JP5732910B2 - Method and apparatus for encoding acoustic signal - Google Patents

Description

  The present invention relates to an audio signal encoding technique, and more particularly to an encoding technique suitable for encoding into code data of MIDI format or the like.

  Conventionally, in order to be able to reproduce an acoustic signal using a MIDI sound source, the acoustic signal is converted into code data such as a MIDI code (see Patent Documents 1 to 3). Since a MIDI sound source is reproduced at a limited frequency such as 32 chords, it is necessary to select and encode a limited number of frequencies when encoding. The applicant has also proposed a technique for selecting and encoding a limited number of frequencies from an acoustic signal (see Patent Document 1).

JP 2002-41037 A Japanese Patent No. 4061070 Japanese Patent No. 4156268

Masato Akagi: “Hearing filter and its model”, IEICE Journal 77 (9), 948-956, 1994-09-25.

  However, in the invention described in Patent Document 1, signal components in the vicinity of a frequency having a high intensity selected previously are uniformly attenuated so that adjacent frequency components are not selected as much as possible. For this reason, the component that is originally necessary for expressing the consonant in the vocal reproduction is also deleted, and the reproducibility is lowered.

  On the other hand, human hearing is frequency-discriminated by 24 auditory nervous systems, and can be theoretically approximated by 24 bandpass filters. It is known that the frequency component is less than (see Non-Patent Document 1). That is, the frequency component included in each band filter can recognize only a single frequency component at the same time, and a plurality of frequency components included in the same band filter are recognized with a time difference (as a beat sound or a beat, However, a minute difference in frequency of about ¼ semitone within the same band is recognized).

Therefore, the present invention provides an audio signal encoding method and apparatus capable of faithfully reproducing a consonant component using a sound source reproduced at a limited number of frequencies by utilizing human auditory characteristics. The issue is to provide.

  In order to solve the above-described problem, in the first aspect of the present invention, when encoding an acoustic signal given as a sample sequence of J time series digitized at a predetermined sampling frequency, for the sample sequence, A unit interval composed of a predetermined number T (T <J) samples is set by overlapping a predetermined number W (W <T) samples in the time axis direction with an adjacent unit interval, and each unit interval For each unit section, by performing frequency conversion for at least N types of frequencies f (n) to be analyzed, spectrum intensities corresponding to the N types of frequencies are calculated for each unit section, and the N types Are divided into a predetermined number of frequency groups so that they do not overlap with each other, and the maximum of the spectrum intensities of the frequencies included in each frequency group for each unit section. Is corrected so as to be attenuated by a predetermined ratio to the spectrum intensity other than the frequency taking the frequency, and the corrected spectrum intensity is created, and the time difference between the start time of the unit interval and the start time of the immediately following unit interval is determined. A code code of a predetermined format is generated based on the corrected spectrum intensity.

  According to the first aspect of the present invention, frequency conversion is performed on each unit section to calculate spectrum intensities corresponding to N types of frequencies, and the N types of frequencies f (n) are determined so as not to overlap each other. The frequency spectrum is divided into a number of frequency groups, and correction is made so as to attenuate the spectrum intensity other than the frequency having the maximum value among the spectrum intensities included in each frequency group by a predetermined ratio, and the obtained analysis results are used as a basis. Since the code code is generated, the consonant component of the acoustic signal can be faithfully reproduced using a sound source (for example, a MIDI sound source) reproduced at a limited frequency such as 32 chords.

  Further, in the second aspect of the present invention, in the first aspect of the present invention, the predetermined number of frequency groups are defined based on the characteristics of the human auditory system and the value of n is defined as a standard note number. , 45, 57, 64, 69, 72, 76, 79, 82, 85, 88, 91, 93, 96, 98, 101, 104, 106, 109, 113, 116, 119, 123, 127, 128 It is characterized by being set by 24.

  According to a third aspect of the present invention, in the first aspect of the present invention, the predetermined number of frequency groups have a frequency f (n) value of 20, 100, 200, 300, based on characteristics of the human auditory system. 400,510,630,770,920,1080,1270,1480,1720,2000,2320,2700,3150,3700,4400,5300,6400,7700,9500,12000,15500Hz It is characterized by that.

  According to the second and third aspects of the present invention, the correction is performed by classifying into 24 frequency groups based on the characteristics of the human auditory system, so that the optimum frequency component suitable for the human auditory characteristics is obtained. It is possible to perform encoding. In particular, according to the second aspect of the present invention, since the frequency group is classified by the note number n corresponding to the MIDI standard of the reproduction sound source, encoding suitable for the reproduction MIDI sound source can be performed.

  Also, in the fourth aspect of the present invention, in any one of the first to third aspects of the present invention, the calculation of the spectral intensity is performed for each unit section at least N types of frequencies f ( n), a first spectral intensity E1 (p, n) corresponding to the N kinds of frequencies is calculated for the unit interval p by performing frequency conversion, and immediately before the unit interval p. Only when the evaluation value based on the change for each corresponding frequency with the first spectrum intensity E1 (p-1, n) in the unit interval p-1 located is larger than the predetermined threshold value, the unit interval p Is selected as the q (q ≦ p) -th selection unit interval q, and at least N types of frequencies f (n) to be analyzed are more accurate than the frequency conversion in the first spectrum calculation stage. frequency By calculating the second spectral intensity E2 (q, n) corresponding to the N kinds of frequencies by performing conversion, the second spectral intensity E2 (q, n) is calculated as the spectral intensity. The spectrum is corrected so as to be attenuated by a predetermined ratio with respect to the second spectrum intensity E2 (q, n), and the corrected spectrum intensity E ′ (q, n) is used as the corrected spectrum intensity. ) Is created.

  According to the fourth aspect of the present invention, a simple first frequency conversion is performed for each set unit section, and the selection is performed when the intensity is larger than a predetermined reference compared to the immediately preceding unit section. Since it was selected as a unit interval, and the second frequency conversion with higher accuracy was performed on the selected unit interval, and a code code was generated based on the obtained analysis result, the entire acoustic signal was fixed at a fixed interval. Thus, only the characteristic part is encoded while analyzing the information over time, so that it is possible to more appropriately analyze the acoustic signal including chords and the frequency change of the audio signal.

  In the fifth aspect of the present invention, in the fourth aspect of the present invention, the first spectrum calculation and the second spectrum calculation do not exceed frequencies adjacent to each of the N types of frequencies f (n). A predetermined M types of sub-frequency f (n, m) are set in a range, and the M types of sub-frequency f1 (p, n) and second spectrum strength E2 (q, n) are set as the first spectrum strength E1 (p, n). An intensity value corresponding to a sub-frequency indicating the greatest intensity among the frequencies is calculated.

  According to the fifth aspect of the present invention, by setting the interval of the frequency to be analyzed finely, more detailed frequency analysis can be performed and an appropriate frequency component can be specified.

  In addition, in the sixth aspect of the present invention, when the code code is generated in the fourth or fifth aspect of the present invention, the selection unit section with respect to two adjacent selection unit sections q and selection unit section q + 1. When q is the p-th unit interval p, it is defined as P (q) = p and corresponds to the frequencies f (n), f (n−1), and f (n + 1) in the selected unit interval q + 1. Any one of the first spectral intensities E1 (P (q + 1), n), E1 (P (q + 1), n-1), E1 (P (q + 1), n + 1) and a position immediately before the selected unit interval q + 1 A difference from the first spectral intensity E1 (P (q + 1) -1, n) corresponding to the frequency f (n) in the unit interval P (q + 1) -1 to be less than a predetermined threshold Ldif, and First spectral intensity E1 (P (q + 1), n) Any of E1 (P (q + 1), n−1), E1 (P (q + 1), n + 1) and the first spectral intensity E1 (P (q + 1) −1, n) is greater than a predetermined threshold value Lmin. If it is larger, the selection unit interval q and the selection unit interval q + 1 are connected, and the start time of the selection unit interval q defined in the selection unit interval q and the selection unit interval q + 1 It is characterized in that the time difference from the head time of the immediately subsequent selection unit section q + 2 is used.

  According to the sixth aspect of the present invention, when the code code is generated, the difference in intensity between the subsequent selected unit section and the immediately preceding unit section of the two adjacent selected unit sections is less than the predetermined threshold value. When the intensity of the subsequent selected unit section and the intensity of the previous unit section are both greater than a predetermined threshold, the two adjacent selected unit sections are connected so that the time is not selected. Therefore, the closest unit section is taken into consideration when determining the connection, and the sound components can be appropriately connected.

  In the seventh aspect of the present invention, in the sixth aspect of the present invention, the first spectrum calculation and the second spectrum calculation do not exceed frequencies adjacent to each of the N types of frequencies f (n). A predetermined M types of sub-frequency f (n, m) are set in a range, and the M types of sub-frequency f1 (p, n) and second spectrum strength E2 (q, n) are set as the first spectrum strength E1 (p, n). An intensity value corresponding to the sub-frequency f1 (n, mmax1) and f2 (n, mmax2) indicating the highest intensity among the frequencies is calculated, and the first spectral intensity is generated when the code code is generated. E1 (P (q + 1), n), E1 (P (q + 1), n-1), at least one of the sub-frequency numbers mmax1, mmax2, mmax3 for determining E1 (P (q + 1), n + 1); Only when the condition that the difference from the sub-frequency number mmax0 for determining the first spectral intensity E1 (P (q + 1) -1, n) is less than a predetermined threshold value Ndif is satisfied, the selection unit interval q And the selected unit interval q + 1 are connected.

  According to the seventh aspect of the present invention, it is possible to perform more detailed frequency analysis by finely setting the frequency interval to be analyzed, and further, as a sound component connection condition, the subsequent selection unit interval and the immediately preceding selection unit interval Since it is added that the difference from the sub-frequency of the unit section is less than the threshold value, it is possible to connect sound components based on a more accurate analysis result.

  Further, in the eighth aspect of the present invention, in the seventh aspect of the present invention, when the selection unit section q is already connected to another selection unit section, the head of the selection unit section q is connected. The selected unit interval is qo, and the encoding step includes the first spectral intensities E1 (P (q + 1), n), E1 (P (q + 1), n−1), E1 (P (q + 1), n + 1). A difference between at least one of the sub-frequency numbers mmax1, mmax2, and mmax3 for determining the first spectral intensity E1 (P (qo), n) and the sub-frequency number mmmax for determining the first spectrum intensity is a predetermined threshold value Nadif The selection unit interval q and the selection unit interval q + 1 are connected only when the condition of less than is further satisfied.

  According to the eighth aspect of the present invention, as the sound component connection condition, when the preceding selection unit interval is already connected to another selection unit interval, the subsequent selection unit interval and the immediately preceding selection unit Since the difference between the sub-frequency of the first selection unit section to which the section is connected is less than the threshold value, the subsequent selection unit section belonging to different sound components whose sub-frequency changes slowly is mistakenly It is possible to prevent the connection and to achieve a more accurate connection of sound components.

  Also, in the ninth aspect of the present invention, in any one of the fourth to eighth aspects of the present invention, the first spectrum calculation is performed using N types of element signals to be constituent elements of the section signal of the unit section. , Each corresponding to an integral multiple of the period of the frequency f (n), prepared as T (n) samples closest to the T, and element signals corresponding to the N frequencies f (n); A first spectrum intensity E1 (p, n) is calculated by performing a correlation operation with a section signal composed of T (n) samples of the corresponding unit section p, and the second spectrum. Is calculated by correlating the prepared element signal corresponding to each of the N frequencies f (n) and the section signal composed of T (n) samples of the corresponding selection unit section q. Calculate the frequency f () with the highest correlation value element) corresponding to max) is selected as a harmonic signal, and T (nmax) samples given by the product of the selected harmonic signal and the correlation value obtained for the harmonic signal are used as the inclusion signal, By subtracting the signal from the interval signal, a differential signal composed of T (nmax) samples is obtained by calculation, and T (n) samples updated to reflect the T (nmax) samples are obtained. As the new interval signal, the selection of the harmonic signal and the calculation of the difference signal are performed to repeatedly obtain the new inclusion signal and the difference signal, thereby obtaining N inclusion signals, and the correlation of the obtained inclusion signals. A second spectrum intensity E2 (q, n) corresponding to the N types of frequencies is calculated based on the value.

  According to the ninth aspect of the present invention, the first spectrum calculation for all unit sections is performed by simple discrete Fourier transform, and the second spectrum calculation for the selected unit section is performed by high-precision generalized harmonic analysis. As a result, it is possible to efficiently obtain information on the selected unit section with high accuracy while referring to the analysis results of all the unit sections.

  In the tenth aspect of the present invention, in the ninth aspect of the present invention, the calculation of the correlation in the first spectrum calculation corresponds to each frequency f (n) in the unit interval p-1 located immediately before. A correlation calculation corresponding to the first W sample in the unit interval p-1 is performed on the previous correlation calculation result, and a correlation value for each frequency is subtracted from the previous correlation calculation result, and T (n in the unit interval p ) Correlation calculation corresponding to each frequency f (n) in the unit interval p by performing correlation calculation corresponding to the last W sample in the sample and adding the correlation value for each frequency to the previous correlation calculation result This is performed by obtaining a result and calculating the first spectral intensity E1 (p, n) based on the correlation calculation result.

  According to the tenth aspect of the present invention, when performing a simple correlation calculation for each unit section in the first spectrum calculation, the correlation calculation result performed for the previous unit section is used, and the previous correlation calculation result is calculated. Since the head part is removed and the correlation calculation is performed on the tail of the unit interval, and the result is added to the previous correlation calculation result, most of the correlation calculation result of the previous unit interval is used. It is possible to speed up the arithmetic processing for all unit sections.

  According to the present invention, there is an effect that it is possible to provide a method and an apparatus for encoding an acoustic signal that can more appropriately analyze a frequency change of a chord signal or an audio signal.

It is a hardware block diagram of the encoding apparatus of the acoustic signal in this embodiment. It is a flowchart which shows the outline | summary of the encoding method of the acoustic signal which concerns on this invention. It is a figure which shows the concept of the expansion of a time-axis direction, the increase in a frequency, and reduction | decrease of time information. It is a figure which shows the concept of the area setting of this embodiment compared with the prior art. It is a figure which shows the relationship of the logical / physical range of the analysis frequency in this embodiment. It is a figure which shows the relationship between the unit area and analysis range in this embodiment. It is a figure which shows the mode of the analysis process of the unit area in this embodiment. It is a figure which shows the correspondence of the sample row | line in the unit area extracted from the acoustic signal after a time-axis expansion process, and a harmonic signal. It is a figure explaining the time-axis extension of the analysis frame in case unit interval length T is more than 1/2 period of a harmonic signal and less than 3/4 period. It is a figure explaining the time-axis extension of an analysis frame in case unit interval length T is 1/4 period or more and less than 1/2 period of a harmonic signal. It is a figure which shows the frequency reference band filter produced based on the nonpatent literature 1. FIG. It is a figure which shows the band filter of the MIDI note number reference | standard which converted the unit of the band filter shown in FIG. It is a flowchart which shows the detail of the auditory filter correction | amendment in S8. It is a figure which shows the relationship between the selection unit area used as the object of a connection judgment, and a unit area.

DESCRIPTION OF EMBODIMENTS Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the drawings.
FIG. 1 is a hardware configuration diagram of an audio signal encoding device according to an embodiment of the present invention. The audio signal encoding apparatus can be realized by a general-purpose computer. As shown in FIG. 1, a CPU 1 (CPU: Central Processing Unit) and a computer main memory RAM 2 (RAM: Random Access Memory) A large-capacity data storage device 3 (for example, a hard disk) for storing data, a program storage device 4 (for example, a hard disk) for storing a program executed by the CPU, and a key input I such as a keyboard and a mouse / F5, a data input / output interface 6 for data communication with an external device (data storage medium), and a display output interface 7 for sending information to a display device (display), each via a bus It is connected.

  The program storage device 4 of the acoustic signal encoding device is mounted with a dedicated program for operating the CPU 1 and causing the computer to function as the acoustic signal encoding device. By executing the dedicated program, the CPU 1 can select the section setting means, the first spectrum calculation means (including element signal preparation means and correlation calculation means), and the second spectrum calculation means (element signal preparation means, harmonic signal selection). Functions as a spectrum correction means and an encoding means. The data storage device 3 stores various data necessary for processing.

  FIG. 2 is a flowchart showing an outline of an audio signal encoding method according to this embodiment. The audio signal encoding method according to the present embodiment is performed by a computer executing a program that records the detailed procedure of each step (each stage) shown in FIG. As a computer, a CPU and memory for performing arithmetic processing, a storage device such as a hard disk for storing programs and data, a data input device for inputting data such as acoustic signals, a keyboard for inputting instructions, a mouse, etc. A general-purpose computer including a device and a display device such as a liquid crystal display that displays necessary information on a screen can be used. Also, the audio signal encoding apparatus according to the present embodiment is realized by a general-purpose computer in which a program recording the detailed procedure of each step (each stage) shown in FIG. 2 is incorporated.

  First, a computer (encoding device) reads a digital audio signal to be processed from a data input device. The digital audio signal is obtained by sampling an analog audio signal at a predetermined sampling frequency and the number of quantization bits. In the present embodiment, the sampling is performed with a sampling frequency of 44.1 kHz and a quantization bit number of 16 bits as an example. I will explain. When sampling is performed at a sampling frequency of 44.1 kHz, the digital acoustic signal is configured as a sample row (sample array: intensity array) having 44100 samples (intensity values) per second.

  After reading the digital sound signal, the computer (encoding device) expands the digital sound signal by a predetermined magnification K (K is an integer) in the time axis direction (S1). Specifically, the number of samples constituting the digital acoustic signal is multiplied by K. Then, every K samples having the same value as the original sample are arranged, and (K-1) sample values between them are linearly interpolated using the values of the original sample located on both sides. give. Assuming that the sample value for each sample j (j = 0... J-1) of the original sound signal is x (j), the computer expands by executing the processing according to the following [Equation 1]. A sample value x ′ (j · K + k) for each sample j · K + k (0 ≦ k ≦ K−1) of the subsequent acoustic signal is calculated. In the following [Equation 1], w is a real value taking a value of 0 ≦ w ≦ 1 given by k / K.

[Formula 1]
x ′ (j · K + k) = (1−w) · x (j) + w · x (j + 1)

  As a result of the processing in S1, J samples constituting the digital audio signal are expanded to J × K. FIG. 3A shows a change in waveform due to the enlargement process in S1. The waveform in FIG. 3A is obtained by connecting sampled values plotted with line segments, but is expressed in a curved line due to the large number of samples. By executing the processing according to the above [Equation 1], the waveform shown on the left side changes to the waveform shown on the right side. In the example of FIG. 3, the case of K = 2 is shown for convenience of explanation. In this way, by performing analysis by expanding the acoustic signal in the time axis direction, the time resolution in the analysis is improved while maintaining the same frequency analysis accuracy as when performing the analysis directly on the original acoustic signal. Thus, it is possible to extract frequency fluctuations with high accuracy.

  The computer (encoding device) performs frequency analysis on a predetermined section in subsequent S2 to S5. In the present embodiment, the section to be subjected to frequency analysis is set by selecting a unit section satisfying a predetermined condition as the selected unit section after setting the unit section. The concept of setting the section to be analyzed in the present embodiment will be described in comparison with the prior art represented by Patent Document 2. FIG. 4 is a diagram showing a concept of section setting of the present embodiment compared with the prior art. As will be described later, both the conventional case and the present embodiment are the same in that a predetermined number of samples obtained by sampling are set as unit sections and frequency analysis is performed for each unit section. As shown in FIG. 4A, in the prior art disclosed in Patent Document 2, a zero crossing point is specified, and the setting position of the unit section is made variable using the zero crossing point. In FIG. 4A, three unit sections are set as an example, but the intervals between these start positions (left ends) are not uniform. Then, both discrete Fourier transform and generalized harmonic analysis are executed for each set unit interval to obtain an analysis result.

  On the other hand, as shown in FIG. 4B, in this embodiment, unit intervals are set at fixed intervals, and discrete Fourier transform is performed on each unit interval to obtain an analysis result. Then, the analysis result is compared with the immediately preceding unit section, and when a predetermined condition is satisfied, it is selected as the selected unit section. In the example of FIG. 4B, unit sections 1, 5, and 6 are selected as selection unit sections 1, 2, and 3, respectively. Then, a generalized harmonic analysis is performed on the selected unit section to obtain an analysis result.

  Specifically, first, the computer (encoding device) sets a unit interval on the sample expanded in the time axis direction (S2). The length of the unit interval (number of samples T) is set in relation to the sampling frequency. However, if the sampling frequency is 44.1 kHz, 4096 samples or more are required to faithfully analyze the low frequency region. . Therefore, in this embodiment, the unit interval is set as the number of samples T per unit interval T = 4096.

  Setting of the unit section is performed by sequentially extracting samples from the head of the digital sound signal as disclosed in Patent Documents 1 and 3. The unit interval is set so as not to leak all samples, and is preferably set so that the samples overlap in continuous unit intervals. In the present invention, the head interval (referred to as shift width) of each unit section is set as a fixed value. That is, the number of samples to be overlapped is set to be constant. In the present embodiment, the shift width W = 64 is a fixed value. Accordingly, when T = 4096, the first unit interval is j = 0 to 4095, the second unit interval is j = 64 to 4159, the third unit interval is j = 128 to 4223, and so on. It will be set with duplicate samples. The value x (j) of each sample is expressed as a value x (p, i) (0 ≦ i ≦ T−1) for each unit interval p (p is an integer of 0 or more).

Next, discrete Fourier transform, which is the first frequency analysis, is executed for each set unit section, and the spectrum of each unit section is calculated (S3). As disclosed in Patent Documents 1 to 3, the spectrum of each unit section is calculated with 128 analysis frequencies f (n) = 440 · 2 ^{(n−69) / 12} corresponding to MIDI note number n. This is done by extracting 128 components by discrete Fourier transform based on the harmonic signal (harmonic function). “128 types” and “128 pieces” are examples. For example, in the case of the MIDI standard, it corresponds to the range of note number n = 0 to 127, but the standard range for reproducing a grand piano is note number n. The range is from 21 to 108. Therefore, in this case, 88 components are extracted using 88 types of analysis frequencies. Further, as described above, when the frequency analysis is performed by enlarging the acoustic signal by a predetermined magnification K in the time axis direction, if the time axis expansion process described later is omitted, the analysis cannot be performed at the analysis frequency located in the bass portion. Therefore, the types of analysis frequencies are further reduced. For example, when K = 4, the analysis frequency is limited to the range of note numbers n = 24 to 127 (104 types) in the MIDI standard, and note numbers n = 24 to 108 (64 types) in the standard range of the grand piano. ).

In the present embodiment, as the acoustic signal is expanded K times in the time axis direction, the upper and lower limits of n are respectively moved downward by α. α is an integer defined by α = 12 · log ₂ K (for example, α = 24 when K = 4). Therefore, in Patent Documents 1 to 3, 0 ≦ n ≦ 127, but in this embodiment, −α ≦ n ≦ 127−α. Thereby, the frequency of each harmonic signal is set to 1 / K times. Here, the relationship of the logical / physical range of the analysis frequency in this embodiment is shown in FIG. As shown in FIG. 5, the standard sound range of the grand piano is in a range of n = 21 to 108. Therefore, when performing a normal analysis, it is performed in a range of n = 21 to 108. However, in the present invention, the time axis is expanded to shift the frequency to the bass side and perform analysis processing. For note number n = 21 or less, since one cycle of the corresponding harmonic signal is longer than the unit interval, a long cycle analysis is performed by extending the time axis (described later). As a result, frequency components are obtained for n = −3 to 84, but by performing correction processing finally, range frequency components of n = 21 to 108 are obtained.

When the analysis frequency is set in correspondence with the note number n, the frequency interval between the note numbers becomes wider as the frequency becomes higher. In particular, when n exceeds 60, the analysis accuracy decreases. Therefore, in the present embodiment, as disclosed in Patent Document 3, 128M element signals f (n, m) = 440 · 2 ⁽ⁿ⁻ ) where the note numbers are divided into M differential sounds (sub-frequency). ^{69 + m / M) / 12} , and 128M components are extracted. Unless special encoding such as addition of a pitch bend code is performed in the code code generation process of S10 described later, information on M differential sounds at each note number is unnecessary, so the maximum value of the components of M differential sounds Is represented as a component in the note number, and as a result, 128 components are extracted.

  As a specific processing procedure by the computer (encoding device), for each unit section p, first, an array of intensity values E1 (p, n) (−α ≦ n ≦ 127−α) corresponding to the note number and sub The frequency array S (p, n) is set and all initial values are set to 0. Subsequently, the processing according to the following [Formula 2] is executed for -α ≦ n ≦ 127-α and 0 ≦ m ≦ M−1 to maximize E1 (p, n, m) (nmax , Mmax). However, if the bass portion is not required due to the relationship with the processing load, the processing load can be reduced by omitting the processing for n <0.

[Formula 2]
A (p, n, m) = (1 / T (n)). Σi _{= 0, T (n) −1} x (p, i) sin (2πf (n, m) (i + pW) / fs)
B (p, n, m) = (1 / T (n)). Σi _{= 0, T (n) −1} x (p, i) cos (2πf (n, m) (i + pW) / fs)
E1 (p, n, m) = {A (p, n, m)} ² + {B (p, n, m)} ²

  In the above [Equation 2], T (n) is the analysis frame length. When one period of the harmonic signal (harmonic function) is equal to or shorter than the unit section length T, the maximum period of the harmonic signal is within a range not exceeding the unit section length T. Set to be an integer multiple of. However, in the present invention, the time axis is expanded to shift the frequency to the bass side and the analysis process is performed, so that one period of the harmonic signal (harmonic function) exceeds the unit interval length T. Specifically, when one period of the harmonic signal is larger than the unit section length T, it is given by T (n) = g × fs / f (n, m), and the value of x (i) at T <T (n) Is set based on the time axis extension process described later. Note that g is an integer value of 1 or more, and fs is a sampling frequency (for example, 44.1 kHz).

  The processing according to the above [Equation 2] is executed for each unit section, and A (p, n, m), B (p, n, m), E1 (p, n, m) can be obtained. It is. Here, the relationship between the unit interval and the analysis range in the present embodiment is shown in FIG. In FIG. 6, the waveform at the upper end schematically shows the original sound signal, and the waveform at the lower end schematically shows the harmonic function. In the example of FIG. 6, only the target unit section that is the target unit section and the immediately preceding unit section that is the immediately preceding unit section are shown, but each correlation calculation range has a rectangular horizontal length. Become. In this embodiment, since the correlation calculation range T is 4096 samples and the shift width W is 64 samples, the overlapping portion is very large. Therefore, in the present embodiment, the efficiency of the analysis process is improved by using the analysis result in the immediately preceding unit section for the overlapping portion.

  FIG. 7 shows a state of the unit interval analysis processing in the present embodiment. As shown in FIG. 7, when obtaining the analysis result in the target unit section, the overlapping part of the immediately preceding unit section is used. Specifically, the head part of the previous unit section that does not overlap with the target unit section is deleted, and only the tail part of the target unit section that does not overlap with the previous unit section is added by performing correlation calculation. Therefore, the correlation calculation is performed only for the head unit interval (p = 0) over the entire unit interval.

  When p ≧ 1, that is, when processing for the second and subsequent unit intervals p, A (p−1, n, m) and B (p−1, n, m) for the immediately preceding unit interval (p−1). m) has already been calculated. In the present embodiment, by using A (p−1, n, m) and B (p−1, n, m), the processing according to the following [Equation 3] is executed, so that the unit interval p is obtained. A (p, n, m) and B (p, n, m) are calculated.

[Formula 3]
A (p, n, m) = A (p-1, n, m) − (1 / T (n)) · Σ _{i = 0, W−1} x (p−1, i) sin (2πf (n , M) (i + (p-1) W) / fs) + (1 / T (n)). Σi _{= T (n) -W, T (n) -1} x (p, i) sin (2πf (N, m) (i + pW) / fs)
B (p, n, m) = B (p-1, n, m) − (1 / T (n)) · Σ _{i = 0, W−1} x (p−1, i) cos (2πf (n , M) (i + (p-1) W) / fs) + (1 / T (n)). Σi _{= T (n) -W, T (n) -1} x (p, i) cos (2πf (N, m) (i + pW) / fs)
E1 (p, n, m) = {A (p, n, m)} ² + {B (p, n, m)} ²

  Subsequently, for each note number n, (p, n, mmax) that maximizes E (p, n, m) in the range of 0 ≦ m ≦ M−1 is obtained, and E1 (p, n) = E1 Processing is performed such that (p, n, mmax) and S (p, n) = mmax. Then, the calculated E1 (p, n) and S (p, n) are temporarily stored in the memory. E1 (p, n) and S (p, n) temporarily stored in the memory are used in a single-tone component connection process to be described later.

  Next, a change between the spectral intensity E1 (p, n) calculated in the unit interval p and the spectral intensity E1 (p-1, n) calculated in the immediately preceding interval (p-1) is evaluated (S4). ). Specifically, first, a change evaluation value dE (p−1, p) with the immediately preceding section (p−1) of the unit section p is calculated by executing processing according to the following [Formula 4]. .

[Formula 4]
dE (p−1, p) = (100 / N) · Σ _{n = 0, N−1} {| E1 (p, n) −E1 (p−1, n) | / (E1 (p, n) + E1 (p-1, n))}

  If the obtained change evaluation value dE (p−1, p) is less than a predetermined threshold value (for example, 40), the process returns to S2 as p ← p + 1, and the next unit interval p is set. .

  On the other hand, when the obtained change evaluation value dE (p−1, p) is equal to or greater than a predetermined threshold value, the unit interval p is selected as the selected unit interval q, and the generalized harmony is selected for the selected unit interval q. Analysis is performed (S5). The value of q is 0 for the first selected unit interval, and thereafter, a value obtained by adding 1 each time it is selected is given. Specifically, first, E1 (p, nmax) that maximizes E1 (p, n) set in S3 is obtained. That is, among all n of −α ≦ n ≦ 127−α, the value of n that maximizes E1 (p, n) is obtained as nmax, and E1 (p, n) at that time is determined as E1 (p, n). nmax). This is performed by executing the process of [Formula 2] for all n and selecting the largest of the calculated n E1 (p, n). Further, using the obtained nmax, mmax = S (p, nmax) is set.

  Then, A (p, nmax, mmax) and B (p, nmax, mmax) are calculated by executing processing according to the following [Equation 5] using the obtained nmax and mmax. When executing the processing according to [Formula 5], first, assuming that the unit interval p is the q-th selected unit interval q, P (q) = p is set, and the selected unit interval q , The correlation intensity array E2 (q, n) for the note number is defined, and the initial values are all set to values less than 0 (eg, −1).

[Formula 5]
A (p, nmax, mmax) = (1 / T (nmax)) · Σi _{= 0, T (nmax) −1} x (p, i) · sin (2πf (nmax, mmax) i / fs)
B (p, nmax, mmax) = (1 / T (nmax)) · Σi _{= 0, T (nmax) −1} x (p, i) cos (2πf (nmax, mmax) i / fs)
E2 (q, nmax) = {A (p, nmax, mmax)} ² + {B (p, nmax, mmax)} ²

  Then, by using the calculated A (p, nmax, mmax) and B (p, nmax, mmax), a process in accordance with the following [Equation 6] is performed, whereby a sample (p , I), the value x (p, i) is updated over 0 ≦ i ≦ T (nmax) −1.

[Formula 6]
x (p, i) ← x (p, i) −A (p, nmax, mmax) · sin (2πf (nmax, mmax) i / fs) −B (p, nmax, mmax) · cos (2πf (nmax , Mmax) i / fs)

  The process of [Formula 6] is a process of removing the contained signal from the original acoustic signal. A new E2 (q, nmax) that maximizes E2 (q, n) is obtained for n other than the processed nmax value for the acoustic signal after removing the contained components, and the new nmax is obtained. Is used to execute processing according to [Equation 5] and [Equation 6]. As a result, the contained signal is further removed from the acoustic signal. The computer (encoding device) executes such a process for all 128 n to obtain E2 (q, n).

  In the present embodiment, in order to reduce the processing load, the value of M is variably set based on the note number, for example, the frequency interval to be analyzed is about 100 Hz. And note number 60 and below are not divided and M = 1. Further, although the accuracy is slightly reduced, S (p, n) is determined by the processing of [Equation 2] for determining the correlation strength array E1 (p, n), and the correlation strength array E2 (q, n) is determined. The processing of [Formula 5] for this purpose may be performed with m = S (p, n) fixed, and the differential sound analysis may be omitted. Further, in the processing of [Equation 5], there is a possibility that a signal component having a different sub-frequency with respect to the same note number may be analyzed multiple times, but a value is already set in E2 (q, n). If it is, it may be excluded from selection candidates of the maximum value of E1 (p, n).

  Here, the setting of the analysis frame in the unit section will be described. The following description is similarly applied to the aforementioned selection unit section. FIG. 8 is a diagram illustrating a correspondence relationship between a sample sequence in a unit section extracted from an acoustic signal after time axis expansion processing and a harmonic signal. Among these, Fig.8 (a) is a sample row | line | column in the unit area extracted from the acoustic signal after a time-axis expansion process. By connecting the sample values (4096) in each sample, a wave shape as shown in FIG. Among the 128 harmonic signals, a harmonic signal selected from the unit interval T when performing a correlation operation with an analysis harmonic signal of a treble part whose period is equal to or less than the unit interval length T as shown in FIG. When subtracting a certain contained signal, the analysis frame length T (n) is a length obtained by multiplying the period of the harmonic signal by an integer up to a range in which the period does not exceed the unit section length T. (N) Extract them and use them as analysis frames.

  When one period of the harmonic signal is larger than the unit section length T, the analysis frame length T (n) that is the correlation calculation section is set to one period of the harmonic signal, so that the unit section length T is T (n) −T. Extend the time axis by adding the number of samples. In this case, one period of the harmonic signal is set as four divided sections Q1-Q4. Then, as shown in FIG. 8D, when the unit section length T corresponds to 3/4 period of the harmonic signal, the harmonic signal division section Q3 (1/2 period to 3/4 period) A sample obtained by inverting the sample of the unit section corresponding to に in the time axis direction at the time of 3/4 period of the harmonic signal (270 degrees: boundary between the divided sections Q3 and Q4) is added. Assuming that one period of the harmonic signal is a sine wave, inversion in the time axis direction at the time of 270 degrees is 270 degrees in divided sections Q3 and Q4 (3/4 period to 1 period). It uses a feature that is symmetrical with respect to the axis in the time axis direction. In this way, the time axis is extended, and a waveform analysis frame (5461 samples) as shown in FIG. 8C is obtained. In the examples of FIGS. 8C and 8D, the case where the unit section length T corresponds to 3/4 period of the harmonic signal is shown, but the unit section length T is 3/4 period or more of the harmonic signal. In this case, the same processing is performed, and the analysis frame length T (n) is 5461 samples at the maximum. In this case, any sample in the unit section of the portion exceeding the 3/4 period of the harmonic signal is used by overlapping one of the samples in the unit section.

  Also, as shown in FIG. 9B, when the unit section length T corresponds to more than 1/2 cycle and less than 3/4 period of the harmonic signal (when the last sample of the unit section belongs to the divided section Q3). Since the entire divided section Q4 and a part of the divided section Q3 are missing, a unit section sample corresponding to the divided section Q2 (a section from a quarter cycle to a half cycle) is converted to 1 / of the harmonic signal. Samples that are inverted in the time axis and amplitude axis directions at the time of two periods (180 degrees: the boundary between the divided sections Q2 and Q3) are added to the divided sections Q3, and the samples of the added divided sections Q3 are harmonized. An inverted sample is added at the time of 3/4 period (270 degrees) of the signal. Inversion in the time axis and amplitude axis directions at the time of 180 degrees assumes that one period of the harmonic signal is a sine wave, the divided sections Q2 and Q3 are 180 degrees in the time axis and amplitude axis directions. A characteristic that is a symmetrical shape rotated 180 degrees is used. The time axis is extended in this way, and a waveform analysis frame (up to 8192 samples) as shown in FIG. 9A is obtained. In this case, any sample in the unit section of the portion exceeding the half period of the harmonic signal is used by overlapping one of the samples in the unit section.

Also, as shown in FIG. 10 (b), when the unit section length T corresponds to not less than ¼ period and less than ½ period of the harmonic signal (when the last sample of the unit section belongs to the divided section Q2). Since the whole of the divided sections Q3 and Q4 and a part of the divided section Q2 are missing, the unit section samples corresponding to the Q1 section (section having a quarter period from the head) are used as the quarter period of the harmonic signal. The sample inverted in the time axis direction at the time of (90 degrees: the boundary between the divided section Q1 and the divided section Q2) is added to the divided section Q2, and the added sample of the divided section Q2 is ½ of the harmonic signal. Samples inverted in the time axis and amplitude axis directions at the time of the period (180 degrees) are added to the divided section Q3, and the added samples of the divided section Q3 are added to the 3/4 period (270 degrees) of the harmonic signal. Time at time To add a sample obtained by reversing the direction. The reason for inversion in the time axis direction at the time of 90 degrees is that if one period of the harmonic signal is a sine wave, the divided sections Q1 and Q2 are symmetrical in the time axis direction about 90 degrees. Is used. The time axis is extended in this way, and a waveform analysis frame (maximum 16384 samples) as shown in FIG. 10A is obtained. In this case, any sample in the unit section of the portion exceeding the quarter period of the harmonic signal is used by overlapping any sample in the unit section.
Note that the unit interval length T may correspond to less than ¼ period of the harmonic signal (when the last sample of the unit interval belongs to the divided interval Q1), but the time axis is extended based on this. Even if the correlation calculation is performed, the amount of information used as a source is too small to obtain a significant correlation value. Therefore, the unit section length T is analyzed for frequencies corresponding to less than a quarter cycle of the harmonic signal. It shall not be covered.

  When frequency analysis is performed while changing the analysis frame for each selected unit section q and the spectrum (128 frequency components) is calculated, N types are selected for each selected unit section based on the spectrum calculated in S5. Corresponding to each frequency, a single-tone component composed of frequency information that can specify each frequency, spectrum intensity corresponding to each frequency, and time information that can specify the start and end of the selected unit section is created ( S6). Specifically, the time and time length information of each note number n is added to the calculated spectrum, and [start time, time length, main frequency n, sub frequency S (P (q), n), intensity E2 A single tone component composed of (q, n)] is created. The “start time” may be any information that can identify the start time of the selected unit section in the entire digital sound signal, and in this embodiment, the digital sound signal attached to the start sample (i = 0) of the unit section. The sample number (corresponding to absolute sample address: j) in the whole is recorded. By dividing this absolute sample address by the sampling frequency (44100), the time from the head of the acoustic signal is obtained. In this embodiment, the length of time is variably given for each selected unit section. The time length is the difference up to the start time of the selected unit section that has been subjected to the generalized harmonic analysis that immediately follows (the start of the subsequent selected unit section). Time-start time of the selected unit section). When there is no subsequent selection unit section immediately after (the last selection unit section), the shift width W of the unit section is given as the time length.

Subsequently, in order to correct the variation caused by the processing being enlarged in the time axis direction, processing for correcting each single tone component is performed (S7). When a single tone component is created, a process of adding 12 · log ₂ K to all frequency information (corresponding to the note number value) is performed. For example, when K = 4, the overall pitch is raised by 24 semitones (2 octaves). This process is performed to restore the original state by multiplying the frequency by K times because the frequency is 1 / K by multiplying the number of samples by K in S1. By this correction, the code code having the note number whose note number exceeds the standard value upper limit 127 is deleted. Specifically, the code code with the note number before correction of 128-12 · log ₂ K or more is deleted.

  Subsequently, all start times and time lengths are multiplied by 1 / K. As a result, when converted to a MIDI code by an encoding process described later, the performance time of the entire MIDI code and the sounding time of each note event are reduced to 1 / K. This process is performed in order to set the time to 1 / K and return to the original state because the total performance time is K times by multiplying the number of samples by K in S1.

  As a result of the processing in S7, the frequency (pitch) becomes K times and the time information becomes 1 / K. FIG. 3B shows how the sound component changes due to the correction process in S7. In FIG. 3B, the change of the sound component in the case of K = 2 is indicated by a note. As a result of the correction processing in S7, the left "mi" note is changed to the "mi" note that is one octave higher (double the frequency) on the right side. On the other hand, the left quarter note is changed to a half eighth note on the right side.

  Once the processing of making the frequency of each single tone component K times and the time axis 1 / K times is performed, auditory filter correction is performed (S8). Specifically, the frequency range of 0 ≦ n ≦ 127 obtained by the processes up to S7 is divided into 24 bands, and correction is performed on E2 (q, n) for each band.

  A band-pass filter used for auditory filter correction in this embodiment will be described. FIG. 11 shows a bandpass filter created based on Non-Patent Document 1. In FIG. 11, band numbers 1 to 24 are assigned to 24 bands, and the lower limit frequency, the center frequency, the upper limit frequency, and the bandwidth are shown for the 24 bands specified by each band number. The lower limit frequency and the upper limit frequency indicate the lower limit frequency and the upper limit frequency of each band, respectively. The lower limit frequency and the upper limit frequency of each band are the upper and lower cut-off frequencies in each band filter, respectively, above or below the corresponding frequency. The frequency component is not passed through (the filter gain dB value is close to minus infinity), which matches the upper limit frequency and the lower limit frequency. The center frequency is a peak frequency through which the signal component passes most in the bandpass filter (filter gain dB value is close to plus infinity). The bandwidth is the width of each band, and is a difference value between the lower limit frequency and the upper limit frequency.

FIG. 12 shows the band filter shown in FIG. 11 converted into MIDI note number units. That is, the relationship between the frequency f (n) (unit: Hz) shown in FIG. 11 and the note number n shown in FIG. 12 satisfies f (n) = 440 · 2 ^{(n−69) / 12.} ing. However, the upper limit pitch of each band is set to a value obtained by subtracting 1 from the lower limit pitch of one higher band so that the note numbers do not overlap in a plurality of bands.

  Each band is specified in the range of note number n such that Nr (b) ≦ n <Nr (b + 1) using band number b. The 25 band boundary values Nr (b) (b = 1,..., 25) are Nr (b) = {17, 45, 57, 64, 69, 72, 76, 79, 82, 85, 88. , 91, 93, 96, 98, 101, 104, 106, 109, 113, 116, 119, 123, 127, 128}. Of the 25 band boundary values Nr (b), the one corresponding to b = 1 to 24 indicates the lower limit pitch of each band, and only the band boundary value “128” corresponding to b = 25 is the center of the band. The pitch is shown. The bandpass filters shown in FIGS. 11 and 12 are typical examples, and the numerical values of the upper limit, the center, and the lower limit can be changed as appropriate. Also, regarding the number of bands, 24 is currently the established theory, but is not limited to this, and can be changed based on new knowledge.

  FIG. 13 is a flowchart showing details of the auditory filter correction in S8. First, the band number b is set to 1 (S21). As a result, processing is performed from band number 1 which is the lowest band. Next, the maximum note number nbmax that maximizes the correlation value array E2 (q, n) within the range from the lower limit pitch Nr (b) corresponding to the band number b to the upper limit pitch Nr (b + 1) −1 is searched. (S22).

  Subsequently, the correlation value array E2 (q, n) corresponding to the note number n other than the maximum note number nbmax within the range from the lower limit pitch Nr (b) corresponding to the band number b to the upper limit pitch Nr (b + 1) −1. ) Is attenuated by a predetermined ratio (S23). Specifically, attenuation correction is performed by multiplying the correlation value array E2 (q, n) corresponding to each note number n other than the maximum note number nbmax by a predetermined real value γ less than 1. Since the real value γ is for performing attenuation correction, any value less than 1 can be set as appropriate. However, in this embodiment, γ = 0.1 is set. Therefore, the correlation value E2 (q, n) corresponding to the note number n other than the maximum note number nbmax is attenuated to 1/10 by the processing in S23.

  When band number b is searched for maximum note number nbmax and correlation values corresponding to other note numbers are attenuated, band number b is counted up (b ← b + 1) (S24). Then, the band number 24, which is the highest band, is compared with the band number b (S25). If b is 24 or less as a result of the comparison in S25, the process returns to S22, and the search for the maximum note number nbmax (S22) and the attenuation process of the correlation value corresponding to another note number (S23) are performed for the next band. . On the other hand, if b is 25 or more as a result of the comparison in S25, the processing has been performed for all the bands, and the auditory filter correction processing is terminated.

  The result of the auditory filter correction process in S8 is output as a corrected correlation strength array E ′ (q, n). Of the 128 E2 (q, n), 104 corrected correlation values in the corrected correlation intensity array E ′ (q, n) are attenuated and the remaining 24 are not attenuated.

  Once the auditory filter correction processing has been performed, processing for connecting (integrating) the single tone components in successive selection unit sections is performed (S9). Specifically, when a single sound component in consecutive selected unit sections satisfies a predetermined connection condition, two single sound components are connected. Here, FIG. 14 shows the relationship between the selected unit section and the unit section that are subject to the connection determination.

  FIG. 14 shows an example in which unit sections 1, 5, and 6 are selected as selected unit sections 1, 2, and 3 among unit sections 1-6, as in FIG. 4 (b). The present embodiment is characterized in that it is performed not between selected unit sections selected as sound components for connection determination, but between selected unit sections and adjacent unit sections. In the example of FIG. 14, when it is determined whether or not the selection unit section 1 and the selection unit section 2 (unit section 5) are connected, the connection condition is set between the selection unit section 1 and the selection unit section 2 (unit section 5). Rather than determining, the connection condition is determined between the selected unit interval 2 (unit interval 5) and the immediately preceding unit interval 4. As a result, the closest unit section in time is taken into consideration when determining the connection, and the sound components can be appropriately connected.

  As a connection condition, a state having continuity as the same sound can be appropriately set. In the present embodiment, four conditions are determined when determining whether or not to connect the selection unit interval q and the selection unit interval q + 1. The first condition is that a value obtained by subtracting the intensity of a single sound component in the immediately preceding unit section P (q + 1) -1 from the intensity of the single sound component in the selected unit section q + 1 (P (q + 1) as a unit section) is a predetermined threshold. Must be less than the value Ldif. The second condition is that the intensity of both is greater than a predetermined threshold value Lmin. The third condition is that the difference of the frequency considering the sub-frequency is less than a predetermined threshold value Ndif in consideration of a shift of 1 up and down with respect to the note number. The fourth condition is that when the selection unit section q is already connected to another selection unit section, the sub-frequency of the top selection unit section qo and the selection unit section q + 1 connected to the selection unit section q is considered. The frequency difference is less than a predetermined threshold value Nadif. When all the above four conditions are satisfied, it is assumed that there is continuity, and the subsequent single sound component is connected to the front single sound component. Also in the first condition and the second condition, a shift of 1 up and down with respect to the note number is considered. In the present embodiment, whether or not to perform the connection is determined by determining whether or not the four conditions according to the following [Equation 7] to [Equation 9] are taken into account in consideration of the transition of the note number.

[Formula 7]
E1 (P (q + 1), n) -E1 (P (q + 1) -1, n) <Ldif
E1 (P (q + 1) -1, n)> Lmin and E1 (P (q + 1), n)> Lmin
| S (P (q + 1) -1, n) -S (P (q + 1), n) | <Ndif
| S (P (qo), n) -S (P (q + 1), n) | <Nadif

  In the above [Equation 7], the expression on the first line indicates that the value obtained by subtracting the intensity of the preceding single sound component from the intensity of the subsequent single sound component is less than a predetermined threshold Ldif. , Indicating that the intensity of both the preceding single sound component and the subsequent single sound component is larger than the predetermined threshold value Lmin, and the expression in the third row indicates that the frequency difference in note number units considering the sub frequency is a predetermined threshold value Ndif. It shows that it is less than. The expression on the fourth line is a note number that takes into account the sub-frequency of the head selection unit section qo and the subsequent selection unit section q + 1 when the front selection unit section is already connected to the front selection unit section. It shows that the difference in unit frequency is less than a predetermined threshold value Nadif.

[Formula 8]
E1 (P (q + 1), n-1) -E1 (P (q + 1) -1, n) <Ldif
E1 (P (q + 1) -1, n)> Lmin and E1 (P (q + 1), n-1)> Lmin
| S (P (q + 1) -1, n) -S (P (q + 1), n-1) -M | <Ndif
| S (P (qo), n) -S (P (q + 1), n-1) -M | <Nadif

  The above [Equation 7] compares the same note numbers, whereas the above [Equation 8] differs in that the following single-tone component is reduced by one note number. The expression in the third row is obtained by calculating the frequency S (P (q + 1), n−1) for determining the first spectral intensity E1 (P (q + 1), n−1) and the first The difference from the sub-frequency S (P (q + 1) -1, n) that determines the spectrum intensity E1 (P (q + 1) -1, n) is less than a predetermined threshold value Ndif. The expression in the fourth row is obtained by calculating the frequency S (P (q + 1), n−1) for determining the first spectral intensity E1 (P (q + 1), n−1) and the first It is shown that the difference from the sub-frequency S (P (qo), n) that determines the spectrum intensity E1 (P (qo), n) is less than a predetermined threshold value Nadif.

[Formula 9]
E1 (P (q + 1), n + 1) -E1 (P (q + 1) -1, n) <Ldif
E1 (P (q + 1) -1, n)> Lmin and E1 (P (q + 1), n + 1)> Lmin
| S (P (q + 1) -1, n) -S (P (q + 1), n + 1) + M | <Ndif
| S (P (qo), n) -S (P (q + 1), n + 1) + M | <Nadif

  While the above [Equation 7] compares the same note numbers, the above [Equation 9] is different in that the subsequent single sound component is increased by one note number. The expression in the third row is obtained by calculating the frequency S (P (q + 1), n + 1) for determining the first spectral intensity E1 (P (q + 1), n + 1) and the first spectral intensity E1 when the frequency is in note number units. It shows that the difference from the sub-frequency S (P (q + 1) -1, n) for determining (P (q + 1) -1, n) is less than a predetermined threshold value Ndif. The expression in the fourth row is expressed as follows: when the frequency is in note number units, the frequency S (P (q + 1), n + 1) for determining the first spectral intensity E1 (P (q + 1), n + 1) and the first spectral intensity E1 It shows that the difference from the sub-frequency S (P (qo), n) for determining (P (qo), n) is less than a predetermined threshold value Nadif.

  As a specific processing procedure in the present embodiment, first, it is determined whether or not all the conditions shown in [Formula 7] are satisfied. And when satisfy | filling, it progresses to a connection process. If any of the conditions shown in [Formula 7] does not satisfy one of the conditions, it is determined whether or not all the conditions shown in [Formula 8] are satisfied. And when satisfy | filling, it progresses to a connection process. If any of the conditions shown in [Formula 8] does not satisfy one of the conditions, it is determined whether or not all the conditions shown in [Formula 9] are satisfied. And when satisfy | filling, it progresses to a connection process. If any of the conditions shown in [Equation 9] does not satisfy one of the conditions, it is determined that the connection is not performed.

  The main frequency, sub frequency, and intensity after connection are each taken from the value of the larger single tone component, and the time length is given as the sum of both. At this time, the intensity is the value after the auditory filter correction is performed. In this embodiment, specific threshold values as connection conditions are as follows: Ldif = 10 [unit: 128 step velocity conversion], Lmin = 1 [unit: 128 step velocity conversion], Ndif = 4/25 [unit: note number] Conversion], Nadif = 8/25 [unit: note number conversion]. Since the concatenation process is performed before conversion to a code code, each threshold value is converted into a note number and velocity. As a result of the connecting process in S9, the single sound components that have not been connected remain. In addition, a connected sound component is obtained by connection, and the connected sound component is [start time, time length, main frequency n, sub frequency S (P (q), n), intensity E, as with the single sound component. '(Q, n)], of which the time length has a value greater than the single tone component. By the connection process, a single sound component and a connected sound component are mixed, and these are hereinafter collectively referred to as a sound component. It should be noted that the connection processing in S9 is generally preferable because it is expressed by a long note, and the amount of codes is reduced, so that a smooth and natural performance can be performed with a MIDI sound source. If a code is not added, subtle temporal changes in sound such as vibrato disappear, and it may sound unnatural with a MIDI sound source. When the connection process in S9 is not performed, all are expressed as short notes.

When the connection process of S9 is completed, the sound component of [start time, time length, main frequency n, sub frequency S (P (q), n), intensity E ′ (q, n)] obtained finally is obtained. And converted into a code code (S10). The format of the code code may be any format as long as it has frequency information, spectrum intensity corresponding to each frequency, and time information that can specify the start and end of a unit section. However, in this embodiment, the data is converted to the MIDI format. In MIDI, sound generation start and sound generation end are generated as separate events, so in this embodiment, one sound component is converted into two MIDI note events. Specifically, a note-on event of note number n is issued at the “start time”, and the velocity value is 128 · {E ′ (q, n) with the maximum value of intensity E ′ (q, n) as Emax. / Emax} is given by ^1/4 . In Standard MIDI File, it is necessary to give the time as a relative time (delta time) with the immediately preceding event, and the time unit can be defined by an arbitrary integer value, for example, converted to 1/1536 [seconds]. Give. Then, the note-off event of the note number n is issued at the end time specified by “start time” + “time length” (the end time given by “time length” in the delta time). At this time, the time length is multiplied by a real number between 0 and 1. This is because a note-off instruction is given early in consideration of the reverberation of the MIDI sound source, although it depends on the tone color of the MIDI sound source to be used. Even if the time length is used as it is, there is no problem in the processing of the MIDI sound source.

  After completing the code code conversion process of S10, next, necessary adjustments are made to the code code (S11). For example, when converting to a MIDI code as a code code, it is necessary to adjust the number of simultaneous sounds in order to consider the number of simultaneous sounds that can be processed by a MIDI sound source. When the number of simultaneous sounds that can be processed by the MIDI sound source is 32, the number of note events during the sound generation period (note-on state) is continuously counted in the time axis direction, and there are simultaneously more than 32 note events. Is found, each pair of note-off events is searched in the neighborhood, and the priority is determined by the product (energy value) of the velocity value and duration value (note-off time-note-on time) of each note event pair. Evaluation is performed, and a note event pair having a low priority (low energy value) is locally deleted so as to be equal to or less than the specified number of chords (in this case, “32”). “Locally” means that only a part where there are more than 32 note event pairs exists. At this time, if either the velocity value or the duration value is lower than the predetermined lower limit value, the deletion process is also performed regardless of the priority. At this stage, the effect of the auditory filter correction proposed in this application works remarkably, and 24 note events among the 32 note events to be selected are included in each of the 24 bands defined in FIG. At least one is selected from each of the note numbers, and the remaining eight note numbers are more likely to be selected redundantly from any of the eight types of bands. In practice, as shown in FIG. 5, the note numbers to be encoded are often limited to the range of 21 to 108 that is the piano range, and in this case, the effective bands in FIG. Therefore, at least one note number is selected from each of the 18 note numbers included in each of the 18 bands, and the remaining 14 note numbers can be selected redundantly from any of the 14 bands. Increases nature. However, this is to the extent that the correction is made so that the probability of being selected in this way is high, and depending on the selected unit section, the note event to be selected is selected from any band of less than 18 types, It is possible that three or more types of note numbers are selected in duplicate from one band. When the MIDI data encoded in this way is reproduced using a MIDI sound source capable of simultaneously generating 32 chords or more, it is not limited to a specific frequency band, and covers the frequency range of all frequency bands that can be heard simultaneously by the human auditory system. Therefore, it becomes possible to hear clearly with a sense of presence equivalent to a human being listening to the original sound signal before encoding as it is.

  Furthermore, the bit rate is adjusted in order to consider the bit rate that can be processed by the code code. When converting to MIDI code, the number of note event pairs is counted in the time axis direction, for example, at 1-second intervals, and the average amount of code data is 5 bytes (40 bits). Is 9000 [bps (bits / second)], when a section in which the number of events per second exceeds 9000/40 = 225 is found, it is paired with a note-on or note-off event existing in the section. Searches for note-off or note-on events that become in the neighborhood, evaluates the priority by the product (energy value) of the velocity value and duration value (note-off time-note-on time) of each note event pair, and designates the event Note event pairs with low priority (low energy value) to be less than the number (in this case "225") It performs a process of deleting Tokoro manner. At this time, if either the velocity value or the duration value is lower than the predetermined lower limit value, the deletion process is also performed regardless of the priority. Even at this stage, the effect of the auditory filter correction proposed in the present application works remarkably, and at least one note included in each of the 24 bands (18 in the case of operation limited to the piano sound range) defined in FIG. Number note events are not deleted and are more likely to remain.

  The preferred embodiments of the present invention have been described above. However, the present invention is not limited to the above embodiments, and various modifications can be made. For example, in the above embodiment, the analysis between the note numbers is performed using M differential sounds (sub-frequency), but the differential sounds are not used and the analysis is performed using only N types of frequencies corresponding to the note numbers. You may do it. In this case, the analysis accuracy is slightly reduced, but the processing load is reduced because the number of frequencies to be analyzed is reduced. In the case where the differential sound is not used, in the determination of the connection processing of the single sound component in S9, in the [Equation 7] to [Equation 9], the expressions of the third and fourth lines are not determined.

  In the above embodiment, before setting the unit section, the number of samples in the time series direction is expanded in S1, and after the creation of the single sound component, the frequency of each single sound component is multiplied by K times as the correction of the single sound component in S7. Although the processing for increasing the time axis to 1 / K is performed, it is not always necessary to perform the processing when a particularly high time resolution is not required.

  In the above embodiment, the frequency analysis is divided into the first frequency analysis and the second frequency analysis. As a result of the first frequency analysis, the second frequency analysis is performed on the selected unit section that satisfies a predetermined condition. Although it was made to perform, you may make it perform the well-known frequency analysis which is disclosed by patent documents 1-3 with respect to each unit area.

  The present invention uses a technology for converting an acoustic signal obtained by PCM or the like into a code code such as a MIDI code, and uses broadcast media (digital radio / television broadcast such as terrestrial / BS) and communication media (CS broadcast, Internet). -Streaming broadcasting, mobile phone service, portable music distribution service, etc.) and audio media production industry for package media (CD, DVD, BlueRay, memory IC card, etc.).

1 ... CPU
2 ... RAM
3 ... Data storage device 4 ... Program storage device 5 ... Key input I / F
6 ... Data I / O I / F
7 ... Display output I / F

Claims (22)

An encoding method for encoding an acoustic signal given as a sample sequence of J time series digitized at a predetermined sampling frequency,
A unit section composed of a predetermined number T (T <J) samples with respect to the sample sequence, while overlapping a predetermined number W (W <T) samples in the time axis direction with an adjacent unit section. Section setting stage to be set,
For each unit section, by performing frequency conversion for at least N types of frequencies f (n) to be analyzed , the first spectral intensity corresponding to the N types of frequencies for the unit section p. A first spectrum calculation stage for calculating E1 (p, n);
An evaluation value based on a change for each frequency corresponding to the first spectral intensity E1 (p-1, n) in the unit section p-1 located immediately before the unit section p is greater than a predetermined threshold value. Only when it is larger, the unit interval p is selected as the q (q ≦ p) -th selection unit interval q, and at least N kinds of frequencies f (n) to be analyzed are selected in the first spectrum calculation stage. A second spectrum calculation step of calculating a second spectrum intensity E2 (q, n) corresponding to the N kinds of frequencies by performing a frequency conversion with higher accuracy than the frequency conversion;
The N types of frequencies f (n) are divided into a predetermined number of frequency groups so as not to overlap each other, and the second spectral intensity E2 (q, n) A spectrum that corrects the second spectrum intensity E2 (q, n) other than the frequency having the maximum value in n) so as to be attenuated by a predetermined ratio to generate a corrected spectrum intensity E ′ (q, n). A correction stage;
An encoding step of generating a code code of a predetermined format based on the time difference between the start time of the unit section and the start time of the immediately following unit section, and the corrected spectrum intensity of the unit section;
A method for encoding an acoustic signal, comprising: In claim 1,
The predetermined number of frequency groups are based on the characteristics of the human auditory system, and the value of n is defined as a standard note number, and 17, 45, 57, 64, 69, 72, 76, 79, 82, 85 , 88, 91, 93, 96, 98, 101, 104, 106, 109, 113, 116, 119, 123, 127, 128. Method. In claim 1,
The predetermined number of frequency groups have frequency f (n) values of 20, 100, 200, 300, 400, 510, 630, 770, 920, 1080, 1270, 1480, 1720, based on characteristics of the human auditory system. An acoustic signal encoding method characterized by being set at 24 with 2000, 2320, 2700, 3150, 3700, 4400, 5300, 6400, 7700, 9500, 12000, and 15500 Hz as boundaries. In any one of Claims 1-3 ,
In the first spectrum calculation stage and the second spectrum calculation stage, predetermined M types of sub-frequency f (n, m) are set in a range not exceeding adjacent frequencies for each of the N types of frequencies f (n). Then, as the first spectral intensity E1 (p, n) and the second spectral intensity E2 (q, n), intensity values corresponding to the sub-frequency indicating the highest intensity among the M types of sub-frequency are obtained. A method for encoding an acoustic signal, comprising: calculating an acoustic signal. In any one of Claims 1-4 ,
The encoding step defines P (q) = p when the selection unit interval q is the p-th unit interval p with respect to two adjacent selection unit intervals q and selection unit interval q + 1. , First spectral intensities E1 (P (q + 1), n), E1 (P (q + 1), n corresponding to frequencies f (n), f (n-1), f (n + 1) in the selected unit interval q + 1. n-1), E1 (P (q + 1), n + 1) and the first spectrum corresponding to the frequency f (n) in the unit section P (q + 1) -1 located immediately before the selected unit section q + 1 The difference from the intensity E1 (P (q + 1) -1, n) is less than a predetermined threshold Ldif, and the first spectrum intensity E1 (P (q + 1), n), E1 (P (q + 1), n -1), E1 (P (q + 1), n + 1) and When the first spectral intensity E1 (P (q + 1) -1, n) is greater than a predetermined threshold value Lmin, the selection unit interval q and the selection unit interval q + 1 are connected, and the time difference that is the basis of the code code And a method of encoding an acoustic signal using a time difference between a head time of a selected unit section q defined in the selected unit section q and a head time of a selected unit section q + 2 immediately after the selected unit section q + 1. . In claim 5 ,
The encoding step includes at least sub-frequency determining the first spectral intensities E1 (P (q + 1), n), E1 (P (q + 1), n−1), E1 (P (q + 1), n + 1). When the condition that the difference in note number units between any one and the sub-frequency that determines the first spectral intensity E1 (P (q + 1) -1, n) is less than a predetermined threshold value Ndif is further satisfied As long as the selection unit interval q and the selection unit interval q + 1 are connected, the method of encoding an acoustic signal is characterized. In claim 6 ,
When the selection unit interval q is already connected to another selection unit interval, the first selection unit interval to which the selection unit interval q is connected is defined as qo,
The encoding step includes at least sub-frequency determining the first spectral intensities E1 (P (q + 1), n), E1 (P (q + 1), n−1), E1 (P (q + 1), n + 1). Only when the condition that the difference in note number units between any one and the sub-frequency that determines the first spectral intensity E1 (P (qo), n) is less than a predetermined threshold value Nadif is satisfied A method for encoding an acoustic signal, comprising connecting the selection unit interval q and the selection unit interval q + 1. In any one of claims 1 to 7,
In the first spectrum calculation step, N element signals to be constituent elements of the section signal of the unit section correspond to integer multiples of the period of the frequency f (n), respectively, and T closest to the T (N) an element signal preparation stage to prepare as samples;
By performing a correlation operation between the element signal corresponding to each of the N frequencies f (n) and a section signal composed of T (n) samples of the corresponding unit section p, It is constituted by a correlation calculation stage for calculating the spectral intensity E1 (p, n),
The second spectrum calculation step includes:
The element signal preparation step;
Correlation is performed between the element signal corresponding to each of the N frequencies f (n) and the section signal composed of T (n) samples of the corresponding selection unit section q, and the correlation value is the highest. A harmonic signal selection step of selecting an element signal corresponding to the high frequency f (nmax) as a harmonic signal;
By using T (nmax) samples given by the product of the selected harmonic signal and the correlation value obtained for the harmonic signal as an inclusion signal, and subtracting the inclusion signal from the interval signal, T (nmax) A differential signal calculation step for obtaining a differential signal composed of a number of samples,
Using the T (n) samples updated to reflect the T (nmax) samples as new interval signals, the harmonic signal selection step and the difference signal calculation step are executed to obtain new inclusion signals and difference signals. N content signals are obtained by repeatedly performing the obtained process, and the second spectrum intensity E2 (q, n) corresponding to the N kinds of frequencies is calculated based on the correlation values of the obtained content signals. A method of encoding an acoustic signal characterized by the above. In claim 8 ,
The correlation calculation step includes:
A correlation calculation corresponding to the first W sample in the unit interval p-1 is performed on the previous correlation calculation result corresponding to each frequency f (n) in the unit interval p-1 located immediately before, and a correlation value for each frequency is obtained. Is subtracted from the previous correlation calculation result, a correlation calculation corresponding to the last W sample in the T (n) samples in the unit interval p is performed, and a correlation value for each frequency is added to the previous correlation calculation result. Thus, a correlation calculation result corresponding to each frequency f (n) in the unit interval p is acquired, and the first spectral intensity E1 (p, n) is calculated based on the correlation calculation result. A method for encoding an acoustic signal. In any one of claims 1 to 9,
The encoding step performs encoding using the MIDI format as the code code, uses a note number as the frequency information of the code code, uses velocity as the second spectral intensity, and selects the selected unit interval q Delta time 1 and delta time 2 that are relative times from the immediately preceding MIDI event, respectively, and the defined note number, velocity, delta A method for encoding an audio signal, wherein a MIDI note-on event is created based on time 1 and a MIDI note-off event is created based on note number and delta time 2. An encoding device for encoding an acoustic signal given as a sample sequence of J time series digitized at a predetermined sampling frequency,
A unit section composed of a predetermined number T (T <J) samples with respect to the sample sequence, while overlapping a predetermined number W (W <T) samples in the time axis direction with an adjacent unit section. Section setting means to set;
For each unit section, spectrum conversion corresponding to the N types of frequencies is calculated for each unit section by performing frequency conversion for at least N types of frequencies f (n) to be analyzed. Spectrum calculation means;
The N types of frequencies f (n) are divided into a predetermined number of frequency groups so as not to overlap with each other, and the maximum value among the spectrum intensities of the frequencies included in each frequency group is obtained for each unit section. Spectrum correction means for performing correction so as to attenuate the spectrum intensity other than the frequency by a predetermined ratio, and creating corrected spectrum intensity;
And time difference between the start time and the start time of the unit segment immediately following the unit segment, based on the corrected spectrum intensity of the unit section, possess encoding means for generating a predetermined format code code, and,
The predetermined number of frequency groups are based on the characteristics of the human auditory system, and the value of n is defined as a standard note number, and 17, 45, 57, 64, 69, 72, 76, 79, 82, 85 , coding of audio signals, characterized in Rukoto set in 24 bounded by 88,91,93,96,98,101,104,106,109,113,116,119,123,127,128 apparatus. An encoding device for encoding an acoustic signal given as a sample sequence of J time series digitized at a predetermined sampling frequency,
A unit section composed of a predetermined number T (T <J) samples with respect to the sample sequence, while overlapping a predetermined number W (W <T) samples in the time axis direction with an adjacent unit section. Section setting means to set;
For each unit section, spectrum conversion corresponding to the N types of frequencies is calculated for each unit section by performing frequency conversion for at least N types of frequencies f (n) to be analyzed. Spectrum calculation means;
The N types of frequencies f (n) are divided into a predetermined number of frequency groups so as not to overlap with each other, and the maximum value among the spectrum intensities of the frequencies included in each frequency group is obtained for each unit section. Spectrum correction means for performing correction so as to attenuate the spectrum intensity other than the frequency by a predetermined ratio, and creating corrected spectrum intensity;
And time difference between the start time and the start time of the unit segment immediately following the unit segment, based on the corrected spectrum intensity of the unit section, possess encoding means for generating a predetermined format code code, and,
The predetermined number of frequency groups have frequency f (n) values of 20, 100, 200, 300, 400, 510, 630, 770, 920, 1080, 1270, 1480, 1720, based on characteristics of the human auditory system. encoding apparatus of an audio signal, wherein Rukoto set in 24 bounded by 2000,2320,2700,3150,3700,4400,5300,6400,7700,9500,12000,15500Hz. An encoding device for encoding an acoustic signal given as a sample sequence of J time series digitized at a predetermined sampling frequency,
A unit section composed of a predetermined number T (T <J) samples with respect to the sample sequence, while overlapping a predetermined number W (W <T) samples in the time axis direction with an adjacent unit section. Section setting means to set;
For each unit section, by performing frequency conversion for at least N types of frequencies f (n) to be analyzed , the first spectral intensity corresponding to the N types of frequencies for the unit section p. First spectrum calculating means for calculating E1 (p, n);
An evaluation value based on a change for each frequency corresponding to the first spectral intensity E1 (p-1, n) in the unit section p-1 located immediately before the unit section p is greater than a predetermined threshold value. Only when it is larger, the unit interval p is selected as the q (q ≦ p) -th selection unit interval q, and at least N types of frequencies f (n) to be analyzed are selected in the first spectrum calculation means. Second spectrum calculation means for calculating a second spectrum intensity E2 (q, n) corresponding to the N types of frequencies by performing a frequency conversion with higher accuracy than the frequency conversion;
The N types of frequencies f (n) are divided into a predetermined number of frequency groups so as not to overlap each other, and the second spectral intensity E2 (q, n) A spectrum that corrects the second spectrum intensity E2 (q, n) other than the frequency having the maximum value in n) so as to be attenuated by a predetermined ratio to generate a corrected spectrum intensity E ′ (q, n). Correction means;
Encoding means for generating a code code of a predetermined format based on the time difference between the start time of the unit section and the start time of the immediately following unit section, and the corrected spectrum intensity of the unit section;
A device for encoding an acoustic signal, comprising: According to claim 1 3,
The first spectrum calculating means and the second spectrum calculating means set predetermined M kinds of sub-frequency f (n, m) within a range not exceeding adjacent frequencies for each of the N kinds of frequencies f (n). Then, as the first spectral intensity E1 (p, n) and the second spectral intensity E2 (q, n), intensity values corresponding to the sub-frequency indicating the highest intensity among the M types of sub-frequency are obtained. An apparatus for encoding an acoustic signal, comprising: calculating an acoustic signal. According to claim 1 3 or claim 1 4,
The encoding means defines P (q) = p when the selection unit interval q is the p-th unit interval p for two adjacent selection unit intervals q and selection unit interval q + 1. , First spectral intensities E1 (P (q + 1), n), E1 (P (q + 1), n corresponding to frequencies f (n), f (n-1), f (n + 1) in the selected unit interval q + 1. n-1), E1 (P (q + 1), n + 1) and the first spectrum corresponding to the frequency f (n) in the unit section P (q + 1) -1 located immediately before the selected unit section q + 1 The difference from the intensity E1 (P (q + 1) -1, n) is less than a predetermined threshold Ldif, and the first spectrum intensity E1 (P (q + 1), n), E1 (P (q + 1), n -1), E1 (P (q + 1), n + 1) and When the first spectral intensity E1 (P (q + 1) -1, n) is greater than a predetermined threshold value Lmin, the selection unit interval q and the selection unit interval q + 1 are connected, and the time difference that is the basis of the code code And a time difference between the start time of the selected unit section q defined in the selected unit section q and the start time of the selected unit section q + 2 immediately after the selected unit section q + 1 is used. . In claim 15 ,
The encoding means includes at least a sub-frequency for determining the first spectral intensity E1 (P (q + 1), n), E1 (P (q + 1), n-1), E1 (P (q + 1), n + 1). When the condition that the difference in note number units between any one and the sub-frequency that determines the first spectral intensity E1 (P (q + 1) -1, n) is less than a predetermined threshold value Ndif is further satisfied As long as the selection unit interval q and the selection unit interval q + 1 are connected, the apparatus for encoding an acoustic signal is characterized. In claim 16 ,
When the selection unit interval q is already connected to another selection unit interval, the first selection unit interval to which the selection unit interval q is connected is defined as qo,
The encoding means includes at least a sub-frequency for determining the first spectral intensity E1 (P (q + 1), n), E1 (P (q + 1), n-1), E1 (P (q + 1), n + 1). Only when the condition that the difference in note number units between any one and the sub-frequency that determines the first spectral intensity E1 (P (qo), n) is less than a predetermined threshold value Nadif is satisfied An apparatus for encoding an acoustic signal, wherein the selection unit interval q and the selection unit interval q + 1 are connected. In claims 1 to 3 any one of claims 1-7,
The first spectrum calculating means corresponds to N element signals to be constituent elements of the section signal of the unit section, each corresponding to an integral multiple of the period of the frequency f (n), and T closest to the T (N) Element signal preparation means for preparing as samples,
By performing a correlation operation between the element signal corresponding to each of the N frequencies f (n) and a section signal composed of T (n) samples of the corresponding unit section p, It is comprised by the correlation calculating means which calculates spectrum intensity | strength E1 (p, n),
The second spectrum calculation means includes:
The element signal preparation means;
Correlation is performed between the element signal corresponding to each of the N frequencies f (n) and the section signal composed of T (n) samples of the corresponding selection unit section q, and the correlation value is the highest. Harmonic signal selection means for selecting an element signal corresponding to a high frequency f (nmax) as a harmonic signal;
By using T (nmax) samples given by the product of the selected harmonic signal and the correlation value obtained for the harmonic signal as an inclusion signal, and subtracting the inclusion signal from the interval signal, T (nmax) Differential signal calculation means for obtaining a differential signal composed of a number of samples,
The T (nmax) samples updated to reflect the T (nmax) samples are used as new section signals, and the processing by the harmonic signal selection means and the difference signal calculation means is executed to obtain new inclusion signals and differences. N content signals are obtained by repeatedly performing the signal obtaining process, and the second spectrum intensity E2 (q, n) corresponding to the N types of frequencies is calculated based on the correlation values of the obtained content signals. An apparatus for encoding an acoustic signal. In claim 18 ,
The correlation calculation means includes
A correlation calculation corresponding to the first W sample in the unit interval p-1 is performed on the previous correlation calculation result corresponding to each frequency f (n) in the unit interval p-1 located immediately before, and a correlation value for each frequency is obtained. Is subtracted from the previous correlation calculation result, a correlation calculation corresponding to the last W sample in the T (n) samples in the unit interval p is performed, and a correlation value for each frequency is added to the previous correlation calculation result. Thus, a correlation calculation result corresponding to each frequency f (n) in the unit interval p is acquired, and the first spectral intensity E1 (p, n) is calculated based on the correlation calculation result. An apparatus for encoding an acoustic signal. In any one of claims 19 claims 1 to 3,
The encoding means performs encoding using the MIDI format as the code code, uses a note number as the frequency information of the code code, uses velocity as the second spectral intensity, and selects the selected unit interval q Delta time 1 and delta time 2 that are relative times from the immediately preceding MIDI event, respectively, and the defined note number, velocity, delta An apparatus for encoding an audio signal, wherein a MIDI note-on event is created based on time 1 and a MIDI note-off event is created based on note number and delta time 2. Programs for executing the encoding method of the audio signal according to the computer of claims 1 to claim 1 0. A program for causing a computer to function as the acoustic signal encoding device according to any one of claims 11 to 20 .

Images