JP3509139B2 - Waveform generator - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a waveform generator for forming a musical tone waveform based on data collected from an actual musical tone signal.

[0002]

2. Description of the Related Art In a sample sound source used in current electronic musical instruments and the like, an actual musical tone signal is digitized and stored in a waveform memory, and if necessary, the digitized musical tone signal is stored in the waveform memory. By reading out, a musical tone signal is formed.

By the way, the fundamental elements that determine the waveform of a musical tone are the fundamental frequency and harmonics that are integral multiples thereof, and these are combined at various levels to form a musical tone waveform. Therefore, when the fundamental frequency of a musical sound changes, the frequency of the harmonics changes accordingly, but in the case of the same resonator (musical instrument), the level of each harmonic is determined by that frequency and is not significantly affected by the order of the harmonics. Not, that is,
Frequency spectrum of one instrument (spectrum envelope)
Is known to be constant regardless of the frequency of the musical sound emitted by the musical instrument.

[0004]

However, in the case of forming a musical tone having a frequency different from that of the sampled musical tone signal with the above-mentioned sample sound source, this is dealt with by changing the frequency of the clock for accessing the waveform memory to change the reading speed. Was. That is, when forming a high-frequency tone signal, a fast clock is read, and when forming a low-frequency tone signal, a slow clock is read. In this way, the frequency of the fundamental wave and its harmonics can be changed by changing the frequency of the read clock, but in this method, the frequency shifts while maintaining the levels of these fundamental waves and harmonics. Therefore, the spectrum envelope, which should be constant regardless of the fundamental wave, is also shifted, and there is a drawback that the timbre changes. For example, when the musical tone data stored in the spectrum envelope shown in FIG. 4A is read at a fast clock (high frequency), the spectrum envelope shown in FIG. It changes and the timbre also changes.

In order to solve this drawback, in the conventional sampled sound source, a plurality of waveforms of different frequencies for the same tone color are sampled and stored in memory, and the sampling waveform of the frequency closest to it is read according to the frequency of the musical tone to be synthesized. By doing so, the amount of shift was reduced to minimize changes in timbre.

However, even with this method, there is a drawback in that a change in tone color cannot be eliminated at all and a large waveform memory is required because a plurality of waveforms must be stored for one tone color. It was

It is an object of the present invention to apply a waveform generator in which the spectrum envelope does not change even if the frequency of the constituent waveform changes.

[0008]

SUMMARY OF THE INVENTION The present invention is directed to synthesizing data for generating a spectrum envelope and a non-harmonic component.
Data generating means for generating the non-harmonic component data, basic frequency specifying means for specifying a basic frequency, and the basic frequency
And determines the frequency of the harmonic component based on the wave number, and decision means for determining the level of harmonics based on the spectrum envelope, Oyo frequency determined by said decision means
And non-harmonic composition
Waveform synthesizing means for synthesizing non-harmonic components based on minute data . In addition, this invention is
Synthesizes data to generate a spectrum envelope and non-harmonic components
Data storage for storing non-harmonic component data for
A step, a basic frequency specifying means for specifying a basic frequency, and
When the frequency of the overtone component is determined based on the fundamental frequency,
Based on the spectrum envelope, the level of the overtone component is
The determining means for determining the
In addition to synthesizing harmonic components by wave number and level,
Waveform that synthesizes non-harmonic components based on non-harmonic component data
And a synthesizing means.

[0009]

[Function] When the fundamental frequency is specified, the frequency of the harmonic component of this fundamental frequency and the
Its level is determined (decision means). The harmonic component and non-harmonic component data whose levels have been determined in this way
The non-overtone component based on the above is synthesized by the waveform synthesizing means to generate a musical tone waveform. The frequency of the overtone component varies depending on the fundamental frequency designated by the fundamental frequency designating means, but the spectrum envelope is the data that generates the spectrum.
It is uniquely determined by the frequency and does not change with frequency.

[0010]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a block diagram showing the configuration of a musical tone generating system which is an embodiment of the present invention. This tone generation system is a real-time type system in which the frequency spectrum extracted by the extraction unit A is synthesized by the synthesis unit C based on the designation of the performance unit B.

The extraction section A is a microphone 1 for inputting a musical tone signal to be analyzed, an A / D conversion section 2 for digitizing the musical tone signal input from the microphone 1, and an FFT analysis of the digitally converted musical tone waveform data. And an overtone / non-overtone separation unit 4 for separating an overtone component and a non-overtone component of a musical tone from the data decomposed into each frequency component by the FFT analysis. Here, the overtone is
It is a fundamental wave of a musical sound and a harmonic of its integer value. The overtone changes in proportion to the frequency of the musical sound. On the other hand, the non-overtone is a vibration peculiar to the musical instrument and has a constant frequency regardless of the frequency of the synthesized musical tone. The harmonic component data is composed of frequency data F _i (i = 0, 1, ...) And level data L _{i of} each harmonic component, and is input to the harmonic component data generator 9 of the synthesizer C. The data of the non-harmonic component is the frequency data NF _j (j = 0, 1,
..) and level data NL _j , and is input to the non-harmonic synthesizer 13 of the synthesizer C.

The performance section B comprises a performance operator 8 such as a keyboard and a pitch designation section 9. Pitch designation section 9
Outputs pitch data based on the operation of the performance operator 8.

The pitch data is data instructing the frequency of the tone signal to be synthesized. The pitch data is input to the harmonic overtone data generator 10.

The synthesizer C includes the harmonic overtone data generator 10,
Musical sound synthesizer 11, mixer 14, and D / A converter 1
It consists of 5. Further, the musical sound synthesizer 11 is composed of a harmonic overtone synthesizer 12 and a non-harmonic overtone synthesizer 13. The mixer 14 is a circuit for adding and synthesizing the waveform data synthesized by the harmonic overtone synthesizer 12 and the non-harmonic overtone synthesizer 13. The added and synthesized data is input to the D / A converter 15 and converted into an analog tone signal.

An analog tone signal is a sound system 1
6 is input.

The harmonic overtone data generator 10 assumes a pseudo spectrum envelope based on the harmonic overtone component data F _i , L _i input from the harmonic overtone / non-harmonic overtone separator 4, and matches the spectrum envelope. As described above, the levels of the fundamental wave and the harmonic wave of the frequency designated by the pitch data are determined and input to the harmonic overtone synthesizer 12 (see FIGS. 4A and 4B). The harmonic overtone synthesizer 12 synthesizes the fundamental wave and its harmonics at a designated level.

The non-harmonic synthesizer 13 also receives the non-harmonic data NF _j and NL _j input from the harmonic / non-harmonic separating section 4.
Non-overtone is synthesized based on.

Further, the recording circuit 6 is provided in the extraction unit A.
Are connected. This is because with the above-described system alone, when synthesizing musical tones, natural musical tones must always be input from the microphone 1, so that there is no freedom in playing. To solve this, the musical tone data input from the microphone 1 and converted into digital data by the A / D conversion unit 2 is stored in the hard disk 7, and when the musical tone is synthesized, this data is read by the FFT analysis unit 3. It is intended to cause the above operation to be carried out.

Therefore, the recording circuit 6 is connected to the A / D
A hard disk 7 and a control operator 5 are connected to the recording circuit 6, which is connected to the conversion unit 2 and the FFT analysis unit 3. The recording circuit 6 stores in the hard disk 7 the musical sound data input from the microphone 1 and converted into digital data by the A / D converter 2.
When synthesizing a musical tone, this musical tone data is read from the hard disk 7 and input to the FFT analysis unit 3. The control operator 5 is for designating a storage area of the hard disk 7, a reading section, and the like.

FIG. 2 is a detailed block diagram of the overtone / non-overtone separating section 4. Further, FIG. 3 is a diagram showing peaks of frequency components of a musical sound. The frequency component data (FFT output) input from the FFT analysis unit 3 is input to the peak extraction unit 20. The peak extraction unit 20 extracts a peak from this curve, and inputs the peak frequency and peak level to the threshold processing unit 21 and the separation unit 24. The threshold processing unit 21
Each peak level is compared with a threshold Th, and only those exceeding the threshold Th are output to the harmonic detection unit 22. In FIG. 3, the peaks of the frequency components marked with ⊚, ◯ or × exceed the threshold Th. Harmonic detector 2
In 2, the frequency difference of all the input peak frequencies is calculated. The calculation result of the frequency difference is output to the majority decision processing unit 23 together with the peak frequency data. Majority processing unit 2
In step 3, the largest number of the input difference data is determined to be the fundamental frequency of the analyzed tone.
In other words, the tone signal is mainly composed of the fundamental wave of the fundamental frequency and the harmonics that are integral multiples thereof, so if the frequencies of the fundamental wave and the harmonics are sequentially subtracted to obtain the difference, all the fundamental wave frequencies are It is calculated. Other values are also calculated by subtraction with other combinations including non-harmonic components, but the frequency of the fundamental wave is calculated because there are the most sequential combinations (F _i and F _{i +1} ) of the fundamental wave and harmonics. It is most often done. Therefore, the fundamental frequency can be determined by detecting the largest frequency difference. In FIG. 3, those with ⊚ are fundamental waves, and those with ∘ are harmonics. Since the frequency of this fundamental wave is 100 Hz, the frequency difference with each harmonic is also 100 Hz. Also,
Those marked with x are non-harmonic components. Majority processing unit 2
The fundamental frequency determined in 3 is input to the separation unit 24.
The separation unit 24 classifies the peak data (peak frequency, peak level) input from the peak extraction unit 20 into overtone components and non-overtone components, and overtone component data (F _i , L _i ) and non-overtone component data (NF), respectively. _j , NL _j )
Output as. In the above description, all the peaks extracted by the peak extraction unit 20 are output as overtone components or non-overtone components, but in order to simplify the process, the threshold processing unit 21 sets the threshold values. Only the peaks that are determined to exceed may be input to the separation unit 24 and output as overtone components and non-overtone components.

Here, the harmonic overtone data generator 10 is specified by the pitch data assuming a spectrum envelope formed by the overtone component data (F _i , L _i : i = 0, 1, ...). A method of generating the fundamental wave and the harmonic overtone data (TF _k , TL _k : k = 0, 1, ...) Of the frequency SF will be described. The frequency data TF0 of the fundamental wave is S
F, and the frequency TF _k of the _kth harmonic is SF × k. On the other hand, the level data TL _k (k = 0, 1, ...) Can be obtained as follows.

First, _x for which F _x ≤TF _k ≤F _{x + 1} is detected, and y = (TF _k -F _x ) / (F
_x + 1 -F _x) is calculated. This y is a number of 0 ≦ y ≦ 1, and is an internally divided value of the section of x and x + 1. Based on this value, TL _k = (1−y) × L _x + y × L _{x + 1} .

These data are input to the harmonic overtone synthesizer 12.

By such calculation, it is possible to shift only the frequency of the overtone while maintaining the spectrum envelope of the tone signal input from the microphone 1. This state is shown in FIG. FIG. 4A is a diagram showing the spectrum envelope of the overtone component extracted by the FFT analysis. Of these, the peak at the left end is the fundamental wave, and the first harmonic and the second harmonic are arranged on the right below. FIG. 3B shows overtone data in the case of synthesizing a frequency higher than the input tone signal. The frequencies of the fundamental wave and the harmonics are shifted to higher frequencies, and the intervals between them are widened, but the levels thereof are determined based on the spectrum envelope of FIG. As a result, since the frequency spectrum is maintained even if the frequency of the synthesized tone changes, it is possible to synthesize a tone that has no change in tone color.

Incidentally, FIG. 4C shows a frequency spectrum when the read clock is raised and the synthesized tone frequency is raised in the conventional waveform memory system. In this method, the frequencies of the fundamental wave and the harmonics are increased, and at the same time, the frequency spectrum is also shifted.

FIGS. 5 and 6 are block diagrams of a tone generating system according to another embodiment of the present invention. This tone generation system is a system in which a recording device for extracting and recording overtone components and non-overtone components from an actual tone and a performance device for synthesizing the tone based on the data recorded by the performance are divided. FIG. 5 shows a recording device, and FIG. 6 shows a performance device.

In FIG. 5, the microphone 31, the A / D converter 32, the FFT analyzer 33, and the overtone / non-overtone separator 34, which input a natural musical tone, operate in the same manner as the extractor A shown in FIG. To do. The overtone components extracted by the overtone / non-overtone separation unit 34 are input to the F _i → ΔF _i conversion unit 38. The non-harmonic component is input to the non-harmonic memory 41. The non-harmonic component data is composed of frequency data NF _j and level data NL _j , and a plurality of sets of these are stored in the non-harmonic memory 41.
In the F _i → ΔF _i conversion unit 38, _i of the fundamental frequency F ₀ of the frequency data F _i of the input harmonic component data is i.
The shift from the double is calculated, and this is used as the frequency shift data ΔF.
_It is stored in the overtone memory 40 together with the corresponding level data L _{i as i} .

The recording operation of the storage unit 39 consisting of the overtone memory 40 and the non-overtone memory 41 is performed by the recording control operator 35,
It is controlled by the recording controller 36 and the writing counter 37. When extracting and recording the overtones and non-overtones of a musical tone, the recording control operator 35 is operated to set the recording address, the recording time, etc., and the recording operation is started. After that, a musical sound is input from the microphone 1. The recording controller 36 outputs data such as a recording start address and a recording frequency to the writing counter 37. The writing counter 37 outputs an address to the storage unit 39 based on these data at regular intervals. The data output from the F _i → ΔF _i conversion unit 38 and the overtone / non-overtone separation unit 34 is stored at this address.

An edit processing section 42 is connected to the storage section 39. An edit instruction operator 43 is connected to the edit processing unit 42. In the edit mode, the user can operate the edit instruction operator 43 to perform processing such as extraction of a recorded portion used for the performance by transcribing it on the performance device and tone color adjustment such as deletion of higher harmonics. .

FIG. 6 is a block diagram of the playing device. The performance operator 50 is a performance device such as a keyboard. When the player turns on any of the keys, the operation detection unit 54 detects that the key is turned on and which key is turned on. The operation detection unit 54 responds to this key ON by the counter 55,
A reset signal is output to 56 and pitch data corresponding to the turned-on key is output to the F number generation unit 57. The F number generating unit 57 outputs the F number corresponding to this pitch data. The F number is numerical data proportional to the frequency, and is used as frequency data in this playing device (electronic circuit). The F number output by the F number generating unit 57 is the musical tone pitch generating unit 6
2 and the tone level generator 63.

Further, a touch detector 51 is connected to the performance operator 50. The touch detection unit 51 detects the strength of key-on (velocity data). Touch detector 5
The reading rate generating units 52, 5
Output to 3. The reading rate generator 52 uses the overtone memory 70.
The read rate generator 53 is a circuit for generating the read rate of the non-overtone memory 71. The read rates generated by these read rate generators 52 and 53 are input to counters 55 and 56.

The counters 55 and 56 receive the note-on pulse (reset signal) from the operation detector 54,
The count addresses are reset to the top addresses of the overtone memory 70 and the non-overtone memory 71, respectively, and the counting operation is performed based on the read rates input from the read rates 52 and 53. This count value is used as an access address in the overtone memory 7
0, input to the non-harmonic memory 71. Counters 55,5
6 stops after counting the final address.
The reading rate generators 52 and 53 are normally the writing counter 3
Although the read rate of the same rate as the count clock of 7 (see FIG. 5) is output, this rate is increased or decreased based on the velocity data input from the touch detection unit 51. That is, the read rate is increased to increase the read speed of the overtone memory 70 and the non-overtone memory 71. As a result, the temporal change of the spectrum envelope becomes fast and the timbre change becomes large. This is the attack part when you hit the piano hard,
This is a simulation of how the tone color of the decay part changes greatly and quickly.

The overtone data ΔF _i , L _i read from the overtone memory 70 are input to the scaling unit 60. In addition, the non-harmonic data N read from the non-harmonic memory 71
F _j and NL _j are input to the scaling unit 61. The scaling units 60 and 61 control the timbre and pitch based on the velocity data detected by the touch detection unit 51.
For example, when the touch strength data is large, the level of the higher harmonics is increased, and the pitch of each harmonic is slightly raised. Modified overtone pitch data ΔF _i
Is input to the harmonic overtone pitch generator 62, and the harmonic overtone level data L _i is input to the harmonic overtone level generator 63. The overtone pitch generating section 62 generates frequency data of the fundamental wave and each harmonic based on the F number input from the F number generating section 57 and ΔF _i (i = 0, 1, ...) And harmonic synthesis. Input to the sound source 64. On the other hand, the harmonic overtone level generation unit 63 assumes a spectrum envelope based on the F number and overtone level data L _i , and generates the levels of the fundamental wave and each harmonic so as to match the spectrum envelope. This level data is input to the harmonic synthesis sound source 64.

The harmonic synthesis sound source 64 generates a plurality of harmonic waves based on the input pitch data and level data and inputs them to the mixer 66.

On the other hand, the non-harmonic frequency data NF _j and non-harmonic level data N read from the non-harmonic memory 71.
L _j is input to the scaling unit 61. The scaling unit 61 scales the non-overtone data according to the velocity data. The scaling method is the same as that of the scaling unit 60, but the degree of scaling and the like are performed independently of the scaling unit 60. The scaled data is input to the harmonic synthesis sound source 65. The harmonic synthesis sound source 65 generates a non-harmonic sound wave based on each non-harmonic sound data, and inputs it to the mixer 66.

The mixer 66 adds and combines the input basic waveforms to form musical tone waveform data. This tone waveform data is input to the D / A converter 67. D / A converter 67
Converts the input musical tone waveform data into an analog musical tone signal and inputs it to the sound system 68. The sound system 68 amplifies this musical tone signal and outputs it as a musical tone.

In the performance device having the above-described structure, when the performance operator 50 is operated, the note-on pulse and the read rate are input to the counters 55 and 56. Counter 55,
The 56 starts counting by this input, and inputs the address to the overtone memory 70 and the non-overtone memory 71. The overtone memory 70 thereby outputs the overtone data ΔF _i , L _i to the scaling unit 60. In addition, the non-overtone sound memory 71 outputs the non-overtone sound data NF _j and NL _j to the scaling unit 61. The non-harmonic data NF _j and NL _j are input to the harmonic synthesis sound source 65 after scaling. On the other hand, overtone data Δ
F _i and L _i are input to the harmonic overtone pitch generator 62 and the harmonic overtone level generator 63. The overtone pitch generating section 62 generates frequency data TF of the musical tone frequency designated by the performance operator 50.
Generate _k (k = 0, 1, ...). Further, the harmonic overtone level generator 63 generates level data TL _k corresponding to each frequency data and matching the spectrum envelope. The generation method of these data is as follows.

First, the tentative frequency XF _k is calculated by the following equation. XF _k = k × SF (SF: F number) Next, the magnification x and the fractional part y of the fundamental wave frequency data F0 of the overtone data stored in the overtone memory 70 for this frequency are obtained. x = INT (XF _k / F ₀ ), y = FRAC (XF
_k / F ₀₎ Based on _{this, TF k = XF k + (} 1-y) × ΔF 'x + y × ΔF' x + 1 TL k = (1-y) × L 'x + y × L' x + Ask for ₁ . Here _{ΔF 'x, ΔF' x +} 1, L 'x, L' x + 1 is
The correction amounts of the frequency and level by scaling are considered from ΔF _x , ΔF _{x + 1} , L _x , and L _{x + 1} , respectively.

TF _k and TL _k thus obtained
Is input to the harmonic synthesis sound source 64. The harmonic 66 and the non-harmonic sound synthesized by the harmonic sound sources 64 and 65 are mixed by the mixer 66.
Is added and synthesized to generate musical tone waveform data. The tone waveform data is converted into an analog signal by the D / A converter 67 and output via the sound system 68.

The above operation is repeatedly executed while the performance operator 50 is being operated. That is, the overtone memory 7
0, a spectrum envelope is assumed based on the data sequentially read from the non-harmonic memory 71, and a change in timbre with the passage of time from the start of sounding is realized. Even in this case, there is no unnecessary change in tone color depending on the frequency of the musical tone to be generated.

Although FFT analysis is used for the above frequency analysis, other methods can be used, for example, a phase vocoder may be used. Further, although the spectrum envelope is estimated using only the level of the overtone component in the above embodiment, the spectrum envelope may be estimated in consideration of the level of the non-overtone component.

Further, in the performance device (FIG. 6) of the tone generation system of FIGS. 5 and 6, time series data sequentially read out to the overtone memory 70 and the non-overtone memory 71 according to the count value of the counter. Although it is memorized, these overtone data and non-overtone data are attacked, decayed,
It can also be simplified by using 3 sets of data of the release part.

[0043]

As described above, according to the waveform generator of the present invention, even if the frequency of the generated waveform is changed, the shape of the spectrum envelope formed by the fundamental wave and the harmonics of the sound does not change, so that the frequency It is possible to eliminate the change in tone color due to.

[Brief description of drawings]

FIG. 1 is a block diagram of a tone generation system according to an embodiment of the present invention.

FIG. 2 is a detailed block diagram of a harmonic overtone / non-harmonic overtone separation unit of the same tone generation system.

FIG. 3 is a diagram showing a configuration of a peak extracted by a peak extraction unit of the musical sound generation system.

FIG. 4 is a diagram showing a spectrum envelope when a musical tone is synthesized using the musical tone generating system and a conventional waveform memory sound source.

FIG. 5 is a block diagram of a recording device of a musical sound generating system according to another embodiment of the present invention.

FIG. 6 is a block diagram of a performance device of the musical sound generation system.

Claims (2)

(57) [Claims] 1. Data for generating a spectrum envelope and
Generates non-harmonic component data for synthesizing non-harmonic components
Data generating means, a fundamental frequency designating means for designating a fundamental frequency, and a frequency of a harmonic component based on the fundamental frequency.
Both the spectrum and determine means for determining the level of harmonics based on the envelope, the determined at the determined frequency and level at a constant means overtones formed
Minutes are combined and based on the non-harmonic component data
And a waveform synthesizing means for synthesizing non-harmonic components . 2. Data for generating a spectrum envelope and
Stores non-harmonic component data for synthesizing non-harmonic components
Data storing means, a fundamental frequency designating means for designating a fundamental frequency, and a frequency of a harmonic component based on the fundamental frequency.
Both the spectrum and determine means for determining the level of harmonics based on the envelope, the determined at the determined frequency and level at a constant means overtones formed
Minutes are combined and based on the non-harmonic component data
And a waveform synthesizing means for synthesizing non-harmonic components .