JPH0239197A - Higher harmonic factor sound source system - Google Patents

Abstract

PURPOSE:To obtain a smooth sound change by providing the title system with an interpolating means for outputting the waveforms of respective parts of a sound signal synthesized by a higher harmonic factor synthesizing means and outputting the interpolated results of wave-forms of respective parts. CONSTITUTION:The higher harmonic factor synthesizing means 2 resynthesizes the waveforms of respective parts of a sound signal on the basis of applied higher harmonic factors. The interpolating means 8 outputs the waveforms of respective parts of the signal synthesized by the means 2 and outputs the interpolated results of the waveforms of respective parts. A higher harmonic factor extraction means 1 extracts the higher harmonic factor of the inputted sound signal and applies the extracted factor to the means 2. Since the storage and reproduction of sound waveforms is executed only on the basis of the higher harmonic factor of the sound signal, the capacity of processing data is extremely less, the feature of an original sound is not lost by the extraction and synthesis of a higher harmonic factor and the waveforms of respective parts of the sound signal can be interpolated. Consequently, a smooth sound change can be attained.

Description

[Detailed Description of the Invention] [Industrial Field of Application] The present invention is directed to a sound source method for extracting and synthesizing harmonic coefficients. By analyzing and synthesizing this sound signal, the amount of processing data is reduced, and by interpolating the waveform of each part of the synthesized waveform, smooth sound changes are realized.

[Prior Art] Conventionally, electronic musical instruments have been implemented as sampling instruments that record external sounds and play using these external sounds as a sound source. The method used was to memorize the information and then read it out and play it back. This PCM sound source method is widely used in musical instruments other than sampling instruments.

[Problem to be solved by the invention] However, when waveforms are stored in the form of PCM data,
Although it is possible to obtain a sound that is very close to the original sound, in order to memorize one waveform, all the level data for each step of that waveform must be memorized, so the data capacity becomes extremely large, making it difficult to record. In addition to requiring longer processing times for playback, etc., the data volume was large, making it difficult to process the sound.

The present invention has been made in order to solve the above-mentioned problems, and provides a sound source method that requires less data capacity and shorter data processing time when storing and reproducing sound waveforms, and therefore allows easier control of sound processing, etc. Our goal is to provide the following.

9. Means for Solving the Problem] In order to achieve the above object, in the present invention, the harmonic coefficient synthesis means resynthesizes the waveforms of each part of the sound signal based on the obtained harmonic coefficients. The interpolation means outputs the waveform of each part synthesized by the harmonic coefficient synthesis means, and also outputs the interpolated waveform of each part. In addition to this, the harmonic coefficient extracting means extracts the harmonic coefficients of the input sound signal, and provides the extracted harmonic coefficients to the synthesizing means.

[Function] In the above configuration, when storing and reproducing sound waveforms, this is done using only the harmonic coefficients of the sound IS number, so the amount of processing data is extremely small, and harmonic coefficient extraction and synthesis is used. Since the characteristics of the original sound are not lost, and the waveforms of each part of the sound signal are interpolated, smooth changes in sound can be achieved.

[Example 1] Hereinafter, an example that does not embody the present invention will be described with reference to the drawings.

FIG. 2 schematically shows a block circuit according to an embodiment of the present invention, and the specific circuit is as shown in FIG. The harmonic coefficient extraction unit 1 extracts formant harmonic coefficient data at each part of the envelope of the sound signal, and the harmonic coefficient synthesis unit 2 generates the original sound based on the extracted harmonic coefficient data. The signals are combined at each part of the envelope, and the waveform interpolation unit 3 outputs the combined waveforms of each part according to changes in the envelope, and also outputs intermediate waveforms obtained by interpolating each waveform. The envelope extraction unit 4 extracts an envelope from the input sound signal, and outputs the extracted envelope in parallel with the output of the waveform data from the waveform interpolation unit 3.
The multiplier 5 multiplies the signals and outputs them as musical tone signals.

Each part of the above configuration will be explained based on FIG.

The input sound signal is input to the A/D converter 12 after removing unnecessary high frequencies, for example, 3.4 kHz or higher, by a low-pass filter 11 .

The A/D converter 12 samples the input sound signal at a predetermined sampling frequency, e.g. 8 kHz, quantizes it with a predetermined number of bits, and stores it in the original waveform memory 13 for a predetermined length of time, e.g. 32 milliseconds. That is, 256 samples are stored. This sampling process is performed at predetermined timing T, for example, every 0.2 seconds, and the waveform of each part of the envelope of the input sound signal is sampled.

The Fourier transformer 14 performs scattering Fourier transform on the electronic sound 1λ in the original waveform memory 13,
It is converted into a frequency domain value and provided to the power spectrum calculator 16. The power spectrum calculator 16 performs calculations such as square addition of the real part and imaginary part of each frequency spectrum component given, calculates a power spectrum as shown in FIG. 5 (1), and stores it in the spectrum memory 17. .

The control unit 15 executes the harmonic coefficient extraction process shown in FIG. 4, which will be described later, on each spectrum data in the spectrum memory 17, and extracts the formant harmonic coefficients, that is, each peak, in the power spectrum characteristic diagram shown in FIG. The wavelength data λ1 (I-1, 2, 3...) and level data Pi (i=1.2.3...) at the point are extracted and stored in the harmonic coefficient memory 18. This extraction process is performed as follows. , respectively, for each part of the waveform of the envelope of the input sound signal.

In harmonic coefficient synthesis, the control unit 15 sequentially outputs wavelength data ^1 and level data Pi at the peak point of each harmonic coefficient from the harmonic coefficient memory 18, and each wavelength data λ1 is stored in the programmable timer 19. , and each level data P is set in the level memory 20. The programmable timer 19 gives an increment signal to the address counter 21 at a period corresponding to the set wavelength data λ, and as a result, a sine wave with a frequency corresponding to the wavelength data λ is output from the sine wave memory 23 through the multiplexer 22. The signal is read out, multiplied by the above-mentioned corresponding level data Pi in the multiplier 24, and then sent to the accumulator 25.
is accumulated. One programmable timer and eight address counters 21 are provided, and the level memory 20 consists of eight stages of registers.Through time-sharing processing, up to eight harmonic coefficient data are cumulatively synthesized to form one waveform. The harmonic coefficients are synthesized and stored in the synthesized waveform memory 26. Composite waveform memory 2 of such harmonic coefficient composite waveform data
6 is executed for each part of the waveform of the envelope of the input sound signal.

In this case, as shown in Figure 3, the harmonic components generally decrease from the attack to the release of the envelope, so the harmonic coefficients are calculated only for one representative waveform part of the envelope of the input sound signal. Extraction and synthesis may be performed, and harmonic coefficient synthesis may be performed for the pond waveform portion by sequentially adding or deleting wavelength data λ1 and level data P1 for appropriate harmonic frequencies.

In addition, for each waveform of each part of the envelope of the input sound signal written in this way, ff(
Harmonic coefficient synthesis of waveforms at each level is also performed: fortissimo), mf (mezzo forte imp (mezzo piano), and pp (pianissimo). Generally, as shown in Figure 3, p
As the sound intensity increases from p to ff, the number of harmonic components increases, so wavelength data λi and level data P1 for appropriate harmonic frequencies are sequentially added or deleted accordingly. Then, harmonic coefficient synthesis can be performed.

Each harmonic coefficient composite waveform written in the composite waveform memory 26 is sequentially read out and outputted by the interpolation control unit 27 under the control of the control unit 15 when a musical tone is emitted, and is also subjected to waveform interpolation shown in FIG. 6, which will be described later. Through the processing, interpolated waveforms of each part are also output. In this case, each waveform is sequentially read out at every extraction timing T when harmonic coefficients of the waveform of each part of the envelope of the input sound signal are extracted. When writing and reading harmonic coefficient composite waveform data from the composite waveform memory 26, access address data is given by the address controller 28,
This address controller 28 controls the control unit 15 during writing;
The interpolation control unit 27 controls the continuous output. In this way, the harmonic coefficient composite waveform data read out and interpolated is multiplied by the envelope data in the multiplier 29, and sent to the sound system 31 via the D/A converter 30, where it is emitted as a musical tone.

On the other hand, the envelope waveform of the input sound signal is extracted by the envelope extraction circuit 32, and then sent to the A/D.
Via the converter 33, the control unit 15 stores the data in the envelope memory 3.4 in the form of digital data. The envelope data in the envelope memory 34 is read out by the control unit 15 when a musical tone is emitted, and is multiplied by touch data, which will be described later, in the multiplier 35 to create an envelope 1 corresponding to the intensity of the sound. and is multiplied by the harmonic coefficient composite waveform data.

Incidentally, among the note data and touch data given from the keyboard 36, the touch data is given to the multiplier 35 and also given to the interpolation control section 27, so that the touch data is given to the multiplier 35 and also given to the interpolation control section 27 to calculate the corresponding sound of each stored waveform in the synthesized waveform memory 26. Waveform data corresponding to the strength is output one after another. In addition, the interpolation control unit 27
The speed at which waveform data is read out from the synthesized waveform memory 26 is controlled, and a musical tone having a pitch corresponding to the note data is generated.

Next, the operation of this embodiment will be described.

FIG. 4 shows a flowchart of the harmonic coefficient extraction and synthesis process. When spectrum data is set in the spectrum memory 17, the control unit 15 selects three successive powers from among the power spectra for each frequency. Step S1 following step Si, in which the level data Pa, Pb, and Pc of the spectrum are read out sequentially, is simply expressed as Sl, and the same applies to the other steps. ), it is determined whether the level data Pb in the middle is larger than the level data Pa and Pc on both sides (32). This is done by searching for the peak point of the power spectrum, as shown in FIG. 5 (2). This is to do so.

Once the location of the peak point is known, the vicinity is approximated by a quadratic equation, and accurate wavelength data λ and level data P of the peak point are obtained.
i will be found (33-35). This is because the exact peak point does not necessarily match the position of the level data pb. Specifically, the coefficient A(PC+Pa)/2-Pb, B=(PC
-Pa)/2 is determined (33), and based on this, wavelength data λ--B/2A and level data Pi=-B2/4A
+Pb is obtained (S4, S5), written to the harmonic coefficient memory 18 (S6), and this harmonic coefficient extraction process is performed on all data of the power spectrum (S7). After extracting multiple harmonic coefficients, next time is harmonic coefficient memory 18
Read each wavelength data λ and level data P1 from (
S8), the programmable timer 19... is set in the level memory 20 (S9), and this is done for all wavelength data λI and level data Pi (310)
In this case, for waveform data of different sound intensities, harmonic coefficients are obtained by sequentially adding or deleting wavelength data λi and level data P for appropriate harmonic frequencies as described above. Synthesis is performed.

In this way, harmonic coefficient extraction and synthesis can be easily performed, and the characteristics of the input sound signal can be extracted from the wavelength data λ of the harmonic coefficients.
i, level data Pi alone can be grabbed and stored, reducing the amount of data and simplifying data processing.

FIG. 6 shows a flowchart of waveform interpolation processing. As shown in FIG. 7, since the waveform data in the composite waveform memory 26 is extracted at each timing T when harmonic coefficients are extracted, this interpolation is performed in several steps during this time T. In this example, we set the time T to 4.
Interpolation is performed in three steps (the step value is n) for each equally divided time interval, and the waveform changes gradually and smoothly from one waveform to the next.

First, at the start of musical sound emission, the interpolation control section 27 reads out the first synthesized waveform data in the waveform data group corresponding to the touch data from the synthesized waveform memory 26.
It continues repeating until time t has elapsed (R2), and after confirming that the waveform currently being output is not the last waveform (R3), the level data P of the waveforms read so far is read. The interpolated intermediate level between the level data P1 of the waveform to be read next and the interpolated value Po+ (Pl-po)t n/4 is repeatedly output (R4), and each time t elapses (R5) , the interpolation step n is set to 1 as shown in FIG.
Advance step by step (R6) and gradually approach the next waveform, when the interpolation step in or r 4 is reached (R7)
, controls the address controller 28 (R8), reads out the next waveform, and performs the waveform successive processing (RlmR) described above.
3) Execute interpolation processing (R4-R7) 9Next, once the last waveform data has started to be successively output (R3), from then on, the waveform data after n will be continuously output until this musical sound emission is finished. continue.

In this way, by interpolating each waveform read from the composite waveform memory 26, smooth sound changes can be realized.
Through this interpolation, intermediate waveforms of each waveform are stored in the composite waveform memory 2.
6, the amount of data stored can be reduced accordingly.

FIG. 8 shows a second embodiment, in which a plurality of formant memories 41 are provided, and based on the formant data (data indicating the formant shape) from the formant memories 41, the harmonic coefficient extraction unit 1
The harmonic coefficients, that is, the wavelength data λ1 and level data Pi extracted in , are corrected in terms of range, strength, etc. in the data correction section 40, and the harmonic coefficient synthesis section 2 converts the waveform using the corrected data λi' and Pi'. The signals are synthesized, interpolated by the waveform interpolation section 3, and output. In this case, the storage contents of the formant memory 41 may be not one type but a plurality of types, and the harmonic coefficients may be corrected using data obtained by interpolating between the plurality of types of formant data.

The present invention is not limited to the above embodiments, and can be modified in various ways without departing from the spirit of the present invention.9 For example, the extraction and synthesis of harmonic coefficients may be performed using other methods, such as step values of waveform interpolation, etc. Other methods may also be used.

[Effects of the Invention] As detailed above, according to the present invention, it is possible to easily perform harmonic coefficient extraction and synthesis, and it is possible to grasp and store the characteristics of an input sound signal only with harmonic coefficient data. It is possible to reduce the amount of data and simplify data processing. By interpolating each harmonic coefficient composite waveform,
Smooth sound changes can be achieved, and this interpolation eliminates the need to store intermediate waveforms for each waveform, reducing the amount of data stored.

[Brief explanation of the drawing]

1 to 8 show embodiments of the present invention.
The figure is a circuit diagram for extracting and synthesizing harmonic coefficients and waveform interpolation; FIG. 2 is a schematic circuit diagram of FIG. 1; and FIG. FIG. 4 is a flowchart of harmonic coefficient extraction and synthesis processing, FIG. 5 is a power spectrum characteristic diagram of the input sound signal of the spectrum memory 17, and FIG. 6 is a flowchart of waveform interpolation processing. , FIG. 7 is a diagram showing the contents of waveform interpolation from one waveform to the next waveform, and FIG. 8 is a diagram showing another embodiment. DESCRIPTION OF SYMBOLS 1... Harmonic coefficient extraction part, 2... Harmonic coefficient synthesis part, 3... Waveform interpolation part, 13... Original waveform memory, 14
...Fourier transformer, 15...control unit, 16...
Power spectrum calculator, 17... Spectrum memory, 18... Harmonic coefficient memory, one... Programmable timer, 26... Synthetic waveform memory, 27... Interpolation control unit. Sound I 83 8 addition defeat glaze return beef point, -L-j-'' two waveform correction dark treatment unzuke sae depo U cedar

Claims (1)

[Claims] 1. Harmonic coefficient synthesis means for synthesizing the waveforms of each part of a sound signal based on the given harmonic coefficients; and outputting the waveforms of each part synthesized by the harmonic coefficient synthesis means. , and an interpolation means for outputting an interpolated waveform of each part. 2. A harmonic coefficient extracting means for extracting harmonic coefficients of the input sound signal, and resynthesizing the waveforms of each part of the original sound signal based on the harmonic coefficients extracted by the harmonic coefficient extracting means. A harmonic coefficient sound source characterized by comprising a harmonic coefficient synthesis means, and an interpolation means for outputting a waveform of each part synthesized by the harmonic coefficient synthesis means and an interpolated waveform of each part. method.