JP2000099009A - Audio signal encoding method - Google Patents

Abstract

(57) [Problem] To efficiently convert non-linear code data such as MIDI data. SOLUTION: An audio signal to be encoded is converted into PCM code, fetched as audio data, and a plurality of unit sections d1 to d5 are set on a time axis (FIG. 1 (a)). Fourier transform is performed for each unit section to obtain a spectrum (Figure
(b)). At this time, 128 note numbers (0 to 127) are defined at equal intervals on the frequency axis f of the logarithmic scale, and the execution strength E is obtained only for the frequency corresponding to the 128 note numbers ( Figure (c)). P note numbers n (d1,1) in order of increasing effective intensity
Ｎn (d1, P) are extracted and arranged at time positions corresponding to each unit section on the P tracks. The note number on each track is represented by MIDI data, and the original sound signal is reproduced as a P-channel stereo sound.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for encoding an audio signal, and more particularly to a technique for encoding an audio signal given as a time-series intensity signal, and decoding and reproducing the audio signal.
In particular, the present invention is suitable for a process of efficiently converting a general acoustic signal into MIDI-format code data, and is expected to be applied to various industrial fields for recording voice.

[0002]

2. Description of the Related Art As a technique for encoding an audio signal, a PC is used.
The M (Pulse Code Modulation) method is the most widespread method, and is currently widely used as a recording method for audio CDs and DATs. The basic principle of this PCM method is that an analog audio signal is sampled at a predetermined sampling frequency, and the signal strength at each sampling is quantized and represented as digital data. The more it is, the more faithful it is possible to reproduce the original sound. However, the higher the sampling frequency and the number of quantization bits, the larger the required information amount. Therefore, as a technique for reducing the amount of information as much as possible, an ADPCM (Adaptive Differentia) that encodes only a signal change difference is used.
l Pulse Code Modulation) is also used.

[0003] On the other hand, MIDI (Musical Instrume) was born from the idea of encoding musical instrument sounds by electronic musical instruments.
The Digital Interface (nt Digital Interface) standard has also been actively used with the spread of personal computers. Code data according to the MIDI standard (hereinafter, MID)
I data) is basically data describing an operation of playing a musical instrument, such as which keyboard key of the musical instrument was played and at what strength, and the MIDI data itself contains the actual sound. No waveform is included. Therefore, when reproducing the actual sound, the MI which stores the waveform of the musical instrument sound is used.
A DI sound source is required separately. However, the P
It has the feature that the amount of information is extremely small as compared with the case where sound is recorded by the CM method, and its high encoding efficiency has attracted attention. The encoding and decoding technology based on the MIDI standard is now widely used in software for playing musical instruments, practicing musical instruments, and composing music using a personal computer, and is also widely used in fields such as karaoke and game sound effects. Have been.

[0004]

As described above, the PC
In the case of encoding an audio signal by the method of M, the amount of information becomes enormous if sufficient sound quality is to be ensured, and the load of data processing must be increased. Therefore, usually
In order to keep the amount of information to a certain extent, we have to compromise on some sound quality. Of course, if the encoding method based on the MIDI standard is adopted, it is possible to reproduce a sound having a sufficient sound quality with a very small amount of information.
Since the I standard itself is originally for encoding the operation of musical instrument performance, it cannot be widely applied to general sound. In other words, in order to create MIDI data, it is necessary to actually play a musical instrument or prepare musical score information.

[0005] As described above, there are advantages and disadvantages in the encoding method of the audio signal in both the conventional PCM method and the MIDI method, and a small amount of information is sufficient for general audio. Sound quality cannot be ensured. However, there is an increasing demand for efficient encoding of general audio. In the field of dealing with human voices and singing voices, so-called vocal sound, such a request has been strongly issued for some time. For example, in fields such as language education, vocal education, and criminal investigation, there is a strong need for a technology for efficiently encoding vocal acoustic signals. To meet such demands, Japanese Patent Application No. 9-273949 discloses MI.
A new encoding method that can use DI data has been proposed. In this method, a procedure is performed in which a plurality of unit sections are set along the time axis of an acoustic signal, a spectrum is obtained by performing a Fourier transform for each unit section, and MIDI data corresponding to the spectrum is created. . However, MIDI data is originally data corresponding to musical notes, and has non-linear characteristics with respect to frequency. On the other hand, the conventional general Fourier transform technique is based on obtaining a spectrum using a linear frequency axis. For this reason, when the conventional general Fourier transform technique is used, there has been a problem that conversion to nonlinear code data such as MIDI data cannot be performed efficiently.

Accordingly, an object of the present invention is to provide an audio signal encoding method capable of efficiently performing conversion into nonlinear encoded data such as MIDI data.

[0007]

(1) A first aspect of the present invention is an audio signal encoding method for encoding an audio signal given as a time-series intensity signal, which is to be encoded. A section setting step of setting a plurality of unit sections on the time axis of the acoustic signal, a plurality of M measurement points are discretely defined on the frequency axis, and a total of M measurement points are respectively associated with the M measurement points. A code definition step of defining the number of code codes; and an intensity calculation step of obtaining, for each unit section, a spectrum intensity of a frequency component corresponding to M measurement points included in an audio signal in the unit section. Based on the spectrum intensity obtained in the intensity calculation stage, for each individual unit section, P representative code codes representing the unit section are extracted from the M total code codes, and these extracted codes are extracted. And an encoding step of expressing an audio signal of each unit section by using the representative code and the spectrum intensity thereof.

(2) A second aspect of the present invention is the above-mentioned first aspect.
In the method for encoding an acoustic signal according to the aspect, in the code definition step, a plurality of M measurement points are discretely defined so as to be equally spaced on a frequency axis of a logarithmic scale.

(3) The third aspect of the present invention is the above-mentioned second aspect.
In the audio signal encoding method according to the aspect, in the code definition step, note numbers used in MIDI data are used as a plurality of M code codes, and in the encoding step, an audio signal of each unit section is represented by a representative number. An MI consisting of data indicating a note number extracted as a code code, a velocity determined based on its spectrum intensity, and a delta time determined based on the length of the unit section.
It is represented by code data in DI format.

(4) The fourth aspect of the present invention is the above-mentioned first aspect.
In the audio signal encoding method according to the third to third aspects, when calculating the spectrum intensity S (m) at the m-th measurement point corresponding to the frequency f (m) in the intensity calculation stage, Are calculated to find correlations with M sine functions and cosine functions having frequencies corresponding to.

(5) A fifth aspect of the present invention is the above-mentioned first aspect.
In the audio signal encoding method according to the fourth to fourth aspects, a weight function defining weighting over the section length of the unit section is prepared in the intensity calculation step, and the audio signal in the unit section is multiplied by the weight function. Is used to determine the spectrum intensity.

(6) The sixth aspect of the present invention is the above-mentioned first aspect.
In the audio signal encoding method according to the fifth to fifth aspects, in the section setting step, setting is performed such that adjacent unit sections partially overlap on the time axis.

(7) The seventh aspect of the present invention is the above-mentioned first aspect.
In the audio signal encoding method according to any one of the first to sixth aspects, the audio signal to be encoded is set at a predetermined sampling frequency F
And the amplitude value of the x-th sample is represented by A
(X) is taken as sound data, and each unit section is set for the taken sound data.
In the intensity calculation step, for a unit section including a total of K samples starting from the h-th sample, the frequency f
When calculating the spectrum intensity S (m) at the m-th measurement point corresponding to (m), a predetermined weight function W
Using (k), S (m) = (1 / K) · Σ _{k = 0 to (K−1)} (W
(K) · A (h + k) · exp (−j2πf (m) ·
(H + k) / F)).

(8) An eighth aspect of the present invention is the above-mentioned first aspect.
In the audio signal encoding method according to any one of the first to sixth aspects, the audio signal to be encoded is set at a predetermined sampling frequency F
And the amplitude value of the x-th sample is represented by A
(X) is taken as sound data, and each unit section is set for the taken sound data.
In the intensity calculation step, for a unit section including a total of K samples starting from the h-th sample, the frequency f
When calculating the spectrum intensity S (m) at the m-th measurement point corresponding to (m), a predetermined weight function W
Using (k), S (m) = (1 / K) · Σ _{k = 0 to (K−1)} (W
(K) · A (h + k) · exp (−j2πf (m) · k
/ F)) is used.

(9) The ninth aspect of the present invention is the above-mentioned first aspect.
In the audio signal encoding method according to the eighth to eighth aspects, in the encoding step, a plurality of P representative code codes extracted for each unit section are distributed and arranged on a plurality of tracks, and are adjacently arranged on the same track. When the representative code arranged in such a manner satisfies a predetermined similarity condition, a process of integrating the representative code arranged adjacent to a single representative code is performed.

(10) A tenth aspect of the present invention provides the audio signal encoding method according to the ninth aspect, wherein
When distributing and arranging the representative code codes on a plurality of tracks, the order of distribution is adjusted so that the probability that the representative code codes arranged adjacently on the same track satisfy the similar condition is increased. It is like that.

(11) According to an eleventh aspect of the present invention, a program for executing the audio signal encoding method according to the first to tenth aspects is recorded on a computer-readable recording medium. It was made.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below based on an embodiment shown in the drawings.

§1. Audio signal encoding method according to the present invention
The basic principle beginning of law, the basic principle of the method of encoding an acoustic signal according to the present invention with reference to FIG. 1 will be described. Now, suppose that an analog sound signal is given as a time-series intensity signal as shown in FIG. In the illustrated example, the horizontal axis represents time t, and the vertical axis represents amplitude (intensity), and this acoustic signal is shown. Here, first, a process of capturing the analog audio signal as digital audio data is performed. This can be done by using a conventional general PCM technique, sampling the analog audio signal at a predetermined sampling period, and converting the amplitude into digital data using a predetermined quantization bit number. Here, for convenience of explanation, PCM
The waveform of the sound data digitized by the method of
Will be shown with the same waveform as the analog audio signal of FIG.

Subsequently, a plurality of unit sections are set on the time axis of the audio signal to be encoded. In the example shown in FIG. 1A, six times t1 to t6 are equally spaced on the time axis t.
Are defined, and five unit sections d1 to d5 having these times as a start point and an end point are set (a more practical section setting method will be described later).

When the unit section is set in this way, a Fourier transform is performed on the acoustic signal of each unit section to create a spectrum (actually, as described in §3, this is different from a general Fourier transform) Method). At this time, it is desirable to apply a filter to the cut-out sound signal using a weighting function such as a Hanning Window or the like to perform a Fourier transform. In general, the Fourier transform is assumed to have an infinite number of similar signals before and after the cut-out section. Therefore, when a weighting function is not used, high frequency noise often appears on a created spectrum. In such a case, it is desirable to use a weighting function such as a Hanning window function in which the weights at both ends of the section become 0. The Hanning window function H (k) is a function given by H (k) = 0.5−0.5 * cos (2πk / L) for k = 1... is there.

FIG. 1B shows an example of a spectrum created for the unit section d1. In this spectrum, the frequency components (0 to F: where F is a sampling frequency) included in the acoustic signal in the unit section d1 are indicated by the frequency f defined on the horizontal axis, and defined on the vertical axis. The complex intensity A indicates the complex intensity for each frequency component.

Next, a plurality of M code codes are discretely defined corresponding to the frequency axis f of the spectrum. In this example, note numbers n used in MIDI data are used as code codes, and 128 code codes from n = 0 to 127 are defined. The note number n is a parameter indicating the musical scale of the note. For example, the note number n = 69 indicates the “ra tone (A3 tone)” at the center of the keyboard of the piano, and corresponds to a sound of 440 Hz. Thus, for 128 note numbers,
In each case, since a predetermined frequency is associated, 128 note numbers n are discretely defined at predetermined positions on the frequency axis f of the spectrum.

Here, note number n indicates a logarithmic scale at which the frequency doubles when the octave is increased by one octave, and thus does not correspond linearly to frequency axis f. Therefore, here, the frequency axis f is represented by a logarithmic scale, and an intensity graph in which a note number n is defined on the logarithmic scale axis is created. FIG. 1C shows an intensity graph for the unit section d1 created in this way. The horizontal axis of this intensity graph is obtained by converting the horizontal axis of the spectrogram shown in FIG. 1 (b) into a logarithmic scale, and note numbers n = 0 to 0.
127 are plotted at equal intervals. On the other hand, the vertical axis of this intensity graph is obtained by converting the complex intensity A of the spectrum shown in FIG. 1B into the effective intensity E, and indicates the intensity at the position of each note number n. In general, the complex intensity A obtained by Fourier transform is represented by a real part R and an imaginary part I, but the effective intensity E is expressed as E = (R ²
+ I ² ) ^1/2 .

The thus obtained intensity graph of the unit section d1 is a graph showing, as an effective intensity, a ratio of each vibration component corresponding to the note number n = 0 to 127 with respect to the vibration component included in the acoustic signal of the unit section d1. be able to. Therefore, based on each effective intensity shown in this intensity graph, P note numbers are selected from all M (M = 128 in this example) note numbers, and the P note numbers n are selected. Is extracted as a representative code representing the unit section d1. Here, for convenience of explanation, it is assumed that P = 3 and three note numbers are extracted as representative code codes from a total of 128 candidates. For example, if the extraction is performed based on the criterion of “extracting P code codes from candidates in descending order of strength”, in the example shown in FIG. 1C, note number is used as the first representative code code. n (d1,1) is the note number n (d
1, 2) are extracted as note number n (d1, 3) as the third representative code.

When the P representative code codes have been extracted in this way, the sound signal of the unit section d1 can be represented by these representative code codes and their effective intensities. For example, in the case of the above example, note numbers n (d1, 1) and n in the intensity graph shown in FIG.
(D1, 2) and n (d1, 3) have an effective intensity of e
If (d1,1), e (d1,2), and e (d1,3), the sound signal of the unit section d1 can be represented by the following three data pairs.

N (d1,1), e (d1,1) n (d1,2), e (d1,2) n (d1,3), e (d1,3) Processing for unit section d1 However, the same processing is performed separately for each of the unit sections d2 to d5, and a representative code and data indicating its strength are obtained. For example, unit section d
For two, three data pairs of n (d2,1), e (d2,1) n (d2,2), e (d2,2) n (d2,3), e (d2,3) can get. The original audio signal can be encoded by the data obtained for each unit section in this manner.

FIG. 2 is a conceptual diagram of encoding by the above method. FIG. 2 (a) shows a state in which five unit sections d1 to d5 are set for the original sound signal, as in FIG. 1 (a), and FIG. Are shown in the form of musical notes. In this example,
Three representative code codes are extracted for each unit section (P = 3), and data relating to these representative code codes is stored in three tracks T1 to T3. For example, the representative code codes n (d1,1), n (d1,2), n (d) extracted for the unit section d1
1, 3) are accommodated in tracks T1, T2, T3, respectively. However, FIG. 2B is a conceptual diagram showing the code data obtained by the present invention in the form of musical notes.
In practice, data relating to the intensity is added to each note. For example, the track T1 has note numbers n (d1,1), n (d2,1), n (d3,
1) e (d1,
1), e (d2, 1), e (d3, 1)... Are stored.

It is not always necessary to adopt the MIDI format as the encoding format in the present invention. However, since the MIDI format is the most widely used as this type of encoding format, the MIDI format code data is practically used. Is most preferably used. In the MIDI format, "note on" data or "note off" data exists with "delta time" data interposed. The “note-on” data is data that designates a specific note number N and velocity V to instruct the start of performance of a specific sound, and the “note-off” data is a specific note number N and velocity V. Is the data for instructing the end of the performance of a specific sound by designating. The “delta time” data is data indicating a predetermined time interval. Velocity V is a parameter indicating, for example, the speed at which a piano keyboard or the like is depressed (velocity at the time of note-on) and the speed at which a finger is released from the keyboard (velocity at the time of note-off). Or it indicates the strength of the performance end operation.

In this embodiment, as described above, for the i-th unit section di, P note numbers n (di, 1), n (di, 2),
, N (di, P) are obtained, and the effective intensities e (di, 1), e (di, 2),.
i, P) are obtained. Therefore, in the present embodiment, the code data in the MIDI format is created by the following method. First, note number N described in "note-on" data or "note-off" data is:
The obtained note numbers n (di, 1), n (di,
2),..., N (di, P) are used as they are. On the other hand, as the velocity V described in the “note-on” data or the “note-off” data, the obtained effective intensities e (di, 1), e (di, 2),.
i, P) is normalized such that the value is in the range of 0 to 1,
The value obtained by multiplying the square root of the normalized effective intensity E by 127 is used. That is, assuming that the maximum value of the effective intensity E is Emax, the value V obtained by the calculation of V = (E / Emax) ^1/2 · 127 is used as the velocity. Alternatively, the value V obtained by the calculation of V = log (E / Emax) · 127 + 127 (however, when V <0, V = 0) may be used as the velocity. The “delta time” data may be set according to the length of each unit section.

After all, in the above-described embodiment, MIDI code data composed of three tracks is obtained. If this MIDI coded data is reproduced using three MIDI sound sources, an acoustic signal is reproduced as six-channel stereo reproduced sound.

The encoding process according to the above-described procedure is actually executed using a computer. A program for implementing the encoding process according to the present invention can be supplied by being recorded on a computer-readable recording medium such as a magnetic disk or an optical disk. The data is likewise
The information can be supplied by being recorded on a computer-readable recording medium such as a magnetic disk or an optical disk.

§2. More practical section setting method So far, the basic principle of the audio signal encoding method according to the present invention has been described. Hereinafter, a more practical encoding method will be described. Here, a more practical method for setting the section will be described. In the example shown in FIG. 2A, five unit sections d1 to d5 are set on the time axis t with six times t1 to t6 defined at equal intervals as boundaries. When encoding is performed based on such a section setting, discontinuity of sound is likely to occur at a boundary time during reproduction. Therefore, in practice, it is preferable to set a section in which adjacent unit sections partially overlap on the time axis.

FIG. 3A shows an example in which a partially overlapping section is set as described above. The unit section d shown
1 to d4 partially overlap each other, and if the above-described processing is performed based on such section setting, encoding as shown in the conceptual diagram of FIG. become. In this example, each note is arranged at each reference position with the center of each unit section as a reference position, but the reference position relative to the unit section does not necessarily need to be set at the center. The conceptual diagram shown in FIG.
Compared with the conceptual diagram shown in (b), it can be seen that the density of notes has increased. When such overlapping sections are set, the number of code data to be created increases, but natural coding that does not cause sound discontinuity during reproduction can be performed.

FIG. 4 is a diagram showing a specific method for setting sections that partially overlap on the time axis. In this specific example, an audio signal is sampled at a sampling frequency of 22 kHz, taken in as digital audio data, the section length L of each unit section is set to 1024 samples (about 47 msec), and the shift amount for each unit section is set. Is set to 20 samples (about 0.9 m).
sec). That is, for any i, the offset on the time axis between the start point of the i-th unit section and the start point of the (i + 1) -th unit section is set as the offset length ΔL. For example, the first unit section d1 includes the 1st to 1024th samples, and the second unit section d2 includes the 21st to 1044th samples shifted by 20 samples.

As described above, when sections that partially overlap on the time axis are set, a considerable number of samples are commonly used in adjacent unit sections, and the spectrum obtained for each unit section is used. No significant difference is expected. For example, in the case of the above example, comparing the first unit section d1 and the second unit section d2, the 21st to 1024th samples are completely used in both unit sections, and both units are used. The difference will depend on only 20 samples. However, fortunately, in the Fourier transform processing described in §3, a phase difference corresponding to 20 samples occurs, so that a great difference occurs in the complex intensity A in both unit sections. However, it is expected that there is not much difference in the effective strength E. in this way,
If a sufficient difference is not obtained between the spectra of adjacent unit sections, it is not possible to follow a rapidly changing acoustic signal, and as a result, there is a problem that the time resolution is reduced. In order to cope with such a problem, a countermeasure may be taken such that a difference of only 20 samples causes a large change on the input side of the Fourier transform.

Therefore, the inventor of the present application has devised that the weighting function described in §1 is modified so as to emphasize 20 changing samples. The well-known Hanning window function described above works in the direction of suppressing the fluctuation of the adjacent section, and therefore has an opposite effect from the viewpoint of coping with the above problem.
Therefore, while inheriting the feature of the Hanning window function that the weights at both ends of the section are reduced, a function that emphasizes 20 samples was devised and actually applied. Specifically, assuming that the section length of the unit section is L and the offset length is ΔL, α and β such that α = L / 2−ΔL / 2 β = L / 2 + ΔL / 2 are determined, and are expressed in the section [α, β]. Section near the center (section of width ΔL defined at the center position of the unit section)
When k = 1... Α, W (k) = 0.5−0.5 * cos (πk / 2α) When k = α.β, W (k) = 0.5−0.5 * cos (π (k−α) / Δ
L + π / 2) When k = β... L W (k) = 0.5−0.5 * cos (π (k−β) / 2
α + 3π / 2) may be used as the weighting function. The improved window function W (k) is a distribution function that is deformed to have a narrow width so that the half-value width becomes exactly ΔL. Experiments using this function have confirmed a sufficient effect.

§3. Efficient calculation of spectral intensity
Act Now, according to the principle described with reference to FIG 1, the basic steps of the encoding method according to the present invention, first, FIG. 1 (a), the plurality of unit sections on a time axis of the acoustic data d1, d2, d
Are set, and the Fourier transform is performed on the acoustic data in the section d1 to obtain a spectrum as shown in FIG. 1 (b). As shown in FIG. 1 (c), the peak frequency of this spectrum is Some corresponding codes n (d1,1), n (d
The sound signal of the section d1 is expressed by (1, 2), n (d1, 3). Here, FIG. 1 (b)
An efficient calculation method for obtaining a spectrum as shown in FIG.

When obtaining a spectrum as shown in FIG. 1 (b) for a signal having a vibration component as shown in FIG. 1 (a), a Fourier transform is generally used. An operation is performed using a fast Fourier transform (FFT) technique. However, general Fourier transform is based on the premise that a spectrum using a linear frequency axis is obtained, and is not necessarily suitable for conversion to nonlinear code data such as MIDI data. This is due to the following reasons.

Now, consider a Fourier spectrum on a linear scale as shown in FIG. This Fourier spectrum is a graph in which the horizontal axis represents frequency f on a linear scale and the vertical axis represents spectrum intensity. Here, a plurality of M measurement points are discretely defined at regular intervals on the horizontal axis (frequency axis), and the spectrum intensity is shown in a bar graph for each measurement point. The lower column of the graph shows the numbers of the respective measurement points, and the lower column of the graph shows the frequency values corresponding to the respective measurement points. In this example, an acoustic signal is taken in as data at a sampling frequency F = 22.05 kHz, and the number M of measurement points is set to 1024. Therefore, each of a total of 1024 measurement points from the 0th measurement point at which the frequency f = 0 to the 1023th measurement point at which the frequency f = 111014 Hz (almost の of the sampling frequency F) In, the spectrum intensity corresponding to the length of the bar graph is obtained. In a general Fourier transform, spectrum intensities are obtained for a large number of measurement points defined at equal intervals on the linear frequency axis.

However, as shown in FIG. 5, a spectrum whose intensity is obtained at measurement points defined at equal intervals on a linear frequency axis is converted to a code having nonlinear characteristics with respect to frequency, such as MIDI data. It is not efficient to use it for conversion to a system. FIG. 6 rewrites the frequency axis of the spectrum shown in FIG. 5 on a logarithmic scale. The lower column of the graph indicates the number of each measurement point, and the lower column of the graph indicates the note number (corresponding to log f) associated with each of the measurement points. The point that the number of measurement points is M = 1024 is the same as that of FIG. 5, but since the frequency axis is a logarithmic scale, the measurement points are not arranged at equal intervals on the horizontal axis. In other words, the arrangement of the measurement points is coarse in the low frequency region, but the arrangement of the measurement points becomes dense as it goes to the high frequency region.

In the low frequency range in the example of FIG. 6, note number n = 4 and second
For the measurement point of, note number n = 16,
For the third measurement point, note number n = 2
4 is assigned. However, since there is no corresponding measurement point for the note number located in the middle of these, the spectrum intensity cannot be obtained.
It is like a comb with missing teeth. Therefore, when the sampling frequency F = 22.05 kHz and the number of measurement points M = 1024, it is not possible to define the intensities for note numbers n = 5 to 15, 17 to 23. Of course, the number of measurement points M =
If the number of 1024s is further increased, it is possible to eliminate the omission state, but it is inefficient to perform the calculation for such a large number of measurement points.

Conversely, in the high frequency region, a total of 54 measurement points from the 970th measurement point to the 1023th measurement point are assigned to the same note number n = 124. Of course, in this case, there is no problem if the average value of the spectral intensities for all 54 measurement points is defined as the intensity for the note number n = 124, but one note number n = 1
Performing calculations on as many as 54 measurement points to find the intensity value for 24 is inefficient.

After all, in order to efficiently convert to a non-linear code code such as MIDI data, a plurality of M measurement points are discretely defined on the frequency axis in accordance with a required code code, and the What is necessary is just to obtain | require only the spectrum intensity about the frequency component corresponding to M measurement points contained in a signal. In particular, when conversion to MIDI data is performed, a plurality of M measurement points may be discretely defined so as to be equally spaced on a logarithmic scale frequency axis. In other words, a plurality M of measurement points may be discretely defined so that the frequency of each measurement point forms a geometric progression. FIG. 7 is a diagram showing a part of the measurement points defined in this way. The measurement points shown are assigned note numbers n = 60 to 65, and these measurement points are equally spaced on the logarithmic scale frequency axis. Focusing on the specific frequency values 262, 278, 294,... Of each measurement point, they form a geometric progression. When calculating the spectrum intensity by the Fourier transform, if only the spectrum intensity at each of these measurement points is calculated, useless calculation can be omitted.

A specific method for performing such an efficient operation without waste will be described below. First, for convenience of explanation, a procedure for applying a general Fourier transform to the encoding method of the present invention will be described. Here, a case is considered where Fourier transform is performed on an acoustic signal as shown in FIG. 8 and encoding is performed. As described above, in the present invention, a unit section is set on the time axis of an audio signal, and this unit section is represented by P representative code codes. A unit section di shown in FIG. 8 indicates an i-th unit section having a section length L. Here, it is assumed that K samples are included in this unit section di. That is, if the sampling frequency is F and the section length L is expressed in units of time, then K / F = L. The reference time t = t0 is set at the left end position of the sound signal, the left end time of the unit section di is set as the section start time t = ts, and the right end time is set as the section end time t = te. Further, the time from the reference time t0 to the section start time ts is Δth, and the number of samples included in the time of Δth is h.

On the other hand, consider the case where a Fourier spectrum as shown in FIG. 9 is obtained by this Fourier transform.
In this Fourier spectrum, M measurement points are defined on the frequency axis, and the m-th measurement point (m = 0, 1, 1)
The measurement point of (2,..., M-1) corresponds to the frequency f (m), and the spectrum intensity is S (m).
As already mentioned, in the conventional general Fourier transform,
The M measurement points are defined at regular intervals on the frequency axis of a linear scale. The basic principle of Fourier transform is to prepare reference signals consisting of a sine function and a cosine function having various frequencies, obtain a correlation between an acoustic signal to be subjected to Fourier transform and various reference signals, and determine a degree of the correlation. It is to be shown as spectral intensity. For example, in FIG. 9, the value of the spectrum intensity S (m) at the m-th measurement point corresponding to the frequency f (m) is the same as the frequency f (m)
Is a value indicating the degree of correlation with the reference signal having. As a result, FIG. 9 shows the sound signal in the unit section di.
In order to obtain a Fourier spectrum as shown in FIG.
(M-1) is compared with each reference signal, and the degree of each correlation is determined by the spectrum intensities S (0) to S (M-
It can be obtained as 1).

A basic method of calculating such a correlation will be described with reference to FIG. The signal waveform shown in the upper part of FIG. 10 is a waveform of an acoustic signal to be subjected to Fourier transform, and the signal waveform shown in the lower part of FIG. 10 is a reference signal having the m-th frequency f (m) (this example). Is a cosine function). Each signal waveform has a reference time t =
t0 is used as a reference on the time axis, and its amplitude value is −1 to 1
It is standardized to take a value within the range of +1. Now, a value indicating the correlation between the acoustic signal waveform included in the unit section di set on the time axis of the upper graph and the reference signal having the frequency f (m) shown in the lower graph, that is, Spectral intensity S (m) at frequency f (m)
Can be obtained by an equation as shown in FIG. Transformation using this equation is called cosine transform (transformation that does not consider imaginary components in Fourier transform).
Actually, the equation representing the Fourier transform is as shown in FIG. 12, but here, for convenience, the equation representing the cosine transform in FIG. 11 will be described first.

In the equation of FIG. 11, the term A (h + k) on the right side is the k-th (k = 0, 1, 2,..., K−1) in the i-th unit section di of the audio signal. 2 shows the amplitude value of the sample. In the upper graph of FIG. 10, the reference time t
The number of samples included in the time Δth from 0 to the section start time ts is h, and the kth sample counted from the section start time ts is the (h + k) th sample counted from the reference time t0. It turns out that. Therefore, the amplitude value of the (h + k) -th sample counted from the reference time t0 is A (h + k), and if the time from the section start time ts to the sample is Δtk, the sample value from the reference time t0 to the sample Time until (Δt
h + Δtk).

Further, cos (2) on the right side of the equation of FIG.
The term π · f (m) · (Δth + Δtk)) indicates an amplitude value of a position corresponding to the sample of the reference signal (cosine function) of the frequency f (m). That is, in the lower graph of FIG. 10, the time (Δth +
Δtk) (the (h + in the upper graph)
(the same position as the k) -th sample). Term A (h + k) on the right side
And the term cos (2π · f (m) · (Δth + Δtk))
The reason for obtaining the product is that the correlation between the two at a specific position on the time axis is obtained. Unit section d
Since i contains a total of K samples,
Similarly, values indicating the correlation are obtained for all K samples, and the sum of them is calculated. The symbol に in the equation shown in FIG. 11 indicates the sum of k = 0, 1, 2,..., (K−1), and (1 / K) on the right side performs division by the number of samples K. This is for obtaining the average of the correlation. As described above, the amplitude value of the acoustic signal is also
Since the amplitude value of the reference signal is also normalized so as to take a value in the range of −1 to +1, the value of the spectrum intensity S (m) increases as the degree of correlation increases. Therefore, the obtained value of the spectrum intensity S (m) is equal to the frequency f (m) included in the sound signal waveform in the unit section di.
This indicates the intensity of the component.

On the other hand, in the Fourier transform, an equation shown in FIG. 12 is used instead of the equation shown in FIG. This FIG.
The term W (k) on the right-hand side of the equation shown in is a weighting function applied over the section length L, and is the k-th sample ((h + k) -th sample counting from the reference time t0) in the unit section di. Of the amplitude value A (h + k). For this weight function W (k),
As described in §2. On the other hand, exp (−j
As shown in FIG. 12, the term 2πf (m) · (h + k) / F) is cos (2π · f (m) · (h +
k) / F) -jsin (2π · f (m) · (h + k) /
F), the amplitude of the cosine function is plotted on the real axis,
It shows the complex intensity of a trigonometric function with the sine function amplitude value on the imaginary axis. Here, since F is the sampling frequency, (h + k) / F = Δth + Δtk, and the term of the cosine function is the same as the term of the cosine function in the equation shown in FIG. Eventually, the cosine transform equation shown in FIG.
While only the correlation with the cosine function was considered, FIG.
In the Fourier transform equation shown in (1), both the correlation with the cosine function and the correlation with the sine function can be considered, and the influence of the phase shift between the acoustic signal and the reference signal can be eliminated. Further, in the equation shown in FIG. 12, by multiplying the weight function W (k) as described above, the difference between adjacent unit sections can be further emphasized.

By using the equation shown in FIG. 12, the spectrum intensity S (m) for the m-th frequency f (m) can be obtained, so that m = 0, 1,
If the same operation is performed for all of 2,..., (M-1), a Fourier spectrum as shown in FIG. 9 is obtained. However, in the conventional general Fourier transform, as described above, M measurement points are defined at regular intervals on the frequency axis of a linear scale. For example, as shown in FIG. = F · m / M (where m = 0,
1, 2,..., M-1). Specifically, the sampling frequency F = 22.05 kHz, M = 1
In the case of 024, M measurement points having frequencies f (m) as shown in the table of FIG. 13 are defined (actually, according to the sampling theorem,
A correct spectral intensity cannot be obtained for a frequency portion exceeding 1/2 of the sampling frequency F). As described above, when the Fourier spectrum obtained by defining measurement points at regular intervals on the frequency axis of the linear scale is used for conversion to a code system having nonlinear characteristics such as MIDI data, as shown in FIG. As described above, in the low frequency region, note numbers are missing, and in the high frequency region, calculation results are obtained with excessively redundant frequency accuracy, which is extremely inefficient.

Therefore, in this embodiment, for example, FIG.
As shown in the equation, f (m) = 440 · 10 ^{γ (n)} (n = 0, 1, 2,..., 127) such that a total of 128 Number of measurement points are defined. Where n is MIDI
It is the note number of the data, and the following equation holds: γ (n) = (n−69) · log 2/12 Here, “12” corresponds to the number of semitones included in one octave (width at which the frequency doubles). The table of FIG. 14 shows the note number n, γ (n),
The relationship with f (m) is shown. As shown in the figure, in the case of the note number 69 (corresponding to the “La sound (A3 sound)” at the center of the piano keyboard, γ (n) = 0, and the frequency f (m) =
It is 440 Hz. The values of the frequency f (m) form a geometric progression and are equally spaced on a logarithmic scale frequency axis.

After all, instead of using measurement points in the conventional general Fourier transform as shown in the table of FIG. 13, the present invention uses measurement points as shown in the table of FIG. Since the spectrum calculation is performed by using the frequency calculation, the necessary calculation is performed only for the frequency value necessary for the encoding. The final object of the present invention is not to obtain a Fourier spectrum but to encode an audio signal, and the frequency required for encoding (the frequency corresponding to the code to be used) is predetermined. Therefore,
This predetermined frequency component (f in the table of FIG. 14)
The basic technical idea of the present invention is to increase the calculation efficiency by performing only the calculation for obtaining the (frequency component shown in column (m)).

However, when a general Fourier transform is performed, a method of shortening the calculation time by using a fast Fourier transform (FFT) calculation method is adopted. This FF
In the calculation method of T, it is assumed that M = K when M measurement points are defined at equal intervals on the linear frequency axis and the number of samples in a unit section is K. For this reason, the method according to the present invention cannot use the FFT calculation method. However, when the sampling frequency F is set to 22.05 kHz and the number of samples K in the unit section is set to 1024, and the Fourier transform according to the present invention is executed based on the equation of FIG.
The calculation was completed in about twice as long as the time required for the Fourier transform using the calculation method (the note number in the low-frequency region is missing). Therefore, the method according to the present invention is sufficiently useful in practical use.

In the example shown in FIG. 14, a total of 128 measurement points are set in order to cover the range of MIDI data note number n = 0 to 127. However, depending on the MIDI sound source for reproduction, Since not all of these note numbers are required, MIDI
If only the spectral intensity calculation for the required note number is performed according to the sound source, the calculation time can be further reduced. For example, when a piano sound source is used as a MIDI sound source for reproduction, the leftmost keyboard of a general piano has a note number n = 21 and the rightmost keyboard has a note number n = 108. It is only necessary to perform a spectrum intensity calculation in the range of ~ 108. Further, for example, by adding a restriction that encoding is performed using only C major,
Since the note number corresponding to the black key of the piano is not required, the calculation time can be further reduced.

Although the basic method of the audio signal encoding method according to the present invention has been described above, the inventor of the present application has found that even better results can be obtained by making fine improvements to the above method. Was found. That is, instead of finding the correlation between the acoustic signal and the reference signal under the phase relationship shown in FIG. 10, the correlation is found under the phase relationship shown in FIG. The difference between the two is that in the former, the reference point on the time axis of the reference signal is set to the reference time t = t0, whereas in the latter, the reference point on the time axis of the reference signal is the section disclosure time ts Is set to. In other words, in the former, the phase relationship between the acoustic signal and the reference signal is fixed, and when performing the operation for any unit section, the correlation with this fixed phase relationship is obtained. . On the other hand, in the latter, the phase relationship between the sound signal and the reference signal changes every time the calculation is performed for each unit section. For example, in FIG. 15, although the reference signal for the unit section di has the phase as shown, the subsequent unit section d (i +
The reference signal for 1) is obtained by slightly shifting the phase of the illustrated reference signal to the right.

When the correlation is obtained under the phase relationship as shown in FIG. 10, the equation shown in FIG. 12 is used as described above. On the other hand, when obtaining the correlation under the phase relationship as shown in FIG. 15, the equation shown in FIG. 16 may be used. The difference between the two is that the term (h + k) in the exponential function in the former is replaced with h. this is,
This is because, as shown in the lower part of FIG. 15, the reference point on the time axis of the reference signal is the section start time ts, and the time term in the trigonometric function is Δtk.

The inventor of the present application has performed MIDI operation on the same vocal acoustic signal to obtain a spectrum intensity based on the phase relationship shown in FIG.
The coded data was compared with MIDI coded data obtained by performing an intensity calculation under the phase relationship shown in FIG. As a result, it was found that the latter MIDI code data generally represented the original sound signal more accurately. Although we did not perform a detailed analysis of the reason, perhaps moving the reference point on the time axis of the reference signal for each individual unit interval would disperse the probability of showing an incorrect correlation for each unit interval. It is considered that correct encoding is performed as a whole. Of course, when the original sound signal is an accurate sine wave, it is more accurate to perform the intensity calculation under a fixed phase relationship as shown in FIG. As expected, for an irregular signal waveform such as a vocal acoustic signal, performing the intensity calculation under a fluctuating phase relationship as shown in FIG. Will be
It is considered that more suitable encoding is performed.

§4. Code Code Integration Process As described in §2 above, when a partially overlapping section is set, the number of code codes to be created increases considerably. Here, a description will be given of an effective integration process for reducing the number of finally generated code codes as much as possible.

For example, consider a case where a code code represented by musical notes as shown in FIG. In the illustrated example, all code codes are composed of eighth notes. This is because, since the section length L is constant, each generated code code has the same length. However, the note group shown in FIG.
Can be rewritten as shown in That is, when a plurality of notes indicating the same scale are arranged consecutively,
This plurality of notes can be integrated into one note.
In other words, notes spanning multiple unit sections
It is possible to replace notes for each unit section.

In the example shown in FIG. 17, only the notes of the same scale are integrated, but the notes to be integrated are not necessarily limited to the notes of the same scale. It may be integrated. For example, a series of notes that are not one scale apart from each other can be replaced by one note as a unit to be integrated. In this case, for example, it may be replaced by a note with a lower scale in a series of notes. In general, if there are representative code codes similar to each other under predetermined conditions for a plurality of adjacent unit sections, these similar representative code codes are converted to an integrated code code extending over the plurality of unit sections. , The number of notes can be reduced.

In FIG. 17, the concept of code code integration processing has been described for an example of integrating notes, but code codes created by the encoding processing according to the present invention include data (for example, In the case of MIDI data, a velocity is added. Therefore, when code codes are integrated, it is necessary to also integrate data indicating strength. Here, when different intensity data are defined for the code codes to be integrated, for example, the largest intensity data may be determined as the intensity data for the integrated code code. Just M
In the case of IDI data, when combining two code codes, if the strength of the succeeding code code is considerably higher than the strength of the preceding code code, it becomes unnatural to integrate these two code codes. This is a normal MID
The playback sound of the I sound source is composed of the performance sounds of the musical instrument,
This is because the sound intensity generally decreases with time. Therefore, when the strength of the succeeding code code is smaller than the strength of the preceding code code, the unnaturalness does not occur even if the integrated code code is replaced with one integrated code code.
In the opposite case, unnaturalness will occur. Therefore,
If the difference between the strengths of the two code codes is greater than or equal to a predetermined reference and the strength of the succeeding code code is greater than the strength of the preceding code code, a condition is set such that integration is not performed. It is preferable to keep it.

By the way, according to the general MIDI standard, a code code can be divided into a plurality of tracks and recorded. Therefore, the code code created in the present invention is recorded in a plurality of tracks in practical use.
For example, FIG. 3B shows a state where representative code codes (notes in the illustrated example) are recorded in three tracks T1 to T3. In this case, when representative code codes arranged adjacently on the same track satisfy a predetermined similarity condition, a process of integrating the representative code codes arranged adjacently into a single representative code code is performed. Become.

As described above, when code code integration processing is performed, the merit of reducing the number of code codes can be obtained. Therefore, it is desirable to take measures to promote the integration processing as much as possible. Therefore, when distributing and arranging a plurality of representative code codes on a plurality of tracks, the distribution order is adjusted so that the probability that the representative code codes arranged adjacently on the same track satisfy the similar condition becomes high. Preferably, it is adjusted. More specifically, the code codes may be sorted based on the frequency and then stored in each track. For example, as shown in FIG. 3B, when three pieces of code data are distributed to three tracks T1, T2, and T3,
If the distribution method is determined so that the lowest frequency among the three is stored in the track T1, the next lowest frequency is stored in the track T2, and the highest frequency is stored in the track T3, the frequency is completely reduced. It is considered that the probability of occurrence of notes to be integrated is improved as compared to the case where notes are unrelatedly distributed.

Further, as shown in the example of FIG. 18, when the signal section is rearranged, the code code integration processing can be further promoted. For example, assume that five code codes n3, n1, n2, n1, and n3 are arranged on one track as shown in FIG. 18A. Here, the width of each code code represented by a rectangle indicates the signal section length of the code code, and the height indicates the signal strength. Here, signal sections are rearranged in the following four stages.

Step: When the signal strength is below a predetermined level,
Further, a code code whose signal section length is equal to or less than a predetermined length is deleted. Specifically, the third in FIG.
Assuming that the nth code code n2 meets this condition,
By deleting this, the state becomes as shown in FIG.

Step: The signal section length of each code code is extended to the right by a predetermined length within a range not overlapping with adjacent code codes. Specifically, the signal section lengths of the four code codes shown in FIG.
The state is as shown in (c).

Step: If the code codes arranged adjacently satisfy a predetermined similarity condition, they are integrated. This is the above-described integration process. Specifically, the second code code n1 and the third code code n1 in FIG. 18C are integrated, and as shown in FIG. A unified code code n1 having a signal section connecting the two is created.

Step: A code code whose signal section length is equal to or less than a predetermined length is deleted. Here, since the predetermined length serving as the reference is set to be longer than the predetermined length of the step, the first code code n3 shown in FIG. 18 (d) is deleted, and finally, FIG. e) The state shown in (e) is reached.

By performing the above-described signal section rearrangement processing, only two code codes are finally left.

[0071]

As described above, according to the encoding method of the present invention, it is possible to efficiently perform conversion into nonlinear code data such as MIDI data.

[Brief description of the drawings]

FIG. 1 is a diagram showing a basic principle of an audio signal encoding method according to the present invention.

FIG. 2 is a diagram showing a code generated based on the intensity graph shown in FIG. 1 (c).

FIG. 3 is a diagram showing code codes created by performing unit section settings so as to partially overlap on the time axis.

FIG. 4 is a diagram showing a specific example of unit section setting that partially overlaps on a time axis.

FIG. 5 is a graph showing an example of a Fourier spectrum in which a frequency axis is displayed on a linear scale.

FIG. 6 is a graph showing an example of a Fourier spectrum in which a frequency axis is displayed on a logarithmic scale.

FIG. 7 is a graph showing a correspondence relationship between a Fourier spectrum in which a frequency axis is displayed on a logarithmic scale and a note number.

FIG. 8 is a diagram showing various settings for calculating a Fourier spectrum.

FIG. 9 is a graph showing spectrum intensities obtained for M measurement points defined on a frequency axis.

FIG. 10 is a diagram illustrating a first calculation method for obtaining a Fourier spectrum using a Fourier transform.

FIG. 11 is a diagram illustrating a basic expression for obtaining a spectrum intensity S (m) at a predetermined frequency f (m).

FIG. 12 is a diagram illustrating a first equation for determining a spectrum intensity S (m) at a predetermined frequency f (m).

FIG. 13 is a table showing specific values of the frequency f (m) of the measurement points defined at regular intervals on the frequency axis of the linear scale.

FIG. 14 is a table showing specific values of the frequency f (m) of measurement points defined at equal intervals on the frequency axis of a logarithmic scale.

FIG. 15 is a diagram illustrating a second calculation method for obtaining a Fourier spectrum using Fourier transform.

FIG. 16 is a diagram illustrating a second equation for obtaining a spectrum intensity S (m) at a predetermined frequency f (m).

FIG. 17 is a diagram illustrating an example in which the amount of code data is reduced by unitary unit integration processing.

FIG. 18 is a diagram illustrating an example in which the amount of code data is reduced by signal section reorganization processing.

[Explanation of symbols]

A: complex intensity A (h + k): amplitude value of the (h + k) th sample counted from the reference time t0 d1 to d5: unit interval E: effective intensity e (i, j) code code n (i, j) F: Sampling frequency f: Frequency f (m): Frequency of the m-th measurement point h: Number of samples included between the section start time ts of the i-th unit section and the reference time t0 K: Number of samples in one unit section k: Sample number of interest in one unit section L: Section length of unit section ΔL: Offset length M: Number of measurement points m: Number of measurement points (m = 0, 1, 2,...) , M-
1) n, n1, n2, n3... Note number n (i, j)... J-th code code extracted for unit section di S (m)... Spectrum intensity at m-th measurement point T1 to T3. Tracks t1 to t6 time t0 reference time te section end time of i-th unit section di ts section start time of i-th unit section di Δth, Δtk time width

Claims (11)

[Claims] 1. An encoding method for encoding an audio signal given as a time-series intensity signal, wherein a plurality of unit intervals are set on a time axis of the audio signal to be encoded. A code definition step of discretely defining a plurality of M measurement points on a frequency axis and defining a total of M code codes corresponding to the M measurement points, respectively, An intensity calculation step of obtaining a spectrum intensity of a frequency component corresponding to the M measurement points included in the sound signal in the unit section; and an individual unit section based on the spectrum intensity obtained in the intensity calculation step. In each case, P representative code codes representing the unit section are extracted from the M total code codes, and the extracted representative code codes and their spectrum intensities are extracted. I, the coding method of the audio signal, characterized in that it has a, a coding step of representing the acoustic signals of the individual unit sections. 2. The encoding method according to claim 1, wherein a plurality of M measurement points are discretely defined in the code definition step so as to be equally spaced on a frequency axis of a logarithmic scale. A coding method of an acoustic signal, characterized in that: 3. The encoding method according to claim 2, wherein a plurality of M code codes are defined as MIDI code in the code definition step.
Using the note number used in the data, in the encoding stage, the audio signal of each unit section is extracted as a representative code, the note number extracted as a representative code, the velocity determined based on the spectrum intensity, and the unit A method for encoding an audio signal, characterized in that the audio signal is represented by MIDI-format code data including data indicating the delta time determined based on the length of the section. 4. The encoding method according to claim 1, wherein a spectrum intensity S (m) at an m-th measurement point corresponding to the frequency f (m) is calculated in the intensity calculation step. In this case, an audio signal encoding method is characterized in that an arithmetic operation for obtaining a correlation with M sine functions and cosine functions having frequencies corresponding to the respective measurement points is performed. 5. The encoding method according to claim 1, wherein a weighting function defining weighting over the section length of the unit section is prepared in the intensity calculation step, and the sound signal in the unit section is provided. A sound signal encoding method, wherein a spectrum intensity is obtained by multiplying the weight function. 6. The encoding method according to claim 1, wherein in the section setting step, setting is performed such that adjacent unit sections partially overlap on a time axis. A method for encoding an audio signal. 7. The encoding method according to claim 1, wherein the audio signal to be encoded is sampled at a predetermined sampling frequency F, and the amplitude value of the x-th sample is represented by A.
(X) is taken as sound data, and each unit section is set for the taken sound data. In the intensity calculation stage, for a unit section starting from the h-th sample and including a total of K samples, Frequency f
When calculating the spectrum intensity S (m) at the m-th measurement point corresponding to (m), a predetermined weight function W
Using (k), S (m) = (1 / K) Σ _{k = 0 to (K−1)} (W
(K) · A (h + k) · exp (−j2πf (m) ·
(H + k) / F)) An audio signal encoding method characterized by using the following formula: 8. The encoding method according to claim 1, wherein the audio signal to be encoded is sampled at a predetermined sampling frequency F, and the amplitude value of the x-th sample is set to A.
(X) is taken as sound data, and each unit section is set for the taken sound data. In the intensity calculation stage, for a unit section starting from the h-th sample and including a total of K samples, Frequency f
When calculating the spectrum intensity S (m) at the m-th measurement point corresponding to (m), a predetermined weight function W
Using (k), S (m) = (1 / K) Σ _{k = 0 to (K−1)} (W
(K) · A (h + k) · exp (−j2πf (m) · k
/ F)) An audio signal encoding method characterized by using the following formula: 9. The encoding method according to claim 1, wherein in the encoding step, a plurality of P representative code codes extracted for each unit section are distributed and arranged on a plurality of tracks. ,
When representative code codes arranged adjacently on the same track satisfy a predetermined similarity condition, a process of integrating the representative code codes arranged adjacently into a single representative code code is performed. Encoding method of the audio signal to be encoded. 10. The encoding method according to claim 9, wherein when a plurality of P representative code codes are distributed and arranged on a plurality of tracks, the representative code codes arranged adjacently on the same track are different. A method for encoding an audio signal, comprising adjusting an order of distribution so as to increase the probability of satisfying a similar condition. 11. A computer-readable recording medium on which a program for encoding an audio signal for executing the encoding method according to claim 1 is recorded.