JPH05265490A - Pitch period extracting device - Google Patents

Abstract

PURPOSE:To provide the pitch period extracting device which can extract a pitch period at a high speed although the constitution is inexpensive and simple. CONSTITUTION:This device consists of an arithmetic means 12 which extracts a specific period of a digital musical sound signal obtained by sampling an analog musical sound signal at specific time intervals and finds and cumulates differences between digital musical sound signals at a distance of each time difference in the specific section, a minimum peak detecting means 13 which detects a minimum peak position according to values corresponding to the respective time differences cumulated and added by the arithmetic means 12, and a pitch period arithmetic means 14 which calculates the pitch period from information showing the minimum peak position detected by the minimum peak detecting means 13.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pitch period extracting device for extracting a pitch period from, for example, a musical sound or a voice.

In recent years, an electronic musical instrument has been developed which has a function of extracting a pitch period from a raw musical sound or vibration and sending it to a digital sound source or outputting it as an MIDI signal to an external device.

An autocorrelation method using an autocorrelation function is known as a pitch cycle extraction method for such a musical sound or vibration, but in this method, a large amount of multiplication and addition are required, and the pitch cycle is extracted. Since it takes time, improvement is desired.

[0004]

2. Description of the Related Art Conventionally, in the auto-correlation method, an input analog musical tone waveform is digitized, and a pre-processed signal is divided into appropriate intervals, and the time difference is used as a variable within the interval to determine the self-timer for each time difference. Compute the correlation function. Then, the time difference when the calculated autocorrelation function becomes maximum is obtained, and this time difference is used as the pitch period, and the reciprocal of this pitch period is detected as the pitch frequency.

An example of the pitch frequency detecting method using the autocorrelation method is disclosed in, for example, Japanese Patent Laid-Open No. 62-2700.
This is described in detail in Japanese Patent Publication No. 00.

Here, the autocorrelation function φ (d) is given by the following equation. In the above equation, d is a time difference, n is a sample number in the analysis section, and X (n) and X (n + d) are sample values of the input waveform (values of digital musical tone signals).

The autocorrelation function φ (d) is N within a predetermined section.
Number of sampled digital tone signals X (n) (n
= 1, ..., N), the musical tone signal X (n) and a digital musical tone signal obtained by shifting the sample number from the musical tone signal X (n) by d, in other words, a digital musical tone signal X (n + d) with a time difference d. ) And the cumulative sum of products.

A pitch period extraction device for extracting a pitch period using the autocorrelation method is constructed as shown in FIG. 6, for example.

In the figure, reference numeral 50 is a musical tone signal input terminal to which an analog musical tone signal having a waveform (example of a piano) as shown in FIG. 4 is inputted. This analog musical tone signal is converted to a digital musical tone signal by the A / D converter 51 and then supplied to the arithmetic unit 52.

The arithmetic unit 52 performs the sum-of-products arithmetic operation shown in the equation (1) to obtain the autocorrelation function value at each time difference d. When an autocorrelation function is drawn based on the value obtained by the calculation unit 52, a waveform as shown in FIG. 7 is obtained. The calculation result of the calculation unit 52 is sent to the maximum peak detection unit 53.

The maximum peak detecting section 53 obtains the maximum peak position of the autocorrelation function supplied from the calculating section 52 and sends it to the pitch period calculating section 54. Pitch cycle calculator 54
Then, the pitch period (the time from the first maximum peak position to the next maximum peak position) is calculated based on the maximum peak position information supplied from the maximum peak detection unit 53. The information indicating the pitch cycle calculated by the pitch cycle calculator 54 is output from the output terminal 55 and used for a predetermined purpose.

[0012]

As is apparent from the above equation (1), the conventional pitch period extracting device having the above-mentioned structure has the function of "X" when calculating the autocorrelation function for each time difference d. The multiplication of (n) × X (n + d) ”needs to be performed“ N−d ”times.

Generally, the time required for multiplication is enormous as compared with the time required for addition and subtraction. Therefore, the calculation of the autocorrelation function also requires a lot of time, and there is a problem that it is not suitable for an application in which the pitch period is detected in real time and used for a predetermined purpose.

In order to solve such a problem, a high-speed arithmetic element may be used. However, the high-speed arithmetic element is expensive and requires a special power source. Not applicable.

The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a pitch period extraction device capable of extracting a pitch period at a high speed in spite of having an inexpensive and simple structure. To do.

[0016]

In order to achieve the above object, the pitch period extracting device of the present invention is obtained by sampling an analog tone signal at a predetermined time interval, as shown in principle in FIG. And a calculating means 12 for extracting a predetermined section of the digital musical tone signal, sequentially obtaining and cumulatively adding a difference between digital musical tone signals at positions separated by the time difference for each time difference within the predetermined section, and the calculating means 12.
The minimum peak detecting means 13 for detecting the minimum peak position based on the values corresponding to the respective time differences cumulatively added in the above, and the pitch cycle calculation for calculating the pitch cycle from the information indicating the minimum peak position detected by the minimum peak detecting means 13. And means 14 for configuring.

More specifically, the calculating means 12 is
d is a time difference, n is a sample number (n = 1, 2, ..., N) attached to each digital musical tone signal within a predetermined section, X (n)
And X (n + d) are sample number n and sample number n
If it is the value of the digital tone signal at + d, It is configured to perform the operation indicated by.

[0018]

According to the present invention, a predetermined section of a given digital musical tone signal is extracted, and for each time difference within the predetermined section, the differences between the digital musical tone signals at positions separated by the time difference are sequentially obtained and cumulatively added. The calculation is performed, in other words, the calculation is performed for each time difference d according to the equation (1), and the minimum peak position is detected based on the calculation result.
The pitch cycle is extracted by performing a predetermined calculation on the detected minimum peak position.

Thus, as is clear from the above equation (1), it is not necessary to carry out multiplication in each operation process, and the pitch period can be obtained only by subtraction and addition (accumulation), so that the pitch can be obtained at high speed. The period can be extracted.
Further, since the multiplier is not required, the hardware can be reduced and the inexpensive pitch period extraction device can be realized.

[0020]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 2 is a block diagram schematically showing the construction of the main part of the embodiment of the pitch period extraction device according to the present invention.

In the figure, reference numeral 10 is an input terminal for converting musical sounds and vibrations generated by various musical instruments into electric signals and supplying them. The musical tone signal supplied to the input terminal 10 is an analog musical tone signal. This input terminal 10
The analog tone signal input from is supplied to the A / D converter 11.

The A / D converter 11 samples the input analog musical tone signal at a predetermined sampling rate and converts it into a digital musical tone signal. This A /
The output of the D converter 11 is supplied to the memories 16 and 17.

The memories 16 and 17 are the A / D converter 11
The digital tone signal supplied from the device is stored only for a predetermined section. The memories 16 and 17 respectively include a counter 18 and an address control circuit 19 described later.
The address generated in step 1 is supplied and read / write control is performed. This memory 16 and 17
The digital tone signal data read from the arithmetic circuit 1
2 is supplied.

The arithmetic circuit 12 obtains a difference by subtracting the output of the memory 16 and the output of the memory 17, and an accumulator internally provided with an absolute value of this difference (hereinafter referred to as "difference value"). (Not shown) is cumulatively added. The contents stored in the accumulator of the arithmetic circuit 12 are sent to the minimum peak detector 13 when the arithmetic operation for one time difference d is completed.

The minimum peak detection section 13 stores the values obtained by cumulatively adding the difference values corresponding to the respective time differences d sent from the arithmetic circuit 12, and checks these values to detect the minimum peak position. is there. The data indicating the minimum peak position detected by the minimum peak detector 13 is supplied to the pitch cycle calculator 14.

The pitch period calculation unit 14 calculates the pitch period by performing a predetermined calculation based on the data indicating the minimum peak position sent from the minimum peak detection unit 13. That is, the time between a predetermined minimum peak position and the next minimum peak position is calculated and used as pitch cycle data. The pitch cycle data calculated by the pitch cycle calculator 14 is output to the output terminal 15. This output terminal 15
The pitch period data output from the device will be used for a predetermined purpose.

Reference numeral 18 denotes a counter, which is a memory 16
To the memory 17 and each address to be supplied to the memory 17 via the address control circuit 19.
The reset pulse and the count clock of the counter 18 are given from the control circuit 20 described later.

The address control circuit 19 passes the output of the counter 18 as it is and supplies it to the memory 17 when the pitch period extraction device is in the memory write mode. On the other hand, when in the read mode from the memory,
The control signal SIG1 output from the control circuit 20 is added to the data output from the counter 18, and the result is supplied to the memory 17. Here, the control signal SIG1 is a signal indicating the time difference d.

As described above, the control circuit 20 supplies the reset pulse and the count clock to the counter 18,
Further, the control signal SIG1 is supplied to the address control circuit 19. The count clock is configured to be output in synchronization with the conversion speed of the A / D converter 11.

Further, the control circuit 20 supplies the arithmetic circuit 12 with a pulse for instructing the timing of executing the subtraction and the cumulative addition, and also instructs the minimum peak detector 13 of the timing for detecting the minimum peak position. The pulse is supplied, and further, the pitch period calculation unit 14 is supplied with a rule for instructing the timing for starting the calculation of the pitch period.

Next, the operation of the present embodiment having the above structure will be described.

In the present invention, basically, the operation circuit 12 performs the operation according to the function S (d) shown in the following equation (1) to obtain the function value for each time difference d, and these are calculated as the minimum peak position. It is used for detection.

In the above equation, d is a time difference, n is a sample number (n = 1, 2, ..., N), X (n) and X (n +) assigned to each digital musical tone signal in the extracted predetermined section.
d) is sample number n and sample number n + d, respectively
Is the value of the digital tone signal in.

The above function S (d) is "N-
"d" number of digital musical tone signals X (n) and "Nd" number of digital musical tone signals at a position where the sample number is shifted by d with respect to the digital musical tone signal X (n), in other words, the time difference d. It is obtained by the cumulative addition value of the difference value from the digital tone signal X (n + d).

Next, the operation of this embodiment will be described in detail with reference to the flowchart of FIG.

First, prior to the calculation of the function S (d), a process of storing the digital tone signal data in the memories 16 and 17 is performed.

That is, when an analog musical tone signal as shown in FIG. 4 (an example of a piano waveform is shown) is input from the input terminal 10, the analog musical tone signal is
The A / D converter 11 samples at a predetermined sampling rate and converts it into a digital tone signal.

The digital tone signal converted by the A / D converter 11 is supplied to the memories 16 and 17. At this time, the control circuit 20 supplies a reset pulse to the counter 18 to clear the contents of the counter 18, and then starts supplying the count clock. The output of the counter 18 is supplied to the memory 16 as a write address.

On the other hand, the control circuit 20 gives the address control circuit 19 a control signal (for example, which is realized by outputting the control signal SIG1 as zero) which should pass the output of the counter 18 as it is. As a result, the output of the counter 18 passes through the address control circuit 19 as it is and is supplied to the memory 17 as a write address.

That is, when writing data to the memories 16 and 17, the same address is supplied to the memories 16 and 17. Therefore, the same digital tone signal data is stored in the memory 16 and the memory 17.

The writing operation is performed by the counter 18 which is incremented in synchronization with the conversion speed of the A / D converter 11.
Is performed for each output (address). In addition, the counter 18
The maximum value counted by is defined by the value of “N” shown in the above equation (1). Therefore, the same digital musical tone signal data is sequentially stored in the memories 16 and 17 from the address 0 to at least the address "N".

When a predetermined amount of digital tone signal data is stored in the memories 16 and 17 in this way, first, the time difference d is initialized to zero (step S10). The initialization of the time difference d is equivalent to the initialization of the control signal SIG1 given from the control circuit 20 to the address control circuit 19 to zero.

Next, the operation of the function shown in the above equation (1) is executed (step S11). Therefore, first, two digital musical tone signal data are read from the same address of the memory 16 and the memory 17 and given to the arithmetic circuit 12, so that the arithmetic circuit 12 subtracts them.
Further, the absolute value is taken and cumulatively added to the accumulator provided inside according to the timing pulse from the control circuit 20.

This subtraction and cumulative addition are performed for "Nd" number of sample number values n = 1, 2, ..., Nd. Then, when the calculation on “Nd” pieces of data is completed, the calculation result is sent to the minimum peak detection unit 13 as a function value. The minimum peak detector 13 temporarily stores the received function value.

Then, it is checked whether or not the time difference d is larger than the time difference maximum value d _max (step S1).
2). Here, the maximum time difference value d _max is a value that defines the range in which the function is calculated, and is arbitrarily determined according to the frequency band to be detected and the required accuracy.

When it is determined in step S12 that the time difference d is not larger than the time difference maximum value d _max , the time difference d is incremented (step S13), and thereafter,
It returns to step S11. As a result, the control signal SIG1 supplied from the control circuit 20 to the address control circuit 19 is set to "1".

Therefore, in the calculation of the function in the second step S11, the reading of the digital tone signal data is started from the address "0" in the memory 16 and from the address "1" in the memory 17, and the function with the time difference d = 1. The value will be calculated.

Thereafter, similarly, steps S11 to S1 are performed.
The function value for each time difference d is calculated by repeating the operation 3 and stored in the minimum peak detection unit 13. When the function value for each time difference d calculated in this way is illustrated, the waveform is as shown in FIG.

When it is determined in step S12 that the time difference d is larger than the time difference maximum value d _{max in the} above-described repeated execution process, the minimum peak detection process is performed (step S14).

That is, the control circuit 20 controls the minimum peak detector 1
By giving a timing pulse instructing that the minimum peak position should be detected to 3, the minimum peak detection unit 13 checks the stored function value for each time difference d, and detects the minimum value from among them. . Then, the time difference d corresponding to the detected minimum value is sent to the pitch cycle calculator 14.

Next, the pitch cycle calculation is performed (step S15). That is, based on the minimum peak position information (time difference d) received from the minimum peak detection unit 13, the time between the first minimum peak position and the next minimum peak position is calculated as shown in FIG. Extract as pitch period. The pitch period extracted here is output from the output terminal 15 and used for a predetermined application.

As described above, according to the above-described embodiment, the predetermined section of the given digital tone signal is extracted, and the digital tone signal at the position separated by the time difference d for each time difference d in the predetermined section. The difference between the two is sequentially obtained and cumulative addition is performed. In other words, each time difference d is calculated according to the above equation (1), the minimum peak position is detected based on this calculation result, and the detected minimum peak is calculated. A predetermined calculation is performed on the position to extract the pitch period.

Thus, as is clear from the above equation (1), it is not necessary to perform multiplication, and the pitch cycle can be obtained only by subtraction and addition (cumulative addition), so that the pitch cycle can be extracted at high speed. In addition, since a multiplier is not required, the hardware can be reduced and an inexpensive pitch period extraction device can be realized.

According to this structure, a good effect can be obtained by being applied to an application for calculating a pitch period at high speed, for example, a synthesizer guitar, a vocoder, a digital pitch tuner or the like.

In the above embodiment, the two memories 16 and 17 are used as the memories for storing the tone signal data, but it is also possible to use one memory by using the two-port memory. In this case, there is an advantage that it is possible to configure a pitch period extraction device with reduced hardware.

[0057]

As described in detail above, according to the present invention, it is possible to provide a pitch period extraction device capable of extracting a pitch period at a high speed, although it has a cheap and simple structure.

[Brief description of drawings]

FIG. 1 is a diagram for explaining the principle of a pitch period extraction device of the present invention.

FIG. 2 is a block diagram schematically showing a configuration of an embodiment of a pitch period extraction device of the present invention.

FIG. 3 is a flow chart for explaining the operation of the embodiment of the present invention.

FIG. 4 is a diagram showing an example of a waveform of an analog musical tone signal input to the pitch period extracting apparatus of the present invention and the conventional one.

FIG. 5 is a diagram showing a function value for each time difference calculated using a function in the example of the present invention.

FIG. 6 is a block diagram showing a configuration of a pitch period extraction device that extracts a pitch period using a conventional autocorrelation function.

FIG. 7 is a diagram showing an autocorrelation function value for each time difference in a pitch period extraction device that extracts a pitch period using a conventional autocorrelation function.

[Explanation of symbols]

 10 Input Terminals 11 A / D Converter 12 Arithmetic Means (Arithmetic Circuits) 13 Minimal Peak Detecting Means (Minimum Peak Detecting Units) 14 Pitch Cycle Calculating Means (Pitch Cycle Calculating Units) 15 Output Terminals 16, 17 Memory 18 Counter 19 Address Control Circuit 20 Control circuit

Claims (2)

[Claims] 1. A digital musical tone signal at a position separated by the time difference with respect to each time difference in the digital musical tone signal obtained by sampling an analog musical tone signal at a predetermined time interval. An arithmetic means for sequentially obtaining differences between them and cumulatively adding them, a minimal peak detecting means for detecting a minimal peak position based on a value corresponding to each time difference cumulatively added by the computing means, and a minimal peak detecting means A pitch period extracting device comprising: a pitch period calculating means for calculating a pitch period from information indicating a minimum peak position. 2. The calculating means is a sample number (n = 1, 2, ..., N), X (n) and X (n + d) assigned to each digital tone signal in a predetermined section, where d is a time difference and n is a predetermined section. 2. The pitch period extraction device according to claim 1, wherein the calculation of the following formula (1) is performed when) is the value of the digital tone signal at sample number n and sample number n + d.