JP2003263155A - Frequency analyzer and acoustic signal encoding device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a frequency analyzer and an acoustic signal encoding device, capable of reducing a total period of time required for correlation calculation by removing overlapped correlation calculation in an overlapped area of unit sections. <P>SOLUTION: Unit sections each of which a reference unit for analysis are set up so that adjacent unit sections (dI, dI<SB>+1</SB>) on time series are mutually overlapped and the time series signals of respective unit sections are successively extracted as section signals. The correlation of the unit section dI<SB>+1</SB>is calculated by calculating the correlation of a signal located on the head part dh of the preceding section signal with a prescribed harmonic function, calculating the correlation of a signal located on the end part db of the current section signal with the prescribed harmonic signal, subtracting the correlation value calculated for the head part dh from the whole correlation value of the preceding section dI, and adding the correlation value calculated for the end part db to the whole correlation value of the preceding section dI. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

The present invention relates to broadcast media (radio, television), communication media (CS video / audio distribution, Internet music distribution, communication karaoke), package media (CD, MD, cassette, video, LD, CD). −
Production of various audio contents provided in ROM, game cassettes, solid-state memory media for portable music players, etc., as well as music performance recording signals, musical score publishing, MIDI data for communication karaoke distribution, and automatic performance data for electronic musical instruments with performance guide function. , Automatic music transcription technology for automatically creating ringing melody data for mobile phones, PHS, pagers, etc.

[0002]

2. Description of the Related Art A time series signal represented by an acoustic signal is
A plurality of periodic signals are included as its constituent elements. Therefore, a method of analyzing what kind of periodic signal is included in a given time series signal has been known for a long time. For example, Fourier analysis is widely used as a method for analyzing frequency components included in a given time series signal.

By using such a frequency analysis method for time series signals, it is possible to encode acoustic signals.
With the spread of computers, it has become easy to sample the analog sound signal that is the original sound at a predetermined sampling frequency, quantize the signal strength at each sampling, and capture it as digital data. If a method such as Fourier analysis is applied to the data and the frequency components included in the original sound signal are extracted, the original sound signal can be encoded by the code indicating each frequency component.

MIDI (Musical Instrume) was born from the idea of encoding musical instrument sounds by electronic musical instruments.
The nt Digital Interface) standard has also been actively used with the spread of personal computers. Code data according to this MIDI standard (hereinafter referred to as MID
Basically, the I data) is data that describes the operation of the musical instrument playing, such as which keyboard key of the musical instrument was played and with what strength. The MIDI data itself contains the actual sound. Waveform not included. Therefore, when reproducing the actual sound, the MI that stores the waveform of the instrument sound is stored.
Although a DI sound source is required separately, its high coding efficiency has been attracting attention, and the MIDI coding and decoding technology is currently used for musical instrument performance, musical instrument practice, composition, etc. using a personal computer. It is widely adopted in software that does.

Therefore, a time-series signal typified by an acoustic signal is analyzed by a predetermined method to extract a periodic signal which is a constituent element thereof, and the extracted periodic signal is M
Proposals have been made to encode using IDI data. For example, JP-A-10-247099, JP-A-11-73199, and JP-A-11-73200.
JP, JP-A-11-95753, JP, 2000
-99009, JP 2000-99092 A, JP 2000-99093 A, JP 2000-
261322, JP 2001-5450 A,
Japanese Unexamined Patent Application Publication No. 2001-148633 proposes various methods capable of analyzing a frequency as a constituent element of an arbitrary time series signal and creating MIDI data from the analysis result.

In the invention described in the above publication, the frequency analysis of the time series signal has been performed by using the short-time Fourier transform method or the generalized harmonic analysis method. The short-time Fourier transform method has a small calculation load, but has a relatively low frequency resolution, and the generalized harmonic analysis method has a problem that the frequency resolution is high but the calculation load is large. Therefore, the applicant of the present application has proposed in Japanese Patent Application No. 2002-9223 a method that uses a correlation calculation table to have a high frequency resolution and a small calculation load.

[0007]

However, in the method using the above correlation calculation table, the adjacent unit sections are set so as to overlap each other. Therefore, if the overlap region is increased to increase the time resolution, the correlation calculation is performed. There is a problem that the number of times of performing is also increased.

In view of the above points, the present invention eliminates the overlap of the correlation calculation in the overlapping area of the unit section and reduces the total time required for the correlation calculation, and the code of the acoustic signal. An object of the present invention is to provide an activation device.

[0009]

In order to solve the above-mentioned problems, in the present invention, as a frequency analysis device for converting a given time-series signal into time-series spectrum data, a unit section which is a basic unit for analysis is used. , Section signal extraction means for setting adjacent unit sections in time series so as to overlap each other and sequentially extracting time series signals of each unit section, and a previous section signal extracted immediately before by the section signal extraction means. Correlation calculation means at the beginning of the previous section for calculating the correlation with a predetermined harmonic function for the signal located at the beginning of the previous section signal that does not match the newly extracted current section signal, and the previous section signal And a newly-extracted current section signal, the current section tail correlation calculation means for calculating a correlation with a predetermined harmonic function for a signal located at the tail of the current section signal that does not match, and a previous section overall The correlation sum for calculating the correlation value of the entire current section by subtracting the value calculated by the correlation calculation section at the beginning of the preceding section and adding the value calculated by the correlation calculation section at the end of the current section to the function value Means, storage means for storing the correlation value of the previous section for holding the correlation value of the entire current section obtained by the correlation summing means, and a predetermined value based on the correlation value of the entire current section obtained by the correlation summing means It is characterized in that it is configured by a spectrum calculation means for performing conversion and calculating spectrum data corresponding to each unit section.

According to the present invention, a plurality of unit sections are set so that adjacent unit sections in the time series overlap with each other, and the correlation value between each unit section and the harmonic function is calculated to obtain the time series. When performing the frequency analysis of a signal, the correlation value of each unit section is calculated by using the correlation value of the immediately preceding unit section instead of calculating the correlation between all section signals in the unit section and the harmonic function. Since the correlation value of the part of the unit section that does not overlap with the current unit section is calculated and subtracted, and the correlation value of the part that does not overlap with the unit section immediately before the current unit section is calculated and added, the calculation is performed. It is possible to eliminate the overlap of the correlation calculation in the overlapping area of the unit section and reduce the total time required for the correlation calculation.

[0011]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below in detail with reference to the drawings. (1. Basic Principle) First, the basic principle of frequency analysis and audio signal coding according to the present invention will be described. Since this basic principle is disclosed in the above-mentioned publications or specifications, only the outline thereof will be briefly described here.

As shown in FIG. 1A, it is assumed that an analog acoustic signal is given as a time series signal. In the example of FIG. 1, the horizontal axis represents time t, and the vertical axis represents amplitude (intensity) to show this acoustic signal. Here, first, a process of taking in the analog acoustic signal as digital acoustic data is performed. This uses the conventional general PCM method,
The analog acoustic signal may be sampled at a predetermined sampling frequency, and the amplitude may be converted into digital data by using a predetermined number of quantization bits.

Then, a plurality of unit sections are set on the time axis of the acoustic signal to be analyzed. In the example shown in FIG. 1A, six times t1 to t6 are equally spaced on the time axis t.
Is defined, and five unit sections d1 to d5 whose start point and end point are the respective time points are set. In the example of FIG. 1, all unit sections having the same section length are set without overlapping on the time axis, but the section setting is performed so that adjacent unit sections partially overlap on the time axis. It doesn't matter. Particularly, in the present invention, it is an essential requirement that the unit sections overlap.

When the unit section is set in this way, a representative frequency is selected for each acoustic signal (hereinafter referred to as section signal) for each unit section. Each section signal usually contains various frequency components,
For example, a frequency component having a large intensity ratio of the components may be selected as the representative frequency. Here, the representative frequency is generally a so-called fundamental frequency, but a harmonic frequency such as a formant frequency of voice or a peak frequency of a noise sound source may be treated as a representative frequency. Although only one representative frequency may be selected, more accurate encoding becomes possible if a plurality of representative frequencies are selected depending on the acoustic signal. FIG. 1B shows an example in which three representative frequencies are selected for each unit section and one representative frequency is encoded as one representative code (in the figure, it is shown as a note for convenience). Has been done. Here, three tracks T1 for accommodating a representative code (note) are provided.
T2 and T3 are provided so that the three representative codes selected for each unit section are accommodated in different tracks.

For example, the representative codes n (d1,1), n (d1,2), n (d1,) selected for the unit section d1.
3) are housed in the tracks T1, T2, T3, respectively. Here, each code n (d1,1), n (d1,
2) and n (d1,3) are codes indicating note numbers in the MIDI code. The note number in the MIDI code takes 128 values from 0 to 127, and each indicates one key on the keyboard of the piano. Specifically, for example, the representative frequency is 440 Hz
Is selected, the frequency is note number n = 6
Since it corresponds to 9 (corresponding to "Ra sound (A3 sound)" at the center of the keyboard of the piano), n = 69 is selected as the representative code. However, FIG. 1B is a conceptual diagram showing the representative code obtained by the above-described method in the form of a musical note, and in fact, each musical note is also provided with data relating to its strength. For example, the track T1 includes note numbers n (d1,1), n (d2,1) ... Pitch data and e (d1,1), e (d
Data indicating the strength of 2, 1, 1 ... The data indicating this intensity is determined by the degree to which the component of each representative frequency is included in the original section signal. Specifically, the data indicating the intensity is determined based on the correlation value of the section signal of the periodic function having each representative frequency. Also, FIG.
In the conceptual diagram shown in (b), the position of each unit section on the time axis is shown by the position of the note in the lateral direction. The data shown as is added to each note.

It is not always necessary to adopt the MIDI format as the format for encoding the acoustic signal, but since the MIDI format is the most popular as this type of encoding format, the MIDI format code data is practically used. Is preferably used. In the MIDI format, “note on” data or “note off” data exists with “delta time” data interposed. The "Note On" data is
The "note-off" data is data for instructing the start of playing a specific sound by designating a specific note number N and velocity V. The "note-off" data is a data of a specific note designated by a specific note number N and velocity V. This is data for instructing the end of performance. The "delta time" data is data indicating a predetermined time interval. Velocity V is a parameter indicating, for example, the speed at which the piano keyboard is pushed down (velocity at note-on) and the speed at which the finger is released from the keyboard (velocity at note-off), and operation to start playing a specific sound. Alternatively, it indicates the strength of the performance ending operation.

In the above method, the i-th unit section di
About J note numbers n (d
i, 1), n (di, 2), ..., N (di, J) are obtained for each of these intensities e (di, 1), e
(Di, 2), ..., E (di, J) are obtained. Therefore, the code data in the MIDI format can be created by the following method. First, as the note number N described in the “note on” data or the “note off” data, the obtained note number n (d
i, 1), n (di, 2), ..., N (di, J) may be used as they are. On the other hand, as the velocity V described in the “note-on” data or the “note-off” data, the obtained intensities e (di, 1), e (d
i, 2), ..., E (di, J) may be standardized by a predetermined method. The “delta time” data may be set according to the length of each unit section.

(2. Concrete Method for Obtaining Correlation with Periodic Function) In the method based on the above-mentioned basic principle, one or a plurality of representative frequencies are selected for the section signal and have the representative frequency. The section signal is represented by the periodic signal. Here, the selected representative frequency is literally a frequency representing the signal component in the unit section. As a specific method for selecting the representative frequency, there are a method using a short-time Fourier transform, a method using a generalized harmonic analysis, and a method using a correlation calculation table. Among these, the method of using the correlation calculation table is the method proposed by the applicant in Japanese Patent Application No. 2002-9223. Since the present invention is particularly effective when a method using a correlation calculation table is used, a specific method for obtaining a correlation with a periodic function using the correlation calculation table will be described.

It is assumed that a trigonometric function as shown in FIG. 2 is prepared as a plurality of periodic functions. These trigonometric functions are composed of a pair of a sine function and a cosine function having the same frequency, and 128 standard frequencies f (0) to
For each of f (127), a pair of sine and cosine functions will be defined. Here, a pair of functions including a sine function and a cosine function having the same frequency will be defined as a periodic function for the frequency. That is, the periodic function for a specific frequency is composed of a pair of sine function and cosine function. Thus, the reason why the periodic function is defined by a pair of sine function and cosine function is to consider that the correlation value is influenced by the phase when the correlation value of the periodic function with respect to the signal is obtained. The variables F and k in each trigonometric function shown in FIG. 2 are variables corresponding to the sampling frequency F and the sample number k for the interval signal X. For example, a sine wave for the frequency f (0) is represented by sin (2πf (0) k / F), and given an arbitrary sample number k, the same time position as the kth sample forming the interval signal is given. The amplitude value of the periodic function at is obtained. Here, 128 standard frequencies f (0)
~ F (127) is defined by the following [Formula 1].

[Formula 1] f (n) = 440 × 2 ^{γ (n)} γ (n) = (n−69) / 12 where n = 0, 1, 2, ..., 127

Defining the standard frequency by such an equation is convenient when finally performing encoding using MIDI data. This is because there are 128 standard frequencies f (0) to f (12) set by such a definition.
7) is to take frequency values forming a geometric series, and M
This is because the frequency corresponds to the note number used in the IDI data. Therefore, 128 shown in FIG.
The standard frequencies f (0) to f (127) are the frequencies set at equal intervals (semitone unit in MIDI) on the frequency axis shown by the logarithmic scale.

Next, a concrete description will be given of how to obtain the correlation of each periodic function with respect to a section signal of an arbitrary section. For example, as shown in FIG. 3, it is assumed that the section signal X is given to a certain unit section d. Here, it is assumed that the unit section d having the section length L is sampled at the sampling frequency F and w sample values are obtained in total, and the sample number is 0, as shown in the figure. 1,2,3, ..., k, ..., w-
2, w-1 (the w-th sample shown by the white circle is
It shall be the sample included at the beginning of the next unit section adjacent to the right). In this case, for any sample number k,
This means that the amplitude value X (k) is given as digital data. Here, w is also described as a constant in the following description, but in general, it is changed according to the value of n and becomes maximum within a range not exceeding the section length L.
It is desirable to set the value to an integral multiple of (n). Normal,
As shown in FIG. 1, when the analysis is performed in a predetermined unit section, that is, in a short time, the central weight is close to 1 and the weights at both ends are close to 0 for each sample with respect to X (k). The window function W (k) is multiplied, which is the short-time Fourier transform method. However, in the present invention, as will be described later, in the case of analyzing a certain unit section, the correlation value of the unit section immediately before that is not directly calculated, but the basic correlation is calculated. From the value,
The correlation value of the portion peculiar to the immediately preceding unit section is subtracted, and the correlation value of the tail portion of the current unit section that does not overlap is added. For this purpose, it is not preferable to multiply by the window function W (k), which is the reason why the short time Fourier transform method and the generalized harmonic analysis method are not used in the present invention.

The principle of obtaining the correlation value with the sine function Rn having the nth standard frequency f (n) for such a section signal X will be described. The correlation value A (n) between the two can be defined by the following [Formula 2].

[Formula 2] A (n) = (2 / w) Σ _{k = 0, w-1} x (k) sin (2πf _n k / F) B (n) = (2 / w) Σ _{k = 0, w-1} x (k) cos (2πf _n k / F) E (n) = {A (n) ² + B (n) ² } ^1/2

In the above [Formula 2], X (k) is the amplitude value of the sample number k in the interval signal X, as shown in FIG. 3, and sin (2πf _n k / F) is on the time axis. It is the amplitude value of the sine function Rn at the same position of. In order to avoid complicated expressions, f
(N) is expressed as f _n . The first arithmetic expression of [Equation 2] is an expression for obtaining the inner product of the amplitude value of the interval signal X and the amplitude vector of the sine function Rn for the dimensions of all sample numbers k = 0 to w−1 in the unit interval d. Can be said.

Similarly, the second arithmetic expression of the above [Equation 2] is an expression for obtaining the correlation value between the interval signal X and the cosine function having the nth standard frequency f (n). The correlation value of is given by B (n). Note that both the first arithmetic expression for obtaining the correlation value A (n) and the second arithmetic expression for obtaining the correlation value B (n) are finally multiplied by 2 / w. Is for normalizing the correlation value, and since w is generally changed depending on n as described above, this coefficient is also a variable depending on n.

The effective value of the correlation between the interval signal X and the standard periodic function having the standard frequency f (n) is the third value of the above [Formula 2].
As shown in the following equation, of the square root of the sum of squares of the correlation value A (n) with the sine function and the correlation value B (n) with the cosine function,
It can be indicated by a positive value, E (n). If the frequency of the standard periodic function having a large effective value of the correlation is selected as the representative frequency, the interval signal X
Can be encoded.

That is, one or a plurality of standard frequencies whose correlation value E (n) is greater than a predetermined standard may be selected as the representative frequency. The selection condition that “the correlation value E (n) is greater than or equal to a predetermined reference” is, for example, a threshold that is set in advance and the correlation value E (n) exceeds the threshold. Frequency f (n)
May be set as a representative frequency, but an absolute selection condition may be set. For example, relative selection conditions such as selecting up to the Qth in the order of the magnitude of the correlation value E (n). May be set.

(2.1. Method using cross-correlation table)
As a method of calculating the correlation between the section signal and the harmonic function in the set unit section, a short-time Fourier transform method,
The method using generalized harmonic analysis is famous. But,
The short-time Fourier transform method does not have sufficient frequency resolution,
In the method using generalized harmonic analysis to solve the problems of the short-time Fourier transform method, the number of correlation operations with the harmonic function, which is a periodic function, is orders of magnitude higher than that of the short-time Fourier transform method. Therefore, there is a problem that the calculation load is large. Therefore, the present applicant has filed Japanese Patent Application No. 2002-9223.
In this issue, we proposed a method for frequency analysis using a cross-correlation table. With this method, it is possible to realize a frequency resolution equivalent to that of generalized harmonic analysis with a computational load equivalent to that of the short-time Fourier transform method, and the extracted signal components that have been a problem in generalized harmonic analysis. It is possible to improve the accuracy of. A method using this cross-correlation table will be described below.

First, as described above, a plurality of standard frequencies are set, and a standard periodic function corresponding to each standard frequency is prepared as a harmonic function. The standard frequency set at this time can be arbitrarily set according to the characteristics of the frequency analysis, but in order to use it for encoding the acoustic signal, as shown in FIG. 2 and [Equation 1], , MIDI standard note number n is preferably set.

Then, the cross-correlation, which is the correlation between the harmonic functions, is calculated for all the combinations, and the cross-correlation table is created. At this time, the cross-correlation R (f _m , f _n ) of the harmonic function of the frequency f (m) with respect to the harmonic function of the frequency f ( _n )
Is calculated by the following [Formula 3].

[Formula 3] A (f _m , f _n ) = (2 / T (n)) Σ _{k = 0, T (n) -1} sin (2πf _m k
_{/ F) sin (2πf n k} / F) B (f m, f n) = (2 / T (n)) Σ k = 0, T (n) -1 sin (2πf m k
/ F) cos (2πf _n k / F) R (f _m , f _n ) = {A (f _m , f _n ) ² + B (f _m , f _n ) ² } ^1/2

The cross-correlation R (f _m , f _n ) calculated by the third expression of the above [Expression 3] represents one element of the two-dimensional cross-correlation table. As shown in FIG. 2, when m and n correspond to note numbers, the correlation value of 128 note numbers corresponding to each note number m is recorded in the cross-correlation table, which is 128 × 128 in total. Correlation values will be recorded.

When the cross-correlation table is prepared, a unit section is set over the entire section of the time series signal to be analyzed, and the time series signal of the set unit section is extracted as a section signal. Differently from the case shown in FIG. 1A, the unit sections are set such that adjacent unit sections overlap each other.

Subsequently, the correlation calculation with the total harmonic function is performed on the extracted section signal. For example, when the standard frequency is set corresponding to the note number as shown in FIG. 2, correlation calculation with 128 harmonic functions is performed. The correlation calculation with the harmonic function at this stage is performed by the following [Formula 4].
Done by. That is, the correlation between the harmonic function and the portion of the interval signal from the beginning that is an integral multiple of the cycle of the harmonic function for which correlation calculation is performed and does not exceed the unit interval length is calculated. The calculated correlation value is stored in the signal correlation array prepared for each unit section. Here, the correlation calculation with each harmonic function is performed for one section signal is
Since it is only once, it contributes to suppressing the number of correlation calculations. The correlation value E (f _n ) between the harmonic function of the standard frequency f (n) and the interval signal x (k) at this stage is calculated by the following [Formula 4].

[Formula 4] A (f _n ) = (2 / T (n)) Σ _{k = 0, T (n) -1} x (k) sin (2πf
_n k / F) B (f _n ) = (2 / T (n)) Σ _{k = 0, T (n) -1} x (k) cos (2πf
_n k / F) E (f _n ) = {A (f _n ) ² + B (f _n ) ² } ^1/2

This [Equation 4] is substantially the same as the above, except that the correlation calculation sample number w in [Equation 2] is replaced by the correlation calculation time T (n).

When the signal correlation array is obtained, the correlation value which is each element in the array is corrected using the cross correlation table. Specifically, the correlation value E with the standard frequency f (n)
correction value _{(f n) E'(f n} ) , the correlation value between the standard frequency _{f (m) E (f m} ), the standard frequency f of the normal frequency f (m)
The cross-correlation R for _{(n) (f m, f} n), the standard frequency f
It is calculated by the following [Equation 5] using the autocorrelation R (f _m , f _m ) of (m).

[Equation 5] E ′ (f _n ) = E (f _n ) −Σ _{m = 0, N−1} E (f _m ) R (f _m , f _n ).
_{/ R (f m, f m} )

Correction value E calculated by the above [Formula 5]
'(F _n) is stored in a position corresponding to the normal frequency f in the correlation sequence (n), and later is used to calculate the correlation value E (f _m) as another correction value E'(f _n) It In this way
A correction value E ′ (f _n ) corresponding to all set standard frequencies is calculated. At this time, which correlation value E (f _n ) is corrected from n = 0 to N−1 basically follows the order of the magnitude of the initial value of the correlation value in the signal correlation array. . In this way, a signal correlation array in which N correlation values are corrected is obtained. However, at this point, some of the elements in the array may have negative values. In that case, the value is set to 0 so that all the values in the signal correlation array become 0 or a positive value, and this is set as the correction correlation array. The reason why the value of the corrected correlation array is set to 0 or more in this way is to delete an unrealistic value, since a negative correlation value is basically impossible. Also,
The process of setting a negative value element to 0 is performed after all the elements in the signal correlation array are corrected, when the correction value E ′ (f _n ) is negative. ′ (F _n ) is [Formula 5]
This is because it is used for the calculation of other correction values as E (f _m ). As a result, if the correction value is negative,
The sum total of Σ on the right side of [Equation 5] decreases, and as a result, the correlation value E (f _n ) before correction increases. In the present invention, even if the correction value is negative in this way,
Since the correction values of other elements are obtained by directly using the values without changing them, the difference signal changes depending on the order of the included signals to be subtracted, and the obtained correlation value is different, as in generalized harmonic analysis. There is no. Therefore, the correction can be performed without depending on the order of the magnitude of the initial value of the correlation value in the signal correlation array.

By performing the above-mentioned correlation calculation and correlation correction for all the set unit intervals, N frequency components in all the unit intervals can be obtained.

Here, the effect of correction of the correlation value using the cross-correlation table will be conceptually described with reference to FIG. In FIG. 4, the horizontal axis corresponds to the note number (frequency), and the vertical axis corresponds to the signal strength or the correlation strength.
Now, consider the case of performing frequency analysis of a single sound source. The original signal spectrum of a single-tone sound source is shown in FIG.
It is represented by one frequency as shown in. When frequency analysis of this single sound is performed, as shown in FIG.
If only one frequency is extracted, the most accurate frequency analysis has been performed. However, when frequency analysis by short-time Fourier analysis is performed on this single sound, the result shown in FIG.
As shown in (b), many frequency components will be extracted. Further, when frequency analysis by generalized harmonic analysis is performed on the single tone shown in FIG. 4A, a large number of frequency components are extracted as shown in FIG. 4C. However, Figure 4
As can be seen from a comparison between (b) and FIG. 4 (c), using the generalized harmonic analysis makes the correlation strength of the frequency to be extracted larger than the other frequencies, rather than using the short-time Fourier transform. Is also more accurate.

In this method, a correction value is obtained for each note number using a cross-correlation table prepared in advance. This correction value is shown in FIG. Figure 4 (d)
The correction value shown in (4) corresponds to -Σ and after on the right side of the above [Formula 5]. The correction values shown in FIG.
By correcting the correlation strength shown in FIG.
A single frequency component to be the target as shown in (e) is extracted.

By the above processing, a frequency group (spectral data), which is a set of intensity values for each frequency, is obtained for each unit section. When a predetermined number of frequency groups are selected in this way, "information indicating the pitch of the sound" corresponding to each frequency of this frequency group, "sound intensity corresponding to the signal strength of each selected frequency""Informationindicating","information indicating the sound production start time" corresponding to the start point of the unit section, and "information indicating sound production end time" corresponding to the start point of the unit section subsequent to the unit section. If code data including information (which will be referred to as phoneme data) is created, the section signal X in the unit section can be coded by a predetermined number of code data. If MIDI data is created as code data, note number is used as "information indicating pitch of tone", velocity is used as "information indicating intensity of tone", and "start time of sound generation" The note-on time may be used as the “information indicating” and the note-off time may be used as the “information indicating the sound production end time”.

(3.1. Frequency Analysis Apparatus and Acoustic Signal Coding Apparatus According to the Present Invention) The frequency analysis apparatus and acoustic signal coding apparatus according to the present invention will be described below. FIG. 5 is a flowchart showing an outline of frequency analysis according to the present invention. First, set multiple standard frequencies,
A standard periodic function corresponding to each standard frequency is prepared as a harmonic function (step S1). The standard frequency set at this time can be arbitrarily set according to the characteristics of the frequency analysis, but in order to use it for encoding the acoustic signal, as shown in FIG. 2 and [Equation 1], , MIDI
It is preferable to set in correspondence with the standard note number n.

Subsequently, a unit section is set over the entire section of the time series signal to be analyzed, and the time series signal of the set unit section is extracted as a section signal (step S2).
The setting of the unit section is different from that shown in FIG.
The unit sections adjacent to each other are overlapped with each other.

Then, the correlation calculation with the total harmonic function is performed on the extracted section signal (step S3). For example,
When the standard frequency is set corresponding to the note number as shown in FIG. 2, correlation calculation with 128 harmonic functions is performed. The correlation calculation with the harmonic function in step S3 is performed by the method using the above [Formula 4].
That is, the correlation between the harmonic function and the portion of the interval signal from the beginning that is an integral multiple of the cycle of the harmonic function for which correlation calculation is performed and does not exceed the unit interval length is calculated. However, the correlation calculation method is different between the first unit section and the second and subsequent unit sections. For the first unit section, the correlation calculation is performed over the entire unit section according to the above [Formula 4], but in the second and subsequent unit sections, the correlation value of the overlapping portion is calculated from the correlation value of the immediately preceding unit section that has already been calculated. Is subtracted, and only the correlation value of a portion that does not overlap with the immediately preceding unit section in the unit section is newly calculated and added.

The concept of calculation of the target correlation value of the current unit section using the correlation value of the immediately preceding unit section will be described with reference to FIG. Here, the immediately preceding unit section is the previous section,
The target current unit section is called a current section or a subsequent section. In FIG. 6, FIG. 6A shows the waveform of the acoustic signal, and FIG. 6B shows the unit section d _i and the unit section d of the fixed length L set in the time-series acoustic signal.
It is a figure which shows the mode of _{i + 1} . Since in FIG. 6 (b), D is the difference between the start time and the unit interval d _{i + 1} of the start time of the unit section d _i, all the unit interval is a fixed length,
The difference between the end time of the unit section d _{i and} the end time of the unit section d _{i + 1} is also D. In this specification, this temporal length D will be referred to as an update interval. As shown in FIG. 6, when the correlation value of the unit section d _{i + 1} of the fixed length L is calculated, the correlation value of the unit section d _i of the fixed length L for which the calculation has already been completed is used.
Specifically, the correlation value between the harmonic function and the interval signal in the head area dh does not overlap with the unit section d _{i + 1} in the unit section d _i, the unit section d _i of the unit section in d _{i + 1} The correlation values of the section signals and the harmonic functions in the non-overlapping tail region db are calculated, the correlation value of the head region dh is subtracted from the correlation value of the unit segment d _i , and the correlation value of the tail region db is added. As a result, it is not necessary to calculate the correlation value of the overlapping region of the unit section d _i and the unit section d _{i + 1} , and the total load of the correlation calculation for the entire time series signal is reduced.

Here, the reduction of the calculation load will be specifically described by using the calculation formula for obtaining the correlation value. Correlation value En for frequency f (n) in the i-th unit section
(I) can be expressed as the following [Equation 6] based on the above [Equation 2]. The correlation value is calculated by the number of prepared harmonic functions. For example, if 128 harmonic functions are prepared as shown in FIG.
Although the correlation value is calculated individually, here, the correlation value En (i) at a certain standard frequency is shown as a representative.

[Equation 6] En (i) = {((2 / L) Σ _{k = 1, L} x (k) sin (2πf _n
k / F)) ² + ((2 / L) Σ _{k = 1, L} x (k) cos (2πf _n k
/ F)) ² } ^1/2

In the above [Formula 6], for convenience of explanation, sample numbers 0, 1, 2, 3, ...
.., k, ..., w-2, w-1 to 1, 2, 3, ...
.., L-1 and L are replaced. That is, in order to make it easy to understand the reduction amount of the calculation load, the number of sample points is set to the same value as the fixed length L. And L is the period F / f
It is given as an integer multiple of (n), and the frequency f is actually
It changes depending on (n).

Here, in order to make it easy to understand the change in the calculation amount, in the above [Equation 6], it is the sum of the product of the sine function of the harmonic function and the amplitude value at each sample point of the interval signal, "(2 / L) Σ _{k = 1, L} x (k) sin (2πf _n k /
F) ”is Sn (i, 1, L),“ (2 / L) Σ _{k = 1, L} x (k)
Replace “cos (2πf _n k / F)” with Cn (i, 1, L). Thereby, [Formula 6] can be transformed into [Formula 7] below.

[Equation 7] En (i) = {Sn (i, 1, L) ² + Cn (i, 1,)
L) ² } ^1/2

In the above [Formula 7], Sn (i, 1,
L) is the correlation between the sine function of the harmonic function and the interval signal calculated from sample point 1 to sample point L, and Cn
(I, 1, L) is the correlation between the cosine function of the harmonic function and the interval signal calculated from sample point 1 to sample point L. That is, Sn (i, 1, L), Cn (i, 1, L)
L) Both are calculating for L points. On the other hand, the correlation value En (i + 1) of the unit section d _{i + 1} can be expressed as the following [Equation 8].

[Equation 8] En (i + 1) = {Sn (i + 1,1, L) ² + Cn
(I + 1,1, L) ² } ^1/2

Since the unit section d _{i + 1} has the same section length L as that of the unit section d _i , the sample points are the sample point 1 at the beginning of the unit section and the sample point L at the end. If this equation is rewritten in detail, the following [Equation 9] is obtained.

[Equation 9] En (i + 1) = {((2 / L) Σ_{k = 1, L}x (k + D) sin
(2πf_n(k + D) / F)) ²+ ((2 / L) Σ_{k = 1, L}x (k +
D) cos (2πf_n(k + D) / F))²}^1/2

In [Equation 9], "k" in [Equation 6] is replaced with "k".
It is replaced with "+ D". This is because the unit section d _i and the unit section d _{i + 1} actually have different sampling points by the update section D, so that the value of the section signal x also changes accordingly. Regarding the harmonic function for performing the correlation calculation with the interval signal, it is usual not to add the offset of the update interval D, but in this embodiment, L is set to an integral multiple of the period F / f (n) as described above. Therefore, theoretically, En (i +
The value of 1) does not change. In the present application, because of the configuration of the invention, it is necessary to change the phase in accordance with the update section of the unit section. Therefore, the offset of the update section D is added and sin (2πf _n
k / F) and cos (2πf _n k / F) are sin (2πf _n )
_n (k + D) / F) and cos (2πf _n (k + D) / F). Then, [Formula 9] can be rewritten as [Formula 10] below.

[Formula 10] En (i + 1) = [{(2 / L) Σ_{k = 1, L}x (k) sin (2π
f_nk / F)-(2 / D) Σ _{k = 1, D}x (k) sin (2πf_nk /
F) + (2 / D) Σ_{k = L-D + 1, L}x (k + D) sin (2πf _n(k
+ D) / F)}²+ {(2 / L) Σ_{k = 1, L}x (k) cos (2πf_n
k / F)-(2 / D) Σ_{k = 1, D}x (k) cos (2πf_nk / F)
+ (2 / D) Σ_{k = L-D + 1, L}x (k + D) cos (2πf_n(k +
D) / F)}²]^1/2

The conversion from [Equation 9] into [Equation 10] is
The sample points 1 to L assigned with the head of the unit section d _{i + 1} as a reference are replaced with the head of the unit section d _i as a reference, and the correlation value in the tail region db of the unit section d _i is phase D Show that you are just staggering. Further, [Formula 10] can be rewritten as shown in [Formula 11] below, similarly to the conversion from [Formula 6] to [Formula 7].

[Equation 11] En (i + 1) = [{Sn (i, 1, L) -Sn (i,
1, D) + Sn (i + 1, L-D + 1, L)} ² + {C
n (i, 1, L) -Cn (i, 1, D) + Cn (i +
1, L-D + 1, L)} ² ] ^1/2

In the above [Formula 11], Sn (i, 1,
L) and Cn (i, 1, L) have already been obtained when calculating [Equation 7], and therefore need not be calculated. Sn (i, 1, D) is newly calculated. ), S
n (i + 1, L-D + 1, L), Cn (i, 1, D),
It becomes Cn (i + 1, L-D + 1, L). Sn (i,
1, L) and Cn (i, 1, L) multiply the interval signal at the L sample points from sample point 1 to sample point L by the amplitude value of the harmonic function, but Sn (i, 1) ，
D) and Cn (i, 1, D) are D sample points from sample point 1 to sample point D, and Sn (i + 1, L-D +).
1, L) and Cn (i + 1, L-D + 1, L), the section signals at the D sample points from the sample point L-D + 1 to the sample point L are also multiplied by the amplitude value of the harmonic function.

As can be seen from this, the unit section d _{i + 1}
When the calculation is performed as it is, the calculation at L sample points must be performed. However, when the correlation value obtained in the immediately preceding unit section d _i is used, 2D
It suffices to perform the calculation at each sample point.
That is, the calculation amount is 2D / L. This means that the larger the overlapping area of the adjacent unit sections in order to improve the time resolution, that is, the smaller the update section D, the more the amount of calculation required can be significantly reduced. Is shown.

The correlation value calculated for each unit section is stored in the signal correlation array prepared for each unit section. Similarly, perform correlation calculation for all unit intervals, and when the correlation value for each unit interval is obtained, correct the value of the signal correlation array using the cross-correlation table as described in section 2.1. Yes (step S4).

By performing the correlation calculation in step S3 and the correlation correction in step S4 for all the unit intervals set in step S3, N frequency components in all the unit intervals can be obtained. That is, the spectrum data of the frequencies in all the unit sections can be obtained.

(3.2. Coding of acoustic signal) The frequency analysis of the time-series signal is performed as described above, and N contained signals are extracted for each unit section. When an acoustic signal is adopted as the time-series signal and the acoustic signal is encoded, the standard frequency f (n) is set at MIDI note numbers, that is, pitch intervals in semitone units, as shown in FIG. Then, N (= 128) frequency components corresponding to each note number are obtained. Then, as described above, by performing conversion into MIDI data in which the frequency of the frequency component is the note number, the correlation value is the velocity, the start point of the unit section is the note-on time, and the start point of the following unit section is the note-off time, The audio signal is encoded.

(3.3. Regarding Frequency Setting) In the above embodiment, the frequency to be extracted is set as the standard frequency corresponding to the MIDI standard note number n as in [Equation 1]. If the intervals are not set finer, highly accurate detection cannot be performed. The reason is that the frequency setting is a logarithmic interval proportional to the note number, as in the above [Formula 1]. Therefore, the higher the frequency, the coarser the calculation interval,
This is because the probability of missing a frequency component increases.
When an acoustic signal is analyzed as a time-series signal and encoded, it is necessary to set at intervals corresponding to note numbers as in [Equation 1]. Then, it is desirable that the fine frequencies set between the adjacent standard frequencies have logarithmic intervals as in [Equation 1]. For example, 12 frequencies between each note number, 1 for each note number
/ 13 interval is set. That is, the frequencies are set at geometric intervals between the standard frequencies. Specifically, the standard frequency f (n) and the standard frequency f
The frequency f (n + m / M) is set between (n + 1). Here, M is an integer of 0 or more, m is 0
It is an integer that takes the value of M-1. In this case, if frequencies are set at geometrical intervals, each frequency is expressed by the following [Equation 12].

[Equation 12] f (n) = 440 × 2 ^{(n-69) / 12} f (n + m / M) = 440 × 2 ^{(n + m / M-69) / 12} f (n + 1) = 440 × 2 ^{(n + 1-69) / 12} However, n = 0,1,2, ..., 127, m = 0,1,
2, ..., M-1

For example, as described above, twelve frequencies are provided between note numbers, each of which is 1/1 of the note number.
The setting of 3 intervals corresponds to the case where M = 13 in [Equation 12]. When such a setting is made, the frequency is set to 128 × 13 in total, and the calculation accuracy of the correlation value is improved. Since there is no frequency corresponding to such a fine frequency setting when encoding into an audio signal, the maximum correlation value is selected and given as an intensity (velocity) component corresponding to the closest note number.

However, if the frequencies are set at fine intervals as described above, it goes without saying that the calculation load for the correlation calculation becomes enormous. Then, the correlation values corresponding to the individual fine frequencies calculated by spending the calculation time are selected, and most of them are not used. That is, the magnitude relation of the correlation values corresponding to each fine frequency is important, but absolute accuracy is not required. Therefore, in the present invention, the frequencies are actually set at fine intervals between the standard frequencies, and the correlation calculation between each set frequency and the section signal is not performed one by one, but each frequency is set based on the correlation value of the adjacent standard frequency. A process of estimating a correlation value corresponding to a fine frequency between standard frequencies and searching for a correlation value of a maximum fine frequency corresponding to each standard frequency is performed. For example, frequency f
(N + m / M) correlation value E _{n + m / M} in is calculated by the following [Equation 13].

[Equation 13] E _{n + m / M} = {((2 / L) Σx (k) sin (2πf _{n + m / M} k /
F)) ² + ((2 / L) Σx (k) cos (2πf _{n + m / M} k /
F)) ² } ^1/2 = {S _{n + m / M} ² + C _{n + m / M} ² } ^1/2

In order to avoid complicated formulas,
In this [Formula 13] and the following [Formula 14], f
(N) is f _n , f (n + m / M) is f _{n + m / M} , f (n +
1) is expressed as f _{n + 1} . Here, the standard frequency f
Linearly approximating a fine frequency harmonic function sin (2πf _{n + m / M} k / F) between (n) and f (n + 1) based on two adjacent standard frequency harmonic functions, that is, sin (2
πf _{n + m / M} k / F) ≈sin (2πf _n k / F) + {sin (2πf
_{If n + 1} k / F) -sin (2πf _n k / F)} m / M is set, the above [Formula 13] is transformed into the following [Formula 14].

[Equation 14] E _{n + m / M} = [{S _n + (S _{n + 1} −S _n ) m / M} ² + {C _n
+ (C _{n + 1} −C _n ) m / M} ² ] ^1/2

This means that, when frequencies are set at geometric series intervals between standard frequencies, correlation calculation for the set number must be performed, whereas correlation calculation is performed only for standard frequencies. It means that the addition and subtraction using the calculated correlation value should be performed. Compared with the load of correlation calculation, the load of addition and subtraction is almost negligible.
The total calculation load for calculating the correlation value is about 1/13 as compared with the case where the correlation calculation is performed by setting two frequencies.

(4. Device Configuration) Next, the device configurations of the frequency analysis device and the acoustic signal coding device according to the present invention will be described. FIG. 7 is a functional block diagram of an audio signal encoding apparatus for converting to a MIDI code according to the present invention. In FIG. 7, 1 is an acoustic signal input means, 2 is a frequency analysis device, 3 is a phoneme connection means, and 4 is a MIDI code conversion means.

The acoustic signal input means 1 is for inputting an acoustic signal which is a time series signal. When the device shown in FIG. 7 is used as a frequency analysis device, not only acoustic signals but also time series signals can be input. The frequency analysis device 2 executes the processing of steps S2 to S4, and has a function of performing frequency analysis of a time-series signal to extract frequency components (frequency and correlation values, that is, spectrum data) for each unit section. Have.

The phoneme connecting means 3 obtains phoneme data consisting of four pieces of information, which are the start time of a unit section, the start time of the following unit section, the frequency, and the intensity value, which are obtained as a result of frequency analysis of the acoustic signal. Based on the similarity of adjacent phoneme data, the phoneme data is connected to each other to form connected phoneme data. The MIDI code conversion means 4 converts the encoded code data into a MIDI format, and generates a note-on event at the sounding start time and a note-off event at the sounding end time of the concatenated phoneme data as described above. It has the function of setting the note number corresponding to the frequency and the velocity corresponding to the intensity value when a note-on event occurs.

Here, the details of the frequency analysis device 2 will be described. In the present invention, the frequency analysis device 2 has three configuration patterns, which will be described respectively. First, the first configuration pattern has a configuration as shown in FIG. The section signal extraction means 11 has a function of performing the process of step S2. In step S3, the correlation calculating means 12 at the beginning of the preceding section has a function of calculating the correlation between the leading area dh of the preceding preceding unit section and the section signal when calculating the target correlation value of the current unit section. Have The correlation calculating means 13 of the tail of the rear section has a function of calculating the correlation between the tail region db of the current unit section and the section signal when calculating the correlation value of the target current unit section in step S3. ing. The previous section correlation value storage unit 14 is a storage unit for storing the previous section correlation value. In step S3, the correlation summing means 15 extracts the correlation value of the previous section stored in the storage section 14 of the previous section correlation value when calculating the correlation value of the target current unit section, and this correlation value On the other hand, it has a function of subtracting the correlation value calculated by the correlation calculating means 12 at the beginning of the preceding section and adding the correlation value calculated by the correlation calculating means 13 at the end of the succeeding section. The spectrum calculation means 16 is
It has a function of outputting spectrum data by using the cross-correlation table.

The second configuration pattern has a configuration as shown in FIG. 8 (b). In FIG.
For those having the same function as (a), the same reference numerals are given and the description thereof is omitted, and only those different from the first configuration pattern will be described. The correlation calculating means 21 with the basic harmonic function has a function of calculating the correlation between the harmonic function of the standard frequency f (n) and the section signal, where f (n) is a certain standard frequency. There is. The correlation calculating means 22 with the difference harmonic function determines a certain standard frequency as f (n).
, The difference signal sin (2πf _{n + 1} ) between the adjacent harmonic function of the standard frequency f (n + 1) and the adjacent harmonic function of f (n).
k / F) -sin (2πf _n k / F) and cos (2πf _{n + 1} k /
It has a function of calculating the correlation between F) -cos (2πf _n k / F) and the interval signal. The correlation adjusting means 23 uses the standard frequency f according to the above [Formula 13] and [Formula 14]
The correlation value corresponding to the frequency existing between (n) and the standard frequency f (n + 1) is the correlation value with the harmonic function of the standard frequency f (n) and the harmonic function of the standard frequency f (n + 1) and f.
It has a function of approximately calculating using the correlation value between the harmonic function of (n) and the difference signal.

The third configuration pattern has a configuration as shown in FIG. The third configuration pattern is a combination of the first configuration pattern and the second configuration pattern. In FIG. 9 as well, those having the same functions as those in FIGS. 8A and 8B are denoted by the same reference numerals and description thereof is omitted, and those different from the first and second configuration patterns explain. In FIG. 9, the correlation calculating means 12a at the beginning of the preceding section, the correlation calculating means 13a at the tail of the following section, the storing means 14a of the correlation value of the preceding section, and the correlation summing means 15a are the same as those of the basic harmonic function shown in FIG. The processing of the correlation calculation unit 21 is executed by the correlation calculation unit 12 at the beginning of the preceding section, the correlation calculation unit 13 at the end of the succeeding section, the storage unit 14 of the correlation value of the preceding section, and the correlation summing unit 15 shown in FIG. 8A. This is for the purpose of the correlation calculation means 12b at the beginning of the preceding section, the correlation calculation means 13b at the tail of the following section, and the storage section 14 for the correlation value of the preceding section.
b, the correlation summing means 15b performs the processing of the correlation calculation means 22 with the differential harmonic function shown in FIG. 8B by the correlation calculation means 12 at the beginning of the preceding section and the processing at the tail of the succeeding section shown in FIG. 8A. This is to be executed by the correlation calculation unit 13, the previous section correlation value storage unit 14, and the correlation summation unit 15.

The frequency analysis device and the acoustic signal coding device shown in FIGS. 7 to 9 are actually realized by installing dedicated software in an arithmetic processing device such as a computer. Specifically, a program for executing the steps shown in the flowchart of FIG. 5 in the above procedure is installed in the computer. Then, after time-series signals such as acoustic signals are digitized by the PCM method or the like, they are taken into an acoustic signal encoding device realized by a computer, the processes of steps S2 to S4 are performed, and then the extracted spectrum data is output. To do. The acoustic signal encoding device further converts the encoded data into MIDI format data and outputs the encoded data. If the output code data is MIDI data, for example, a MIDI sequencer, M
It is reproduced as an acoustic signal using an IDI sound source.

[0082]

As described above, according to the present invention,
As a frequency analysis device for converting a given time-series signal into time-series spectrum data, a unit section that is the basic unit for analysis is set so that adjacent unit sections on the time series overlap each other, and each unit Section signal extracting means for sequentially extracting time-series signals of a section, and a leading portion of the previous section signal that does not match between the previous section signal extracted immediately before by the section signal extracting section and the newly extracted current section signal For the signal located at, the correlation calculation means at the beginning of the previous section for calculating the correlation with the predetermined harmonic function, and the current section signal that does not match between the previous section signal and the newly extracted current section signal For the signal located at the tail, the correlation calculating means for calculating the correlation with a predetermined harmonic function, and the correlation value for the entire previous interval, the value calculated by the correlation calculating means for the beginning of the previous interval Reduced Then, the correlation summing means for calculating the correlation value of the entire current section by adding the values calculated by the correlation calculation means at the end of the current section, and the correlation value of the whole current section obtained by the correlation summing means are held. And a spectrum calculating means for performing a predetermined conversion based on the correlation value of the entire current section obtained by the correlation summing means, and calculating spectrum data corresponding to each unit section. Thus, there is an effect that it is possible to eliminate the overlap of the correlation calculation in the overlapping region of the unit section and reduce the total time required for the correlation calculation.

[Brief description of drawings]

FIG. 1 is a diagram showing a basic principle of a frequency analysis device and an acoustic signal encoding device according to the present invention.

FIG. 2 is a diagram showing an example of a periodic function used in the present invention.

FIG. 3 is a diagram showing a method of calculating a correlation between a signal to be analyzed and a periodic signal.

FIG. 4 is a conceptual diagram when the effect of using the correlation correction table is compared with other methods.

FIG. 5 is a flowchart showing processing operations of the frequency analysis device and the acoustic signal encoding device according to the present invention.

FIG. 6 is a diagram showing a state in which adjacent unit sections are set to overlap each other with respect to a time-series signal.

FIG. 7 is a diagram showing a device configuration of an audio signal encoding device according to the present invention.

FIG. 8 is a diagram showing first and second configuration patterns of the frequency analysis device according to the present invention.

FIG. 9 is a diagram showing a third configuration pattern of the frequency analysis device according to the present invention.

[Explanation of symbols]

d, d1 to d5, d _i , d _{i + 1} ... Unit section D ... Update section L ... Unit section length n ... Note number En (i), En (i + 1), En _{+ m / M:} Correlation value X, X (k): Section signal

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁷ Identification code FI theme code (reference) H03M 7/30 G10L 7/02 A F term (reference) 5D045 AC10 DA20 5D082 BB01 5D108 BA39 5D378 MM41 5J064 AA02 AA03 BA16 BC01 BC27 BD02 BD03

Claims (6)

[Claims] 1. A frequency analysis device for converting a given time-series signal into time-series spectrum data, wherein unit sections that are basic units for performing analysis overlap adjacent unit sections on the time series. Thus, the section signal extracting means for sequentially extracting the time-series signal of each unit section, and the previous section signal extracted immediately before by the section signal extracting means and the newly extracted current section signal are matched. The correlation calculation means at the beginning of the previous section for calculating the correlation with the predetermined harmonic function for the signal located at the beginning of the previous section signal, and the previous section signal and the newly extracted current section signal For the signal located at the tail of the current section signal that does not match, the correlation calculating means of the tail of the current section for calculating the correlation with a predetermined harmonic function, and the correlation value of the entire previous section, Correlation calculation means Correlation summing means for calculating the correlation value of the entire current section by adding the values calculated by the correlation calculation means at the end of the current section, and the entire current section obtained by the correlation summing means A storage unit for storing the correlation value of the previous section for holding the correlation value of, and a predetermined conversion is performed based on the correlation value of the entire current section obtained by the correlation summing unit, and spectrum data corresponding to each unit section is calculated. And a spectrum calculating means for performing the frequency analysis. 2. A frequency analysis device for converting a given time-series signal into time-series spectrum data, wherein unit sections which are basic units for performing analysis overlap adjacent unit sections on the time series. Set as such, the interval signal extraction means for sequentially extracting the time-series signal of each unit interval, and the basic harmonic function for calculating the correlation between the interval signal extracted by the interval signal extraction means and a predetermined basic harmonic function Correlation calculation means, the section signal, the correlation calculation means of the difference harmonic function to calculate the correlation with the difference function of the other harmonic function frequency adjacent to the predetermined basic harmonic function, of the basic harmonic function The correlation value calculated by the correlation calculation unit is multiplied by the correlation value obtained by the correlation calculation unit with the difference harmonic function and added, and the weight is changed at regular intervals to obtain one value. Correlation adjusting means for calculating a plurality of correlation values in which the frequency slightly changes with respect to the frequency of the fundamental harmonic function, and based on the correlation value obtained by the correlation adjusting means, a predetermined conversion is performed, and each unit interval And a spectrum calculation means for calculating spectrum data corresponding to the frequency analysis device. 3. The correlation calculating means with the basic harmonic function and the correlation calculating means with the differential harmonic function are respectively the correlation calculating means at the beginning of the preceding section, the correlation calculating means at the tail of the current section, and the correlation. The frequency analysis device according to claim 2, wherein the frequency analysis device is configured by a summing unit and a storage unit of the previous section correlation value. 4. The harmonic function, the fundamental harmonic function and the differential harmonic function are composed of a sine wave function and a cosine wave function, and each of the correlation calculating means is composed of a two-dimensional vector composed of a sine wave component and a cosine wave component. A correlation value is output, the correlation summing means and the correlation adjusting means perform vector addition and subtraction on each of the correlation values to output a correlation value composed of a two-dimensional vector, and the spectrum calculation means is composed of a two-dimensional vector. 2. A process of converting a correlation value according to the above into spectral data composed of a scalar value is performed.
Alternatively, the frequency analysis device according to claim 2. 5. The spectrum calculating means is provided with spectrum data for each harmonic function as a cross-correlation table, and the calculated spectrum data is corrected based on the cross-correlation table. The frequency analysis device according to claim 1 or 2. 6. An acoustic signal input means for inputting a given acoustic signal to obtain a time-series signal, a frequency analysis device having the configuration according to any one of claims 1 to 5, and A phoneme linking unit that creates phoneme data by connecting the spectrum data of the series in time series, and an MI that converts the obtained phoneme data into a MIDI format code.
An audio signal encoding apparatus, comprising: DI code conversion means.