for detecting the absolute values of the signals, smoothing units 112a to 112h for smoothing the outputs of the ABSs 110a to 110h, and smoothing units for smoothing the outputs of the ABSs 111a to 111h, respectively. 113a to 113h, coincidence detection units 114a to 114h for detecting the presence or absence of coincidence of signals in the same band between the first wavelet transform unit 10a side and the second wavelet transform unit 10b side, and each coincidence detection unit Counters (CNT) 115a to 115h that count up correspondingly when they match at 114, normalizers 116a to 116h that respectively normalize the output of the CNT 115, and random numbers that generate arbitrary random numbers from 0 to 1, for example When the output of the generation unit 200 and the normalization unit 116 is multiplied by the parameters a to h, In addition, multipliers 117a to 117h and 117i that multiply and output the output of the random number generator 200 by the parameter i, an adder 118 that adds the output value of each multiplier 117, and a counter (CNT that counts the number of comparisons K) ) 119. The scoring unit 11 accumulates the increase / decrease value conversion unit 201 that increases / decreases the addition value x output from the adder 118 and the increase / decrease value y of the addition value x increased / decreased by the increase / decrease value conversion unit 201. A section 202 is also provided, and the value accumulated by the accumulation section 202 is output to the display section 12 as a final scoring value.
[0033]
The normalizing unit 116 normalizes the output of the CNT 115 to 1 / K based on the comparison count K from the CNT 119.
[0034]
Each parameter a to h and i can be arbitrarily variably set on condition that a total relationship of 1 is satisfied by a parameter setting unit (not shown), thereby changing the scoring evaluation criteria.
[0035]
Next, operation | movement of 1 A of karaoke scoring apparatuses of the said structure is demonstrated.
[0036]
First, a processing operation until a signal such as a model musical tone signal (Vr + A) input in the karaoke scoring apparatus 1A reaches the first and second wavelet transform units 10a and 10b will be described.
[0037]
First, a digital model tone signal (Vr + A) is input to both input terminals 4a and 4b by CD3 reproduction. Also, the analog scoring target vocal signal Vm from the microphone 7 is input to the analog input terminal 8 by the actual singing of the scoring target person.
[0038]
The left channel signal L and the right channel signal R of the sample musical tone signal (Vr + A) input to both input terminals 4a and 4b are mixed by the first mixing circuit 5a and input to the first wavelet transform unit 10a. The The first wavelet transform unit 10a performs wavelet transform on the model musical tone signal (Vr + A) to generate a signal W ₁ Thru W ₇ , E ₇ Is output to the scoring unit 11.
[0039]
In addition, the sample vocal signal Vr is removed from the sample musical tone signal (Vr + A) inputted to both the input terminals 4a and 4b by the voice cancel unit 6, and the musical tone signal A having only accompaniment is inputted to the second mixing circuit 5b. . On the other hand, the scoring target vocal signal Vm input to the analog input terminal 8 is converted to digital by the A / D converter 9 and input to the second mixing circuit 5b. The second mixing circuit 5b mixes the accompaniment-only musical tone signal A from the voice canceling unit 6 with the scoring target vocal signal Vm from the A / D converter 9, and outputs the second scoring target musical signal (Vm + A). Input to the wavelet transform unit 10b. The second wavelet transform unit 10b performs wavelet transform on the scoring target musical sound signal (Vm + A) to generate a signal W ₁ Thru W ₇ , E ₇ Is output to the scoring unit 11.
[0040]
Next, processing operations after the scoring unit 11 will be described.
[0041]
Signal W from first wavelet transform unit 10a ₁ ~ W ₇ , E ₇ First, the absolute values of the signals are converted into absolute values in the absolute value detectors (ABS) 110a to 110h of the scoring unit 11, and then smoothed in the smoothers 112a to 112h. Each signal after smoothing is sent to the coincidence detection units 114a to 114h.
[0042]
Further, the signal W from the second wavelet transform unit 10b ₁ ~ W ₇ , E ₇ First, the absolute values of the signals are converted into absolute values by the absolute value detectors (ABS) 111a to 111h, and then smoothed at the smoothers 113a to 113h. Each signal after the smoothing is sent to the coincidence detection units 114a to 114h in the same manner as the signal from the first wavelet transform unit 10a.
[0043]
The coincidence detection units 114a to 114h detect whether or not the signals transmitted from the first wavelet transform unit 10a side and the second wavelet transform unit 10b side coincide in the same band. When the coincidence detectors 114a to 114h almost detect coincidence in the same band, the counters (CNT) 115a to 115h then count up the coincidence detected signals. The count value counted up by the CNT 115 is normalized to 1 / K by the normalizing units 116a to 116h based on the comparison count K counted by the counter 119. The normalized value has a value of 0 to 1, for example, and is output to the multipliers 117a to 117h and further multiplied by the parameters a to h. The multiplied value is output to the adder 118.
[0044]
On the other hand, an arbitrary random number from 0 to 1, for example, generated by the random number generation unit 200 is multiplied by the parameter i in the multiplier 117i, and output to the adder 118 in the same manner as the outputs from the multipliers 117a to 117h. The random number output from the random number generation unit 200 is for adding a human scoring element to the final scoring value, so that the mechanical scoring in the conventional karaoke scoring device is performed by humans. Can be approximated to a typical scoring. The parameters a to h and i can be arbitrarily variably set on condition that the relationship (a + b + c... + I = 1) that is 1 in total is satisfied by a parameter setting unit (not shown). Can be changed.
[0045]
The multiplication values from the multipliers 117a to 117h and the multiplier 117i output to the adder 118 are added by the adder 118, respectively, and sent to the increase / decrease value conversion unit 201 as an addition value x (for example, a value of 0 to 1). . The increase / decrease value conversion unit 201 has a plurality of conversion tables or calculation means, and thereby determines an increase / decrease value y corresponding to the addition value x and outputs it to the accumulation unit 202. In addition, the conversion process from the addition value x to the increase / decrease value y in the increase / decrease value conversion unit 201 can change the increase / decrease rate by providing a plurality of conversion tables or calculation means, and the scoring evaluation criteria can also be changed accordingly It becomes.
[0046]
The increase / decrease value y output from the increase / decrease value conversion unit 201 is accumulated to an initial value (for example, 50 points) set in advance by the accumulation unit 202, and the value of “initial value + accumulated value of increase / decrease value y”. Is output to the display unit 12 as a final scoring result. Note that the scoring value output to the display unit 12 may be not only the final scoring result but also the scoring value that changes over time. Thereby, the scoring target person can know the score in the middle of singing.
[0047]
According to the karaoke scoring device 1A according to the first embodiment described above, the orthogonal wavelet transform output of the model tone signal (Vr + A) based on the reproduction of the CD3 and the scoring target tone signal (Vm + A) based on the input from the microphone 7 are used. Since the degree of coincidence of the orthogonal wavelet transform output is obtained for each of the same frequency band, an algorithm for pitch extraction becomes unnecessary and the configuration is simplified. Moreover, since orthogonal wavelet transform is performed, scoring corresponding to human hearing can be performed.
[0048]
(Second Embodiment)
FIG. 6 is a block diagram showing a karaoke scoring apparatus according to the second embodiment of the present invention.
[0049]
As in the first embodiment, the karaoke scoring device 1B includes a left channel / digital input terminal 4a, a right channel / digital input terminal 4b, first and second mixing circuits 5a and 5b, a voice cancel unit 6, an analog Although the input terminal 8, the A / D converter 9, the first and second wavelet transform units 10a and 10b, the scoring unit 11, and the display unit 12 are provided, the connection relationship is different from that of the first embodiment.
[0050]
That is, the karaoke scoring apparatus 1B includes an adder 13 that adds an output signal of the first mixing circuit 5a and a signal obtained by subtracting an output signal of the second mixing circuit 5b, and outputs the output of the adder 13 The input of the first wavelet transform unit 10a is used, and the output of the A / D converter 9 is used as the input of the second wavelet transform unit 10b.
[0051]
Next, operation | movement of the karaoke scoring apparatus 1B of the said structure is demonstrated. In the following description of the operation, the description of the same part as the operation in the first embodiment is omitted.
[0052]
First, as in the first embodiment, a digital model tone signal (Vr + A) is input to both digital input terminals 4a and 4b, and an analog scoring target vocal signal Vm is input to the analog input terminal 8.
[0053]
The left channel signal L and the right channel signal R of the sample musical tone signal (Vr + A) input to both the input terminals 4a and 4b are mixed by the first mixing circuit 5a. At the same time, the sample vocal signal Vr is removed from the sample musical tone signal (Vr + A) inputted to both the input terminals 4a and 4b by the voice cancel unit 6, and the musical tone signal A with accompaniment only is output to the second mixing circuit 5b. Is done. The second mixing circuit 5 b mixes the left and right signals of the musical tone signal A with accompaniment from the voice cancel unit 6 and outputs the mixed signal to the adder 13. The adder 13 adds the sample musical tone signal (Vr + A) from the first mixing circuit 5a and the negative tone signal A from the second mixing circuit 5b, and adds the sample vocal signal Vr to the first vocal signal Vr. 1 to the wavelet transform unit 10a. The first wavelet transform unit 10a performs wavelet transform on the model vocal signal Vr to generate a signal W ₁ Thru W ₇ , E ₇ Is output to the scoring unit 11.
[0054]
On the other hand, the scoring target vocal signal Vm input to the analog input terminal 8 is converted to digital by the A / D converter 9 and input to the second wavelet converter 10b. The second wavelet transform unit 10b performs wavelet transform on the scoring target vocal signal Vm to generate a signal W ₁ Thru W ₇ , E ₇ Is output to the scoring unit 11.
[0055]
The scoring unit 11 receives signals W from the first and second wavelet transform units 10a and 10b. ₁ Thru W ₇ , E ₇ The degree of coincidence between each other is obtained, and a scoring value is obtained based on each degree of coincidence and displayed on the display unit 12.
[0056]
According to the karaoke scoring device 1B according to the second embodiment described above, the orthogonal wavelet transform output of the model vocal signal Vr based on the reproduction of the CD3 and the orthogonal wavelet transform output of the scoring target vocal signal Vm based on the input from the microphone 7 are performed. Since the degree of coincidence is obtained for the same frequency band, the same effect as in the first embodiment can be obtained.
[0057]
(Third embodiment)
FIG. 7 is a block diagram showing a karaoke scoring device according to the third embodiment of the present invention.
[0058]
This karaoke scoring device 1C is configured to input analog signals reproduced from a laser disk (LD) 14 to a left channel / analog input terminal 15a and a right channel / analog input terminal 15b, and an analog signal from the left channel / analog input terminal 15a. From an A / D converter 16a that converts the analog signal into a digital signal, an analog input terminal 8 that inputs an analog scoring target vocal signal Vm from the microphone 7, an analog signal from the analog input terminal 8, and a right channel / analog input terminal 15b An adder 17 for adding the analog signals of A, a A / D converter 16b for converting the analog signals of the adders 17 into digital signals, and a first wavelet transform unit 10a connected to the A / D converter 16b. And a second wavelet transform connected to the A / D converter 16b. And part 10b, and is configured by including the same scoring unit 11 and the display unit 12 in the first embodiment.
[0059]
In the analog audio channel of the laser disc (LD) 14, a model tone signal (Vr + A) based on the singing and accompaniment of a model person is recorded on the left channel, and a tone signal A of accompaniment only is recorded on the other channel. Is.
[0060]
Next, the operation of the karaoke scoring device 1C configured as described above will be described.
[0061]
First, by playing the LD 14, an analog model tone signal (Vr + A) is input to the left channel / analog input terminal 15a, and an analog tone signal A only for accompaniment is input to the left channel / analog input terminal 15a. Also, an analog scoring target vocal signal Vm is input from the microphone 7 to the analog input terminal 8 by the actual singing of the scoring target person.
[0062]
The analog model tone signal (Vr + A) input to the left channel / analog input terminal 15a is converted to digital by the A / D converter 16a and input to the first wavelet converter 10a. The first wavelet transform unit 10a performs wavelet transform on the model musical tone signal (Vr + A) to generate a signal W ₁ ~ W ₇ , E ₇ Is output to the scoring unit 11.
[0063]
On the other hand, the accompaniment-only musical tone signal A input to the right channel / analog input terminal 15b and the scoring target vocal signal Vm input to the analog input terminal 8 are added by the adder 17, and the scoring target musical signal (Vm + A) is added. ) Is input to the A / D converter 16b. The A / D converter 16b converts the scoring target musical sound signal (Vm + A) from analog to digital and inputs it to the second wavelet transform unit 10b. The second wavelet transform unit 10b performs wavelet transform on the scoring target musical sound signal (Vm + A) to generate a signal W ₁ ~ W ₇ , E ₇ Is output to the scoring unit 11. The scoring unit 11 receives signals W from the first and second wavelet transform units 10a and 10b. ₁ Thru W ₇ , E ₇ The degree of coincidence between each other is obtained, and a scoring value is obtained based on each degree of coincidence and displayed on the display unit 12.
[0064]
According to the karaoke scoring device 1C according to the third embodiment described above, the orthogonal wavelet transform result of the model tone signal (Vr + A) based on the reproduction of the LD 14 and the scoring target tone signal (Vm + A) based on the input from the microphone 7 are obtained. Since the degree of coincidence of the orthogonal wavelet transform results is obtained for each identical frequency band, the same effect as in the first embodiment can be obtained.
[0065]
(Fourth embodiment)
FIG. 8 is a block diagram showing a karaoke scoring device according to the fourth embodiment of the present invention.
[0066]
As in the third embodiment, the karaoke scoring device 1D includes a left channel / analog input terminal 15a, a right channel / analog input terminal 15b, A / D converters 16a and 16b, an analog input terminal 8, the first and first channels. Although two wavelet transform units 10a and 10b, a scoring unit 11 and a display unit 12 are provided, the connection relationship is different from that of the third embodiment.
[0067]
In other words, the karaoke scoring device 1D adds the sample musical tone signal (Vr + A) from the left channel / analog input terminal 15a and the negative tone signal A from the right channel / analog input terminal 15b. Is connected to the adder 18 via the A / D converter 16a, and the second wavelet converter is connected to the analog input terminal 8 via the A / D converter 16b. 10b is connected.
[0068]
Next, the operation of the karaoke scoring device 1D configured as described above will be described.
[0069]
As in the third embodiment, an analog model tone signal (Vr + A) is input to the left channel / analog input terminal 15a, and a tone signal A having only an analog accompaniment is input to the right channel / analog input terminal 15b. An analog scoring target vocal signal Vm is input to the analog input terminal 8.
[0070]
The adder 18 adds the model tone signal (Vr + A) input to the left channel / analog input terminal 15a and a signal obtained by subtracting the accompaniment tone signal A input to the right channel / analog input terminal 15b, The sample vocal signal Vr is input to the A / D converter 16a. The A / D converter 16a converts the model vocal signal Vr from analog to digital and inputs it to the first wavelet transform unit 10a. The first wavelet transform unit 10a performs wavelet transform on the model vocal signal Vr to generate a signal W ₁ Thru W ₇ , E ₇ Is output to the scoring unit 11.
[0071]
On the other hand, the scoring target vocal signal Vm input to the analog input terminal 8 is converted to digital by the A / D converter 16b and input to the second wavelet converter 10b. The second wavelet transform unit 10b performs wavelet transform on the scoring target vocal signal Vm to generate a signal W ₁ Thru W ₇ , E ₇ Is output to the scoring unit 11.
[0072]
The scoring unit 11 receives signals W from the first and second wavelet transform units 10a and 10b. ₁ Thru W ₇ , E ₇ The degree of coincidence between each other is obtained, and a scoring value is obtained based on each degree of coincidence and displayed on the display unit 12.
[0073]
According to the karaoke scoring apparatus 1D according to the fourth embodiment described above, the orthogonal wavelet transform result of the sample vocal signal Vr based on the reproduction of the LD 14 and the orthogonal wavelet transform result of the scoring target vocal signal Vm based on the input from the microphone 7. Since the degree of coincidence is obtained for the same frequency band, the same effect as in the first embodiment can be obtained.
[0074]
In addition, this invention is not limited to the said 1st-4th embodiment, A various deformation | transformation implementation is possible.
[0075]
For example, the number of band divisions accompanying the wavelet transform is not limited to eight bands, and can be increased or decreased.
[0076]
【The invention's effect】
According to the present invention described in detail above, the degree of coincidence between the wavelet transform output of the model vocal signal or model tone signal and the wavelet transform output of the score target vocal signal or score target music signal is obtained for each same frequency band. Therefore, an algorithm for pitch extraction is not required, the configuration is simplified, and scoring that matches human hearing characteristics, that is, matches human evaluation criteria can be performed.
[Brief description of the drawings]
FIG. 1 is a block diagram of a karaoke system showing an application example of a karaoke scoring device according to an embodiment of the present invention.
FIG. 2 is a block diagram showing a configuration of a voice cancel unit according to the present invention.
FIG. 3 is a block diagram showing a configuration of a wavelet transform unit according to the present invention.
FIG. 4 is a diagram illustrating frequency characteristics of an output signal of a wavelet transform unit according to the present invention.
FIG. 5 is a block diagram showing a configuration of a scoring unit according to the present invention.
FIG. 6 is a block diagram showing a karaoke scoring device according to a second embodiment of the present invention.
FIG. 7 is a block diagram showing a karaoke scoring device according to a third embodiment of the present invention.
FIG. 8 is a block diagram showing a karaoke scoring device according to a fourth embodiment of the present invention.
[Explanation of symbols]
1A, 1B, 1C, 1D Karaoke scoring device
3 Compact disc
5a First mixing circuit
5b Second mixing circuit
6 Voice cancellation part
7 Microphone
10a First wavelet transform unit
10b Second wavelet transform unit
11 scoring department
12 Display section
13, 17, 18 Adder
14 Laser disk
A Accompanied musical tone signal
Vr Model vocal signal
Vr + A Model music signal
Vm Scoring target vocal signal
Vm + A Music signal for scoring

Claims (7)

A first wavelet transforming means for wavelet transforming a model tone signal based on the singing and accompaniment of a model person;
A second wavelet transforming means for wavelet transforming a musical signal to be scored based on the same accompaniment as the singing of the scored person and the accompaniment;
The degree of coincidence between the outputs of the first and second wavelet transforming means is repeatedly obtained over time by comparing the same frequency bands ,
Each matching score obtained is normalized by dividing it by the number of comparisons so that it falls within a predetermined range.
Find an increase / decrease value to increase / decrease the scoring value based on each normalized degree of matching and a random number that takes a value within the predetermined range ,
A scoring means for accumulating these increase / decrease values to a preset initial value each time the degree of coincidence is obtained, and outputting the accumulated value as a scoring value ;
A karaoke scoring device characterized by comprising: A first wavelet transform means for wavelet transforming a model vocal signal based on the singing of a model person;
A second wavelet transforming means for wavelet transforming a vocal signal for scoring based on the singing of the scoring person;
The degree of coincidence between the outputs of the first and second wavelet transforming means is repeatedly obtained over time by comparing the same frequency bands ,
Each matching score obtained is normalized by dividing it by the number of comparisons so that it falls within a predetermined range.
Find an increase / decrease value to increase / decrease the scoring value based on each normalized degree of matching and a random number that takes a value within the predetermined range ,
A scoring means for accumulating these increase / decrease values to a preset initial value each time the degree of coincidence is obtained, and outputting the accumulated value as a scoring value ;
A karaoke scoring device characterized by comprising: A first wavelet transforming means for wavelet transforming a model tone signal based on a singing and accompaniment of a model person reproduced from a recording medium;
Removing means for removing a model vocal signal based on the singing of the model person from the model music signal reproduced from the recording medium and outputting a musical signal only for accompaniment;
Mixing means for mixing the musical sound signal only of the accompaniment output from the removing means and the scoring target vocal signal based on the singing of the scoring target person inputted from the microphone, and outputting the scoring target musical sound signal;
Second wavelet transforming means for wavelet transforming the scoring target tone signal output from the mixing means;
The degree of coincidence between the outputs of the first and second wavelet transforming means is repeatedly obtained over time by comparing the same frequency bands ,
Each matching score obtained is normalized by dividing it by the number of comparisons so that it falls within a predetermined range.
Find an increase / decrease value to increase / decrease the scoring value based on each normalized degree of matching and a random number that takes a value within the predetermined range ,
A scoring means for accumulating these increase / decrease values to a preset initial value each time the degree of coincidence is obtained, and outputting the accumulated value as a scoring value ;
A karaoke scoring device characterized by comprising: A sample musical sound signal based on the singing and accompaniment of the model person reproduced from the recording medium is input, and the musical vocal signal based on the singing of the model person is removed from the input signal to output a musical signal only of the accompaniment Removing means to perform,
Arithmetic means for subtracting the musical tone signal of accompaniment output from the removing means from the musical tone signal reproduced from the recording medium and outputting a representative vocal signal;
First wavelet transforming means for wavelet transforming the model vocal signal output from the computing means;
A second wavelet transforming means for wavelet transforming a vocal signal to be scored based on a singing voice of the person to be graded input from a microphone;
The degree of coincidence between the outputs of the first and second wavelet transforming means is repeatedly obtained over time by comparing the same frequency bands ,
Each matching score obtained is normalized by dividing it by the number of comparisons so that it falls within a predetermined range.
Find an increase / decrease value to increase / decrease the scoring value based on each normalized degree of matching and a random number that takes a value within the predetermined range ,
A scoring means for accumulating these increase / decrease values to a preset initial value each time the degree of coincidence is obtained, and outputting the accumulated value as a scoring value ;
A karaoke scoring device characterized by comprising: A model tone signal reproduced from the one channel of the recording medium in which a model tone signal based on the singing and accompaniment of the model person is recorded in one channel and the tone signal of only the accompaniment is recorded in the other channel is recorded. First wavelet transform means for performing wavelet transform;
An arithmetic means for adding a musical tone signal of accompaniment reproduced from the other channel of the recording medium and a scoring target vocal signal based on the singing of the scoring subject inputted from a microphone and outputting a scoring target musical sound signal; ,
Second wavelet transforming means for wavelet transforming the scoring target tone signal output from the computing means;
The degree of coincidence between the outputs of the first and second wavelet transforming means is repeatedly obtained over time by comparing the same frequency bands ,
Each matching score obtained is normalized by dividing it by the number of comparisons so that it falls within a predetermined range.
Find an increase / decrease value to increase / decrease the scoring value based on each normalized degree of matching and a random number that takes a value within the predetermined range ,
A scoring means for accumulating these increase / decrease values to a preset initial value each time the degree of coincidence is obtained, and outputting the accumulated value as a scoring value ;
A karaoke scoring device characterized by comprising: A model tone signal based on the singing and accompaniment of a model person is recorded on one channel, and a tone signal of the accompaniment only is reproduced from the one channel of the recording medium recorded on the other channel. Calculating means for subtracting the musical tone signal of the accompaniment reproduced from the other channel and outputting a model vocal signal based on the model person's singing;
First wavelet transforming means for wavelet transforming the model vocal signal output from the computing means;
A second wavelet transforming means for wavelet transforming a vocal signal to be scored based on a singing voice of the person to be graded input from a microphone;
The degree of coincidence between the outputs of the first and second wavelet transforming means is repeatedly obtained over time by comparing the same frequency bands ,
Each matching score obtained is normalized by dividing it by the number of comparisons so that it falls within a predetermined range.
Find an increase / decrease value to increase / decrease the scoring value based on each normalized degree of matching and a random number that takes a value within the predetermined range ,
A scoring means for accumulating these increase / decrease values to a preset initial value each time the degree of coincidence is obtained, and outputting the accumulated value as a scoring value ;
A karaoke scoring device characterized by comprising: 7. The scoring unit can set the degree of coincidence and the ratio of the random numbers to arbitrary values when obtaining the increase / decrease value. Karaoke scoring device.

Images