JPH10171462A - Method and device for analyzing and synthesizing sine wave - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable an efficient analysis and a natural synthesis by preventing spectrum from being intensively selected due to an effect of performance such as vibratos, etc. SOLUTION: When a plurality of sine wave candidates for synthesizing original sound signal Si are extracted as a result of an analysis by selecting a sine wave of which error from the original signal Si is a minimum in frequency analysis part 3 and repeating a processing fixed times by which a difference waveform between the original sound signal Si and the selected sine wave is made to be the next signal for analysis, a performance method detection part 8 extracts the modulated component to the sine wave included in the signal for analysis based on the selected sine wave as performance method information such as vibratos, tremolos, etc. The sine wave in the neighborhood frequencies of the selected sine wave is not selected, but modulated components are adopted as an analysis result as well as the selected sine wave. Also, the result of the analysis is modulated by an amplitude modulator 90 and a frequency modulator 10 based on the modulation component from which the selected sine wave has been detected, and is subtracted from the signal for analysis by a subtractor 11, and the difference waveform is adopted as the next signal for analysis. In the case of synthesis, a modulation processing based on the modulation component is provided to the selected sine wave before a synthetic processing of the waveforms is executed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a sine wave analysis / synthesis method and apparatus for sine wave analysis and synthesis of original sound signals such as musical sounds and voices used in a sound source device and the like.

[0002]

2. Description of the Related Art In a sound source apparatus of the analysis synthesis type, a sine wave included in an original sound signal is obtained by analyzing a sine wave of an original sound signal.
There is a sine wave analysis / synthesis method for performing synthesis based on these selected sine waves. In this type of sine wave analysis / synthesis method, an original sound signal is cut out, for example, by 10 ms at the time of analysis, and a sine wave (first sine wave) that minimizes an error from the cut out waveform is selected. Next, the same operation is performed on the difference waveform between the original sound signal and the selected first sine wave to select the second sine wave. Hereinafter, the same operation is performed a predetermined number of times to select a plurality of sine waves. At the time of synthesis, the original signal is synthesized by adding all the sine waves obtained by the analysis processing. The number of sine wave candidates to be selected is a maximum of 1 in the section of 10 to 100 ms.
When the number of candidates is small, natural synthesis becomes impossible, and when the number is large, analysis efficiency is reduced.

[0003]

However, in the above-described conventional sine wave analysis / synthesis method, when the original sound signal is subjected to sine wave analysis, the original sound signal has a variation in frequency or amplitude due to a vibrato or tremolo effect on the playing method. If a component is included, the spectrum concentrates near a specific frequency, and many similar sine wave candidates with only slightly different frequencies and amplitudes are selected in this part, and correspondingly, other frequency sine wave candidates are selected. There is a problem that information on the components is lost.

The present invention has been made in view of such a problem, and prevents a spectrum from being intensively selected due to a performance effect such as vibrato, thereby enabling efficient analysis and natural synthesis. It is an object of the present invention to provide a simple sine wave analysis and synthesis method and apparatus.

[0005]

A sine wave analyzer according to the present invention outputs frequency component information obtained by analyzing frequency components contained in an original sound signal and amplitude information thereof. Based on the specific frequency component information extracted in advance, a modulation component for the specific frequency component caused by the playing method is extracted, and a frequency component near the specific frequency component is analyzed when analyzing the frequency component. The method is characterized in that the component information is not extracted, and the modulation component caused by the playing method is analyzed together with the frequency component information and its amplitude information.

[0006] A sine wave synthesizer according to the present invention comprises:
A sine wave signal is generated based on the frequency component information and the amplitude information, and the generated sine wave signal is combined with a waveform modulated based on the modulation component information to generate a tone signal to which an effect by a specific playing method is added. It is characterized by combining.

Further, a sine wave analysis method according to the present invention
By repeating a predetermined number of times of selecting a sine wave with the least error from the analysis target signal and processing a difference waveform between the analysis target signal and the selected sine wave as a new analysis target signal, the original sound signal is obtained. In the sine wave analysis method of extracting a plurality of sine wave candidates for synthesis as analysis results, extracting a modulation component for the sine wave included in the analysis target signal based on the selected sine wave, and Without selecting a sine wave having a frequency close to the sine wave, the modulation component is used as an analysis result together with the selected sine wave, and the selected sine wave is modulated by the modulation component and subtracted from the signal to be analyzed. The differential waveform is used as the next signal to be analyzed.

In other words, although other useful information has been lost due to the concentration of spectra based on playing methods such as vibrato and tremolo, according to the sine wave analyzer according to the present invention, the frequency component contained in the original sound signal is reduced. At the time of analysis, a modulation component based on a specific frequency component generated by the playing method is detected from the original sound signal, and an analysis result is obtained by a combination of the frequency component and the modulation component.
Only representative frequency components are extracted even in a portion where the spectrum is concentrated, thereby preventing other useful frequency components from being lost.

Further, at the time of synthesizing, if a synthesizing process is performed after performing a modulating process based on a modulating component on a sine wave generated by the frequency component information and the amplitude information, an approximate waveform of the original sound signal can be easily synthesized. Can be. Moreover, according to the present invention, other useful frequency component information can be prevented from being lost, and time continuity and frequency continuity can be ensured by a modulation component as compared with discrete spectrum detection at regular intervals. Natural smooth synthesis becomes possible.

The sine wave analysis is executed by repeating a process of selecting a sine wave approximating the signal to be analyzed and a process of setting a difference waveform between the signal to be analyzed and the selected sine wave as the next signal to be analyzed. In this case, when calculating the difference waveform, the next signal to be analyzed can be efficiently obtained by subtracting the waveform obtained by modulating the selected sine wave with the modulation component from the signal to be analyzed.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a configuration of a sine wave analyzer according to one embodiment of the present invention. The original sound signal Si as the signal to be analyzed is
For example, PCM data sampled at a sampling frequency of several to several tens KHz, which is a FIFO
(First in First out) The data is sequentially stored in the buffer 1.
From the FIFO buffer 1, an original sound signal Si corresponding to, for example, 10 ms is sequentially cut out and stored in the register 2. The original sound signal Si stored in the register 2 is frequency-analyzed by the frequency analysis unit 3. As a frequency analysis method, a well-known FFT (Fast Fourier Transform) can be used. Here, the largest frequency component is selected from the extracted spectrum, and information on the frequency f and the amplitude v is output. The sine wave generator 4 generates a sine wave based on the frequency f and the amplitude v provided from the frequency analysis unit 3.

On the other hand, the 10
The ms original sound signal Si is also supplied to the band pass filter 5. The band-pass filter 5 is an adaptive filter that passes a band near the frequency f obtained by the frequency analysis unit 3. The output of the band pass filter 5 is provided to an AM demodulator 6 and an FM demodulator 7.
The AM demodulator 6 and the FM demodulator 7 together with the band pass filter 5 constitute a playing method detection unit 8, and the amplitude modulation component A for the first sine wave output from the sine wave generator 4.
M and the frequency modulation component FM are detected as performance method information.

The first sine wave output from the sine wave generator 4 is modulated by the AM modulator 9 and the FM modulator 10 based on the detected amplitude modulation component AM and frequency modulation component FM, respectively. This modulated signal is subtracted from the original sound signal Si by the subtracter 11, and the difference waveform is stored in the register 2 as the next signal to be analyzed. In the next process, the second sine wave and its modulation component are obtained by the same process as described above. By repeating the above processing, frequency information f, amplitude information v, and performance information AM and FM are obtained for each of the first to n-th sine waves, and these are transferred or stored as analysis results. Then, the next 10 ms analysis target signal is taken out from the FIFO buffer 1 to the register 2, and the same processing is repeated.

According to the sine wave analyzer, for example, FIG.
As shown in FIG. 2A, when the original sound signal Si includes a frequency fluctuation component Δf due to vibrato, a specific frequency fa detected first is set as a representative frequency, and the representative frequency is set as a center frequency. After passing through the band-pass filter 5 as shown in FIG. 2B and cutting off other components, the frequency modulation component FM and the amplitude modulation component AM from the representative frequency fa are extracted as performance method information.
It is possible to prevent a large number of frequency components in a portion where the spectrum near the representative frequency fa indicated by the dotted line (a) is concentrated. This can prevent other frequency components from being lost.

As for the playing method, there are tremolo, wah wah, etc. in addition to vibrato.
It is possible to synthesize mainly the representative frequency component and the amplitude modulation component AM, and for wah-wah, by detecting the amplitude modulation component AM for all sine wave candidates,
Can be synthesized.

FIG. 3 is a block diagram showing the configuration of a sine wave synthesizer corresponding to the above sine wave analyzer. Frequency information f and amplitude information v for specifying the first to n-th sine waves
Is supplied to the sine wave generator 21, and a sine wave corresponding to the frequency information f and the amplitude information v is output from the sine wave generator 21. The output sine wave is modulated by the AM modulator 22 and the FM modulator 23 based on the amplitude modulation component AM and the frequency modulation component FM, which are performance method information, and is input to the adder 24. The output of the adder 24 is
5 and is added to the next sine wave. This allows
The register 25 finally stores the waveform obtained by synthesizing all the modulated sine waves. This composite waveform is output via the FIFO buffer 26.

Note that f and v and performance information at the time of synthesis need not necessarily be obtained at the time of analysis. By changing f, v, and performance information, it is possible to synthesize voices having different pitches, levels, or effects. The above sine wave analyzer and sine wave synthesizer are:
The digital signal processor (DSP) and necessary software may be configured to have substantially the same function. Further, in the above embodiment, both the AM modulation component and the FM modulation component are extracted as performance method information. However, even when only one of the modulation components is extracted, a certain effect can be obtained.

[0018]

As described above, according to the present invention,
When analyzing the frequency components included in the original sound signal, a modulation component based on a specific frequency component generated by the playing method is detected from the original sound signal, and the analysis result is obtained by a combination of the frequency component and the modulation component. Only the representative frequency components are extracted even in the portion where the spectrum is concentrated, which can prevent the loss of other useful frequency components, and can reduce the amplitude variation more than the sampling of the discrete spectrum pattern. In addition, since the time continuity of the frequency fluctuation can be improved, it is possible to perform an efficient analysis and a natural synthesis.

[Brief description of the drawings]

FIG. 1 is a block diagram of a sine wave analyzer according to an embodiment of the present invention.

FIG. 2 is a diagram showing a spectrum extracted by the apparatus and characteristics of a bandpass filter.

FIG. 3 is a block diagram of a sine wave synthesizer according to one embodiment of the present invention.

[Explanation of symbols]

1, 26 ... FIFO buffer, 2, 25 ... register, 3
... frequency analysis unit, 4, 21 ... sine wave generator, 5 ... band-pass filter, 6 ... AM demodulator, 7 ... FM demodulator, 8 ...
Performance method detector, 9,22 ... AM modulator, 10,23 ... F
M modulator, 11 ... subtractor, 24 ... adder.

Claims (3)

[Claims] 1. A sine wave analyzer that outputs frequency component information and amplitude information obtained by analyzing frequency components included in an original sound signal, wherein a specific frequency component information previously extracted from the original sound signal is Based on the performance method, a modulation component for the specific frequency component due to the performance method is extracted. When analyzing the frequency component, frequency component information is not extracted for a frequency component near the specific frequency component, and the frequency component is not extracted. A sine wave analyzer, wherein a modulation component caused by the playing method is obtained as an analysis result together with the information and its amplitude information. 2. A sine wave signal is generated on the basis of frequency component information and amplitude information, and a waveform obtained by modulating the generated sine wave signal on the basis of modulation component information is synthesized to obtain an effect of a specific playing method. A sine wave synthesizing device for synthesizing the assigned tone signal. 3. A method of repeating a predetermined number of times of selecting a sine wave with the least error from the signal to be analyzed and setting a difference waveform between the signal to be analyzed and the selected sine wave as a new signal to be analyzed. In the sine wave analysis method of extracting a plurality of sine wave candidates for synthesizing an original sound signal as an analysis result, extracting a modulation component for the sine wave included in the analysis target signal based on the selected sine wave And selecting the modulation component together with the selected sine wave as an analysis result without selecting a sine wave of a frequency near the selected sine wave, modulating the selected sine wave with the modulation component, and performing the analysis. A sine wave analysis method, wherein a difference waveform subtracted from a target signal is used as a next signal to be analyzed.