JPH04329599A - pitch extraction device - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pitch extraction device for extracting the pitch of a waveform from an input audio signal.

0002

[Prior Art] There has been a desire to extract the pitch frequency or pitch period (hereinafter simply referred to as "pitch") from an audio signal consisting of a repetitive waveform, such as a human singing voice or a musical tone collected from a microphone. is strong. If a pitch can be extracted, the extracted pitch information can be used in various processes, such as allowing an electronic musical instrument to produce a musical tone corresponding to the pitch of that pitch.

[0003]Various methods have been proposed as techniques for extracting pitch from audio signals, but among them, methods for directly extracting pitch from audio signals in the time domain require a small amount of calculations and are easy to process in real time. This is possible.

A first conventional example of a time-domain pitch extraction method is, for example, a method in which the time between adjacent zero-crossing points (time positions where the amplitude becomes 0) of an input waveform is determined as the pitch period of that waveform. There is.

In addition, as a second conventional example, there is a method in which the input waveform is passed through a low-pass filter to remove overtone components, and then the zero-cross interval is extracted. Furthermore, the third
As a conventional example, a method has also been proposed in which the autocorrelation value of an input waveform is calculated and the pitch period is extracted from the deviation amount of the autocorrelation that reaches the peak value.

[0006]

[Problem to be Solved by the Invention] However, human voices contain a considerable amount of overtones, and zero-crossing points appear several times during one pitch period of the input waveform. The method of extracting the interval has the problem that the correct period cannot be obtained and pitch extraction errors are large.

Furthermore, in the second conventional example, when the cutoff frequency of the low-pass filter is lowered, the range from which pitches can be extracted is limited to the lower range, and conversely, when the cutoff frequency is raised, overtone components remain. This has the contradictory problem of increasing pitch extraction errors.

Furthermore, although the third conventional example is a time domain process, it requires a very large amount of calculation, and is not suitable for extracting the pitch of an input waveform in real time. have.

An object of the present invention is to eliminate the influence of zero crossings caused by overtone components as described above, and to enable correct measurement of zero-crossing intervals of an input waveform corresponding to one pitch period.

0010

[Means for Solving the Problems] The present invention provides a method for firstly starting from a zero-crossing point where an input waveform zero-crosses from a negative or positive first polarity to the other second polarity, and then from the second polarity to the first polarity. It has a first interval measuring means that measures an interval to a zero-crossing point where the polarity zero-crosses as a first interval.

Next, the first amplitude is measured by measuring the maximum absolute value of each amplitude of the input waveform while the first interval measuring means measures the first interval as the first amplitude maximum value. It has a maximum value measuring means.

Next, after the measurement by the first interval measuring means is completed, the input waveform zero-crosses from the first polarity to the second polarity, and then zero-crosses from the second polarity to the first polarity. It has a second interval measuring means that sequentially measures the interval of each section up to the zero crossing point as a second interval.

[0013] Next, the second interval measuring means sequentially measures the maximum absolute value of each amplitude of the input waveform as the second amplitude maximum value in each section in which each second interval is sequentially measured. It has a second amplitude maximum value measuring means.

[0014] Furthermore, the third interval measuring means sequentially measures the interval from the zero cross point at which the measurement by the first interval measuring means has started.
It has interval measuring means. Each of the above-mentioned first, second, and third interval measuring means is constituted by, for example, a counter.

[0015] Also, an interval storage means is provided which sequentially updates and stores the intervals measured by the third interval measuring means as periodic information at each point in time when the second measuring means starts measuring each second interval. have

In addition, the first interval measured by the first interval measuring means and each second interval sequentially measured by the second interval measuring means match within a predetermined tolerance range, and Sequentially determining whether the first amplitude maximum value measured by the amplitude maximum value measuring means and each second amplitude maximum value sequentially measured by the second amplitude maximum value measuring means match within a predetermined tolerance range. It has a determination means.

[0017] Then, at the time when the determining means detects a match, the period information stored in the time storage means is outputted as pitch information of the input waveform, and thereafter, each process is performed in order from the measurement process by the first interval measuring means. It has a repetition control means that sequentially extracts pitch information from an input waveform by repeating the repetition.

In the configuration of the invention described above, at the time when the determining means does not detect a match, the interval measured by the third interval measuring means exceeds the interval corresponding to the preset pitch period of the lowest note. It may be configured to include a limiting means that determines whether or not the interval has exceeded the interval, and when it is determined that the interval has been exceeded, causes each process to be restarted from the measurement process by the first interval measuring means.

[0019]

[Operation] The audio signal is constructed as a waveform in which one or more waveform crests repeat in each pitch period, and each waveform crest within each pitch period exhibits a similar shape in consecutive pitch periods. Further, the pitch between the zero crossing points at both ends of the crests of substantially similar waveforms of adjacent pitch periods and the heights of the crests are substantially equal to each other.

[0020] The present invention focuses on this point, and first, the first
The interval measuring means measures the interval between zero-crossing points at both ends of one waveform peak, and the first amplitude maximum value measuring means measures the first amplitude maximum value corresponding to the height of the peak. Further, the second interval measuring means sequentially measures the intervals between the zero crossing points at both ends of the peaks that appear sequentially from then on, and the second maximum amplitude value measuring means measures the second maximum amplitude value corresponding to the height of each peak. measure.

[0021]The determining means determines that the first interval and each second interval match within a predetermined tolerance range, and the first amplitude maximum value and each second amplitude maximum value are equal to each other within a predetermined tolerance range. It is sequentially determined whether or not there is a match within a predetermined tolerance range, and the interval from the peak corresponding to the first interval to the top of the peak corresponding to the second interval at the time of the match is extracted as pitch information. . This pitch information can be obtained as period information that is each measurement interval of the third measuring means stored in the interval storage means.

[0022] In this way, the pitch of the input waveform can be stably extracted regardless of the influence of overtone components.

[0023]

Embodiments Hereinafter, embodiments of the present invention will be described with reference to the drawings. <Configuration of Embodiment> FIG. 1 is an overall configuration diagram of an embodiment of the present invention.

In the figure, an analog signal such as the singing voice of one person or the sound of a single note played by an analog musical instrument collected by a microphone (not shown) is input to A via a capacitor 1 for removing DC components. The signal is input to the /D converter 2 and converted into a digital signal.

[0025] The CPU (central processing and control unit) 3 has a RAM
(Random Access Memory) 4 is used as work memory, ROM (Read Only Me
By executing the program stored in the memory) 9, information regarding the pitch of the input audio signal is extracted.

Next, the sound source 5 is controlled by the CPU 3.
A musical tone signal having a note number based on the above-mentioned pitch information and a tone selected by the performer using a tone selection switch (not shown) (for example, a tone of a wind instrument such as a trumpet that is completely different from the above-mentioned singing voice). generate. Thereafter, the generated musical tone signal is converted into an analog musical tone signal by a D/A converter 6, and is emitted as a musical tone from a speaker 8 via an amplifier 7.

[0027] In addition to the above-mentioned programs, the ROM 9 also contains various data used for pitch extraction, a table showing the correspondence between the period of the obtained input signal and the note number, etc., as will be described later. .

With the above-described circuit configuration, pitch data (hereinafter simply referred to as pitch) is extracted from the input audio signal in accordance with the basic operating principle described below. <Basic operating principle of the embodiment> Normally, a voice that has a clear sense of pitch and is played as a single note, such as a human singing voice, is repeated every pitch cycle (for example, ■) as shown in FIG. , are composed of substantially similar waveform elements (for example, a and d, b and e, and c and f in the figure).

Such a voice has many zero-crossing points within one pitch period where its waveform crosses a reference level with zero amplitude due to the overtone components contained in the voice. Therefore, in order to find the pitch period from a waveform with such periodicity, it is necessary to determine which zero-crossing points should be used as the pitch period. There are characteristics like.

That is, the time of the section (for example, ■ and ■) determined by the respective zero-crossing points (for example, Z1 to Z2 and Z7 to Z8) of substantially the same waveforms of adjacent pitch periods (for example, a and d above), and A The maximum values of the /D-converted amplitude values (for example, Pa and Pd) are approximately equal.

Focusing on this point, the present invention finds two sections (for example, ■ and ■) in which both the time between adjacent zero crossing points and the maximum value of amplitude are approximately equal, and thereby The respective starting points of the interval (e.g. Z1 and Z
7) to extract the pitch period.

The operation of this embodiment, which operates according to the basic operating principle as described above, will be explained below. <Summary of the Embodiment> In FIG. 2, for example, the time in section 2 between zero crossing points Z1 and Z2 of waveform a is counted, and the count value COUNT 1 is held in a first counter. Next, waveforms b and c having the same polarity as waveform a
, d..., the time of each interval ■, ■, ■ is sequentially counted, and the count value COUNT 2 is CO
It is held in the second counter while being sequentially updated until a value approximately equal to UNT 1 is obtained.

In this case, from zero cross point Z1 to zero cross points Z3, Z5, and Z7 of waveforms b, c, and d, respectively.
The times ■, ■, and ■ are counted by a third counter. Since the count value COUNT 3 has the possibility of becoming the pitch period of this input audio signal, the variable PITCOUNTER in RAM 4 is updated sequentially until the count value that becomes the pitch period is obtained.
will be saved in Also, the maximum value Pa of the amplitude of waveform a is RA
The maximum values Pb, Pc, and Pd of waveforms b, c, and d are written to the variable MAX2 in the RAM4 while being sequentially updated to the variable MAX1 in the M4.

Then, an interval ■ having a time almost equal to the time of the interval ■ is found, and the values written in the variables Pa and Pd corresponding to the maximum values of the respective amplitudes of both the intervals ■ and ■ are as follows. When it is confirmed that they are almost equal, the counter value corresponding to the time ■ between the zero crossing points Z1 and Z7 of both sections ■ and ■ is stored in the memory variable PIT CO.
It is read from UNTER and determined as the pitch period. Then, the note number corresponding to that pitch period is determined from the table in the ROM 9. <Operation of this Embodiment> Next, the operation of determining the pitch of an input waveform according to the above-mentioned operating principle will be explained using the operation flowcharts of FIGS. 3 to 7. These operation flowcharts are realized as operations in which the CPU 3 in FIG. 1 executes a control program stored in the ROM 9.

In FIG. 3, first, the RAM 4 is initialized. In other words, a status flag (STATUS) indicates which state of the signal waveform the time point at which the audio signal is read corresponds to.
FLAG (see Figure 2), a variable ENV that indicates the size of the waveform envelope of the input audio signal, and a variable P that holds the note number corresponding to the pitch obtained as a result of pitch extraction.
IT, variable O that holds the currently sounding note number
N NOW, variable COUNT 1, CO that holds each count value of the first, second, and third three counters.
UNT 2, COUNT 3, and previous A/D
The contents of the variable ADMEM, which holds the converted value, are all cleared (step S301). In addition, PIT
A value of 0 or ONNOW indicates that there is no pitch-extracted sound, or that there is no sound currently being produced.

Next, after waiting for the completion of A/D conversion for one sampling of the input audio signal (S302), the process proceeds to the next step S303, where the register reg1 in the CPU 3 is
A/D converted data is set to .

Thereafter, the process is branched into three depending on the status flag (step S304). Initially,
Since the status flag in FIG. 2 is 0, the process advances to step S305. When the status flag = 0, in order to detect the zero-crossing point Z1 from the negative side to the positive side of the input waveform, the previous A/D conversion value ADMEM is negative and the current A/D conversion value ADMEM is negative. reg where the value is stored
It is determined whether or not 1 is positive.

In this case, the determination in step S305 is NO.
Therefore, the process advances to step S307. And this time A/D
The value of reg1 with the conversion value set is the variable ADMEM
(step S307), envelope extraction (step S308), musical tone note-on/note-off processing (sound generation/mute processing, step S309), and then A again.
The process waits for the completion of /D conversion (step S302).

[0039] Envelope extraction (step S308)
) and musical tone note-on/note-off (step S309
) are shown in the operation flowcharts of FIGS. 5 and 6.

In FIG. 5, first, the absolute value of the current A/D conversion value (the contents of reg1) is calculated (step S5).
01), and then passed through a low-pass filter (by performing a digital filtering operation) to shape the input waveform (step S502). The value obtained as a result is stored in the variable ENV (step S503).

In the note-on/note-off process shown in FIG.
First, the size of the variable ENV is set to a predetermined threshold (THRESH
A variable P that exceeds OLD) and holds the note number
It is determined whether the value of IT is greater than 0 (step S601).

If the determination is NO, the muting of all sounds being produced by the sound source 5 in FIG. 1 is instructed, and the variable ON NOW is cleared to indicate that no sounds are being produced (step S606), the process returns to the main routine (FIG. 3).

If the determination in step S601 is YES, it is determined in step S602 whether or not there is a sound currently being produced. If the determination is YES, step S60
Proceed to step 5. If the determination in step S602 is NO, the values of the variable ON NOW indicating the note number currently being sounded and the variable PIT indicating the extracted note number are compared (step S603). As a result of the comparison, if the two match, there is no need to re-sound, and the process returns to the main routine (Figure 3); if they do not match, the note number to be sounded has changed, and the currently sounding note is muted. is instructed to the sound source 5 (step S604), and then the process advances to step S605.

In step S605, a sound generation instruction is given in accordance with the variable PIT indicating the note number, and the value of the variable PIT is assigned to the variable ON NOW, so that a musical tone having a pitch corresponding to the variable PIT is currently being produced. It is shown that Thereafter, the process returns to the main routine (FIG. 3).

The above processing is repeated until the zero crossing point from the negative side to the positive side (Z1 in FIG. 2) is reached, and eventually the determination in step S305 in FIG. 3 becomes YES, and the status flag STATUS FLAG becomes 1. (step S306). Here, the count value COUNT of the time in section ■ shown in FIG.
1, and the count value CO of time ■, or ■, ■
The initial value 1 is set to UNT 3, and COU
In order to obtain the maximum amplitude value of the input waveform during the count-up of NT 1 , the value of the register reg 1 which is the current A/D conversion value is written into the variable MAX 1 .

[0046] After this, the same operation as described above is performed, and the process returns to step S304. Here, STATUS
Since the state is FLAG = 1, step S3
Move on to 10. In this state, the variable COUNT 1 (hereinafter referred to as the first counter COUNT
1) is also used as a third counter to count the time exemplified in section ■.
(referred to as UNT 3 ) is started.

First, in step S310, in order to detect the zero-crossing point Z2 from the positive side to the negative side of the input waveform, that is, the end point of section 2 in FIG. Variable ADMEM with value written
It is determined whether the value of is positive and the value of register reg 1 to which the current A/D conversion value is written is negative.

If the determination is NO, the first counter CO
The value of UNT 1 is increased by 1 (step S311
), the process proceeds to the maximum value detection routine in FIG. 7 (step S31
2). In FIG. 7, it is determined whether the current input value is greater than or equal to the value of the variable MAX 1 (step S701), and M
If it is greater than or equal to AX 1 , the current input value is assigned to MAX 1 (step S702). On the other hand, MAX 1
If it is smaller, nothing is done and the process returns to FIG. 3. In this way, as the value of the variable MAX 1, A/D
The maximum value of the converted values is determined.

After the process of step S312 described above, the process proceeds to step S314, where the value of the third counter COUNT3 is incremented by one. After that, step S310→S3
11→S312→S314→S307→S308→S3
The loop processing of 09→S302 to S304→S310... is repeated, and the first counter CO is
The values of UNT 1 and the third counter COUNT 3 are incremented by one.

[0050]Then, the determination in step S310 is YES.
When the zero crossing point Z2 from the positive side to the negative side is reached, the status flag is set to 2 (step S31
2). After that, the process advances to step S314, and the COUNT
After the value of 3 is increased by 1, step S307→S3
08→S309, and returns to step S302.

After the above loop processing is performed, the image shown in FIG.
A counter value corresponding to the time of interval (3) is held in the first counter COUNT 1 , and a maximum value of the amplitude value of waveform a is held in the variable MAX 1 .

Next, in step S312, the status
Since the flag is set to 2, this time step S302
From ~S304, the process proceeds to steps S401→S402 in FIG. In step S402, in order to detect the zero crossing point from the negative side to the positive side (Z3 in FIG. 4), the same process as in step S305 in FIG. ) is negative and register re
It is determined whether the value of g1 (current A/D conversion value) is positive.

[0053] While the determination is NO, S402→S
314 (Figure 3) → S307 to S309 → S302 to S3
The loop processing of 04→S401 (FIG. 4)→S402... is repeated, and the third counter COUNT
The count value of T 3 is increased by 1 (S31
4).

[0054] Then, the determination in step S402 is YES.
When the zero crossing point Z3 from the negative side to the positive side is reached, the status flag is set to 3, and the CO at this time is
The value of UNT 3 (corresponding to time ■ in Figure 2) is the variable PIT
Saved in COUNTER (step S403)
. The value of this PIT COUNTER is held as a candidate for the cycle corresponding to the pitch to be extracted, that is, as a value that has the possibility of becoming the count value of section 2 in FIG. Then, the variable COUNT2, which serves as the second counter, starts to be incremented by one.

After that, steps S403→S314 (FIG. 3) to S309→S302 to S304→S401 (FIG. 4)
) and proceed. This time, the state is STATUS FLAG = 3, so steps S401 to S40 are performed.
Proceed to step 4. In this state, STATUS FLAG = 1
Through the same processing as in the state of , the count of COUNT 2 is increased and the maximum value of the input waveform is determined as MAX 2 .

First, in step S404, the value of the variable ADEM (one It is determined whether the previous A/D conversion value) is positive and the value of register reg1 (current A/D conversion value) is negative (step S404).
).

[0057] While this determination is NO, step S404
→S405-S406→S314 (Figure 3)-S309→
The loop processing of S302 to S304 → S401 (FIG. 4) → S404 is repeated, and in each repetition, step S
Every time the process 405 is performed, the second counter COUN
The value 1 is added to T 2 and step S40
6, the maximum value of the A/D conversion value is determined in the variable MAX 2 .

The routine for determining the maximum value of MAX 2 is the same process as in the state of STATUS FLAG = 1, and is shown from step S703 onward in FIG. Here, it is determined whether the current input value is greater than or equal to the value of the variable MAX 2 (step S703), and MAX
If it is 2 or more, the current input value is substituted into MAX 2 (step S704). On the other hand, if it is smaller than MAX 2 , nothing is done and the process returns to the process of FIG. 4 . In this way, the maximum value of the A/D conversion value is determined as the value of the variable MAX 2 .

By the above-described processing, the time of section 3 of the input waveform in FIG. 2 and the maximum value Pb of waveform b are now determined. Note that the value of the third counter COUNT 3 is also increased by 1 for each repetition described above (S314).
).

[0060] If the determination in step S404 becomes YES, it means that the zero cross point Z4 has been detected, and the status flag is changed to 4 here (step S407).
). After that, the process advances to step S314 in FIG.
After T 3 is increased by 1, step S307→S
308→S309 to S401 (FIG. 4).

[0061] This time, since STATUS FLAG = 4, steps S401 to S4 are executed.
Proceed to 08. In this state, the sections ■, ■, and ■ are sequentially compared with the section ■ in FIG. The values of variables MAX 1 and MAX 2 , which are the maximum values of the A/D conversion values of a and d, are compared, and by determining that they are almost equal, the pitch of the input waveform is determined by A processing operation defined as a period is performed.

First, check whether the values of the first counter COUNT 1 and the second counter COUNT 2 obtained so far are approximately equal, that is, COUNT
It is determined whether the absolute value of the difference between the values of 1 and COUNT 2 is within the allowable error range (step S408). This tolerance range depends on the sampling frequency and pitch frequency of the input signal, but is determined empirically.

[0063] In this case, COUNT 1 and COUNT
Since T 2 holds the count values of sections ■ and ■ in FIG. 2, respectively, the determination in step S408 is NO and the process proceeds to step S409. Here, if noise or the like is mixed into the input audio signal, this will cause the step S
There are cases where the NO determination in step 408 continues for an abnormally long time, and no instruction to produce musical tones is given.

[0064] In order to prevent the occurrence of such abnormal conditions,
the value of a third counter COUNT 3 counting the time to be taken as the pitch period of the audio signal is constantly monitored;
It is determined whether the value exceeds a predetermined limit value LIMIT which is much larger than the pitch period of the lowest tone of the audio signal (step S409).

[0065] This limit value is similar to the above-mentioned allowable error range,
Determined empirically. For example, set the sampling frequency to 20
KHz, minimum extracted pitch (pitch) is 110.5Hz
(note name: equivalent to A1), then 20000 ÷ 110
．． 5 = 180.995... Therefore, the limit value may be set around 180.

[0066] If the determination in step S409 is YES,
That is, in the case of an abnormal state, the process advances to step S315 in FIG. 3, where the sound source 5 in FIG. 2 is instructed to mute all the sounds being produced, and the variable ON NOW is cleared. After that, the process returns to step S301, and the above-described pitch extraction processing operation is executed again.

[0067] If the determination in step S409 is NO,
Proceeding to step S410, the value of the variable MAX2 is rewritten to the current value reg1, and the second counter COUNT
The value of T 2 is cleared and the status flag is set back to 2. Then, the process begins again to find the next zero cross. At this time, the variable PIT COUNT
Although the value held in ER becomes invalid, the third counter COUNT 3 continues to count up, and the first counter COUNT 1 also holds the count value in section 2 in FIG.

2 of the status flags as described above.
After returning to the state, the status flag progresses through the 3 state and 4 state as before, and the second counter COU
NT 2 is the count value of section ■ in Figure 2, variable PIT
COUNTER holds the count value (■ in FIG. 2) from the start of section (2) in the figure to the zero cross point Z5 of section (2). Also, A written in the variable MAX 2
The maximum value of the /D conversion value is cleared, and the maximum value of waveform c is newly determined.

However, this time as well, the value of COUNT 1 (corresponding to the time in category ■) and the value of COUNT 2 (corresponding to the time in category ■) do not match, so the determination in step S408 is NO. Therefore, the process advances to step S410 via step S409, and the execution is restarted from the state where the status flag is 2.

Next, when the status flag becomes 4 again, COUNT 2 retains the count value of section ■ in FIG. 2, while PIT C
OUNTER holds the count value at time ■ in the figure. Then, in step S408, Y for the first time.
It is determined as S, and the process advances to step S411.

In step S411, the variable MA
The maximum value of the A/D conversion value of waveform a held in X 1 and the A/D conversion value of waveform d held in variable MAX 2
It is determined whether the maximum values of the D-converted values substantially match.

When this determination is NO, it is determined that the waveform does not correspond to the correct pitch period, and the process proceeds to step S409, where STATUS FLAG =2.
The process is re-executed from this state and a new waveform is searched for.

On the other hand, if the determination in step S411 is YES, it is determined that a waveform corresponding to the correct pitch period has been found, and the process advances to step S412. The correct value of the pitch period of the input waveform at this time is PIT CO
The third counter COUNT held in UNTER
The value of T 3 (■ in FIG. 2) is obtained. Then, the note number corresponding to this value is read from the table in the ROM 9 in FIG. 1 and is transferred to the variable PIT (step S4
12).

Next, the status flag, all counters, and the contents of variables MAX 1 and MAX 2 are cleared and reset to the initial state (step S
413). After that, steps S307 to S30 in FIG.
Proceed to note on/off processing in step 9, and in this process,
A new production instruction for a musical tone having a note number corresponding to the pitch (pitch) of the above-mentioned input audio signal is given.

That is, in the flow of FIG. 6 corresponding to step S309, after the determination in step S601 is YES, it is determined in step S602 that there is no sound currently being produced (YES).
S), as described above, step S60
At step 5, a pronunciation instruction is given in accordance with the variable PIT indicating the note number. Furthermore, if it is determined in step S602 that there is a sound currently being produced (NO), as described above,
In step S603, the values of the variable ON NOW indicating the note number currently being sounded and the variable PIT indicating the extracted note number are compared. As a result of the comparison, if the two match, there is no need to re-sound, and the process returns to the main routine (Figure 3); if they do not match, the note number to be sounded has changed, and the currently sounding note is The sound source 5 is instructed to mute (step S604), and then, in step S605, a new sound generation instruction is given in accordance with the variable PIT in which the new note number is stored.

After the above pronunciation instructions, step S3 in FIG.
09 → Proceed to S302-S304 and STATUS FL
From the state of AG=0, the same operation as before is repeated. In the embodiment described above, for the waveform portions on the positive side (for example, waveforms a, b, c, d, etc. in FIG. 2) with respect to the reference level of the waveform amplitude value 0, the time of each section is Although the pitch is extracted by comparing the maximum value of the A/D conversion values, the present invention is not limited to this, and the pitch extraction operation can be performed in exactly the same way using the negative side waveform portion.

Furthermore, the above-described pitch extraction operation is performed in parallel for each of the waveforms on the positive side and negative side with respect to the reference level, and the respective third counters COUNT 3
For example, if the values of 2 and 2 agree with each other with a predetermined error allowed,
If the average value of the two COUNT 3 values is determined as the pitch period of the input waveform, pitch extraction with higher accuracy is possible.

In addition, in the note on/off process shown in FIG. 6, if the pitch of the musical tone being sounded changes, the process is performed such as muting the sound and then reproducing it. For example, if the envelope has not changed much, it is also possible to instruct the sound source 5 in FIG. 2 to change only the pitch without muting the sound.

[0079]

According to the present invention, the pitch of the input waveform can be adjusted regardless of the influence of overtone components by extracting the point in time when the intervals between zero crossing points are almost equal and the peaks of the waveform are determined to be similar. It becomes possible to extract stably. In this case, in particular, more accurate pitch extraction is possible by using not only the interval between zero-crossing points but also the maximum amplitude value corresponding to the height of the peak of the waveform as a determining factor.

[0080] This makes it possible to reliably extract, for example, a pitch indicating the pitch from a voice waveform such as a singing voice or musical tone that includes overtones.

[Brief explanation of the drawing]

FIG. 1 is an overall configuration diagram of an embodiment of the present invention.

FIG. 2 is a waveform diagram of an input audio signal.

FIG. 3 is an operation flowchart (Part 1) of this embodiment.

FIG. 4 is an operation flowchart (Part 2) of this embodiment.

FIG. 5 is an operational flowchart regarding envelope detection in this embodiment.

FIG. 6 is an operation flowchart regarding note on/off in this embodiment.

FIG. 7 is an operation flowchart regarding maximum value detection of an input waveform according to the present embodiment.

[Explanation of symbols]

1 Capacitor 2 A/D converter 3 CPU 4 RAM 5 Sound source 6 D/A converter 7 Amplifier 8 Speaker 9 ROM

Claims (2)

[Claims] [Claim 1] An interval from a zero-crossing point where an input waveform zero-crosses from a negative or positive first polarity to the other second polarity to a zero-crossing point where the input waveform zero-crosses from the second polarity to the first polarity. The first interval is measured as the first interval.
and a first amplitude that measures the maximum value of each amplitude of the input waveform while the first interval measuring means measures the first interval as a first amplitude maximum value. After the measurement by the maximum value measuring means and the first interval measuring means is completed, the input waveform zero-crosses from the first polarity to the second polarity, and then from the second polarity to the second polarity. a second interval measuring means that sequentially measures intervals of each section up to a zero crossing point where the zero crosses to a polarity of 1 as a second interval; and the second interval measuring means sequentially measures the respective second intervals. measurement by a second amplitude maximum value measuring means that sequentially measures the maximum absolute value of each amplitude of the input waveform as a second amplitude maximum value in each interval in which the measurement is performed; and a measurement by the first interval measuring means. a third interval measuring means for sequentially measuring the interval from the zero crossing point at which the second interval is started; an interval storage means for sequentially updating and storing measured intervals as cycle information; and the first interval measured by the first interval measuring means and each of the intervals sequentially measured by the second interval measuring means. the first amplitude maximum value measured by the first amplitude maximum value measuring means and the second amplitude maximum value measuring means successively measured by the first amplitude maximum value measuring means and the second amplitude maximum value measuring means. determining means for sequentially determining whether or not the second maximum amplitude values match within a predetermined tolerance range;
At the time when the determination means detects the coincidence, the period information stored in the time storage means is outputted as pitch information of the input waveform, and thereafter, each of the period information is sequentially measured starting from the measurement processing by the first interval measurement means. A pitch extraction device comprising: repetition control means for sequentially extracting pitch information from the input waveform by repeating processing. 2. At the time when the determining means does not detect the coincidence, it is determined whether the interval being measured by the third interval measuring means exceeds a preset interval corresponding to the pitch period of the lowest note. If it is determined that the
2. The pitch extraction device according to claim 1, further comprising a limiting means for redoing each of the processes from the measurement process by the first interval measuring means.