JP3092198B2 - Pitch extraction device - Google Patents

Abstract

PURPOSE:To correctly measure zero-cross intervals of an input waveform corresponding to a one-pitch period by eliminating the influence of the zero crossing of harmonic components when pitches are extracted from a voice signal of a human voice, a musical sound, etc. CONSTITUTION:A CPU measures the cumulative values of the interval between zero-cross points at both ends of, a wave form part (a) whose polarity is, for example, positive and the amplitude of the waveform part (a) and then extracts the interval between the start points (zero-cross point) of waveforms (a) and (d) as the pitch period of the input waveform when finding the waveform part (d) which is at the shortest distance in terms of time while both the cumulative values of the intervals between zero-cross points at both ends of equal-polarity waveform parts (b), (c), and (d) and the amplitudes of the waveform parts are nearly equal to the cumulative values of the interval and amplitude of the waveform part (a).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

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

[0002]

2. Description of the Related Art Conventionally, there has been a demand for extracting a pitch frequency or a pitch period (hereinafter simply referred to as "pitch") from an audio signal having a repetitive waveform such as a singing voice sung by a person or a musical tone collected from a microphone. Is strong. If the pitch can be extracted, the extracted pitch information can be used for various processes, for example, a musical tone corresponding to the pitch of the pitch can be generated from an electronic musical instrument or the like.

Conventionally, various techniques have been proposed as techniques for extracting a pitch from an audio signal. Among them, a method for directly extracting a pitch from an audio signal in a time domain requires a small amount of calculation and facilitates real-time processing. It is feasible.

[0004] As a first conventional example of a time domain pitch extraction method, for example, a time between adjacent zero cross points (time positions where the amplitude becomes 0) of an input waveform is obtained as a pitch period of the waveform. There is.

As a second conventional example, there is a method in which an input waveform is passed through a low-pass filter to remove a harmonic component, and then a zero-cross interval is extracted. Furthermore, the third
As a conventional example, there has been proposed a method of calculating an autocorrelation value of an input waveform and extracting a pitch period from a shift amount of the autocorrelation at which the value becomes a peak.

[0006]

However, since a human voice or the like contains considerable overtones and a zero-cross point appears several times during one pitch period of the input waveform, the zero-cross point is simply obtained as in the first conventional example. The method of extracting an interval has a problem that a correct period cannot be obtained and a pitch extraction error is large.

In the second conventional example, when the cutoff frequency of the low-pass filter is lowered, the range in which the pitch can be extracted is limited to a low range. Conversely, when the cutoff frequency is raised, harmonic components remain. There is a contradictory problem that pitch extraction errors increase.

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

SUMMARY OF THE INVENTION It is an object of the present invention to eliminate the above-mentioned effect of zero crossing due to harmonic components and to enable correct measurement of the zero crossing interval of an input waveform corresponding to one pitch period.

[0010]

According to the present invention, first, an input waveform starts from a zero crossing point where the input waveform crosses zero from a first polarity, negative or positive, to another second polarity, and then from the second polarity to the first polarity. And a first interval measuring means for measuring an interval to a zero-crossing point at which a zero-crossing is made to the polarity as a first interval.

Then, the first interval measuring means measures the accumulated value of the absolute value of each amplitude of the input waveform as the first amplitude accumulated value while the first interval measuring means measures the first interval. Has an amplitude accumulated value measuring means.

Next, after the end of the measurement by the first interval measuring means, the input waveform zero-crosses from the second polarity to the first polarity from the zero-cross point where the input waveform crosses zero from the first polarity to the second polarity. There is provided a second interval measuring means for sequentially measuring an interval of each section up to the zero cross point as a second interval.

Next, the accumulated value of the absolute value of each amplitude of the input waveform is set as the second amplitude accumulated value in each section where the second interval measuring means sequentially measures each second interval. It has a second amplitude accumulated value measuring means for sequentially measuring.

Further, a third step is to sequentially measure an interval from the zero-cross point at which the measurement by the first interval measuring means is started.
Interval measuring means. Each of the first, second, and third interval measuring means described above is constituted by, for example, a counter.

Further, at each time when the second measuring means starts measuring each second interval, an interval storing means for storing the intervals measured by the third interval measuring means while sequentially updating them as cycle information is provided. 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 coincide with each other within a predetermined allowable error range, and Whether the first accumulated amplitude value measured by the accumulated amplitude value measuring means and each second accumulated amplitude value sequentially measured by the second accumulated amplitude value measuring means coincide with each other within a predetermined allowable error range. It has a determination means for sequentially determining whether

Then, when the determination means detects the coincidence, the period information stored in the time storage means is output as pitch information of the input waveform, and thereafter, each processing is sequentially performed from the measurement processing by the first interval measurement means. It has a repetition control means for successively extracting pitch information from the input waveform by repeating.

In the configuration of the invention described above, when the determination means does not detect a match, the interval measured by the third interval measurement means exceeds the interval corresponding to the preset pitch cycle of the lowest tone. It is possible to have a limiting means for determining whether or not the processing has been performed and, when it is determined to have exceeded the limit, restarting each processing from the measurement processing by the first interval measuring means.

[0019]

The audio signal is formed as a waveform in which one or more peaks of the waveform repeat at each pitch period, and the peaks of each waveform in each pitch period exhibit a similar shape between successive pitch periods. The interval between the zero-cross points at both ends of the peaks having substantially similar waveforms in adjacent pitch periods and the peak area are substantially equal to each other.

In the present invention, focusing on this point, first, the first
The interval measuring means measures the interval between zero cross points at both ends of one peak of one waveform, and the first amplitude accumulated value measuring means measures the first amplitude accumulated value corresponding to the area of the peak. Further, the second interval measuring means sequentially measures the intervals of the zero cross points at both ends of the peaks which appear sequentially thereafter, and the second amplitude accumulated value measuring means measures the second amplitude accumulated value corresponding to the area of each peak. Is measured.

The determination means determines that the first interval and each second interval match within a predetermined allowable error range, and that the first amplitude accumulation value and each second amplitude accumulation It is sequentially determined whether or not the values match within a predetermined allowable error range, and the interval from the peak corresponding to the first interval to the head of the peak corresponding to the second interval at the time of matching is determined as pitch information. Extract. This pitch information can be obtained as period information which is each measurement interval of the third measurement means stored in the interval storage means.

Thus, the pitch of the input waveform can be stably extracted regardless of the influence of the harmonic components.

[0023]

Embodiments of the present invention will be described below with reference to the drawings. <Structure of Embodiment> FIG. 1 is an overall structural diagram of an embodiment of the present invention.

In FIG. 1, for example, an analog signal such as a singing voice singed by one person or a single sound of an analog musical instrument, which is picked up by a microphone (not shown), is passed through a capacitor 1 for removing a DC component. The signal is input to the / D converter 2 and converted into a digital signal.

The CPU (Central Processing Control Unit) 3 has a RAM
(Random Access Memory) 4 as work memory, RO
By executing the program stored in the M (Read Only Memory) 9, information on the pitch of the input audio signal is extracted.

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

The ROM 9 has, in addition to the above-mentioned programs, various data used for pitch extraction, a table indicating the correspondence between the cycle of the obtained input signal and the note number, as described later. .

With the above-described circuit configuration, pitch data (hereinafter simply referred to as pitch in the abbreviated form) is extracted from an input audio signal in accordance with the basic operation principle described below. <Basic operating principle of the embodiment> Normally, a voice having a clear pitch feeling and being played as a single sound, such as a human singing voice, is repeated at every pitch cycle (for example) as shown in FIG. It is composed of substantially similar waveform elements (for example, a and d, b and e, c and f in the same figure).

Such a sound has many zero-cross points in one pitch cycle whose waveform crosses a reference level having an amplitude of 0 due to harmonic components contained in the sound.
Therefore, in order to determine the pitch period from such a periodic waveform, it is necessary to determine which of the zero-cross points should be the pitch period. It has the following features.

That is, the time of a section (for example) defined by the respective zero-cross points (for example, Z1 to Z2 and Z7 to Z8) of substantially the same waveforms (for example, a and d) of adjacent pitch periods, and two waveforms Each area (for example, a, d) (corresponding to the accumulated value of the A / D converted amplitude values)
Are almost equal.

By focusing on this point, the present invention finds two sections (for example,) having substantially the same time and waveform area as the time between adjacent zero-cross points, so that the start points (for example, Extract the time interval between Z1 and Z7) to extract the pitch period.

The operation of this embodiment, which operates according to the above-described basic operation principle, will be described below. <Outline of Embodiment> In FIG. 2, for example, the time of the section of the zero cross point Z1 to Z2 of the waveform a is counted, and the count value COUNT 1 is held in the first counter. Next, the time of each section of the waveforms b, c, d,... Having the same polarity as the waveform a is sequentially counted,
The count value COUNT 2 is held in the second counter while being sequentially updated until a value substantially equal to COUNT 1 is obtained.

In this case, the time from the zero-cross point Z1 to the zero-cross points Z3, Z5 and Z7 of the waveforms b, c and d is counted by the third counter. Since the count value COUNT 3 has a possibility of being the pitch cycle of the input voice signal, it is sequentially updated and stored in the variable PITCOUNTER in the RAM 4 until a count value that becomes the pitch cycle is obtained. Until the above section is found, A corresponding to the area of waveform a
The accumulated value of the / D-converted amplitude value is stored in the variable SU in the RAM 4.
The accumulated value of the A / D-converted amplitude values corresponding to the areas of the waveforms b, c, and d are written to the variable SUM 2 in the RAM 4 while being sequentially updated.

Then, a section having a time substantially equal to the time of the section is found, and the waveforms a,
When it is confirmed that the accumulated values of the amplitude values written in SUM 1 and SUM 2 corresponding to the area of d are substantially equal, the time corresponding to the time between the respective zero-cross points Z1 and Z7 of both sections is obtained. The counter value to be read is read from the memory variable PIT COUNTER, and is determined as the pitch cycle. And
A note number corresponding to the pitch cycle is obtained from a table in the ROM 9. <Operation of the Present Embodiment> Next, the operation of obtaining the pitch of the input waveform in accordance with the above-described operation principle will be described with reference to the operation flowcharts of FIGS. These operation flowcharts are realized as operations in which the CPU 3 of FIG. 1 executes a control program stored in the ROM 9.

In FIG. 3, first, an initial setting of the RAM 4 is performed. That is, a status flag (STATUS FLAG, see FIG. 2) indicating which state of the signal waveform corresponds to the time when the audio signal is read, a variable ENV indicating the size of the waveform envelope of the input audio signal, and a pitch. A variable PIT that holds the note number corresponding to the pitch obtained as a result of the extraction, a variable ON NOW that holds the currently sounding note number, and a count value of each of the first, second, and third counters. Variable to keep
All contents such as COUNT 1, COUNT 2, COUNT 3 and a variable ADMEM holding the previous A / D conversion value are cleared (step S301). When the values of PIT and ONNOW are 0, it indicates that there is no pitch-extracted sound or no currently sounding sound.

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

Thereafter, the process branches into three depending on the status flag (step S304). First, Figure 2
Since the status flag of is 0, the step
Proceed to S305. When the status flag is 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 is performed. It is determined whether or not reg1 in which the value is stored is positive.

In this case, the determination in step S305 is NO, and the process proceeds to step S307. Then, the value of reg1 in which the A / D conversion value is set this time is moved to the variable ADMEM (step
S307), envelope extraction (step S308), note-on / note-off processing of sound (sound generation / mute processing, step
The process proceeds to S309) and waits for the end of A / D conversion again (step S302).

The processes relating to the above-described envelope extraction (step S308) and musical note note-on / note-off (step S309) are shown in the respective operation flowcharts of FIGS.

In FIG. 5, first, the absolute value of the current A / D conversion value (the contents of reg1) is calculated (step S501).
Then, the input waveform is shaped by passing through a low-pass filter (by executing a digital filtering operation) (step S502). And the resulting value is the variable EN
V (step S503).

In the note-on / note-off process of FIG.
First, it is determined whether or not the value of the variable ENV exceeds a predetermined threshold (THRESHOLD) and the value of the variable PIT holding the note number is larger than 0 (step S604).

If the determination is NO, the variable ON NOW is cleared in order to mute all sounds being sounded by the sound source 5 in FIG. 1 and to indicate that there is no sound being sounded (step S1). S609), and returns to the main routine (FIG. 3).

If the determination in step S604 is YES, it is determined in step S605 whether there is any currently sounding sound. If the determination is YES, the flow advances to step S608.
If the determination in step S605 is NO, the variable ON NOW indicating the note number being sounded is compared with the value of the variable PIT indicating the extracted note number (step S606). As a result of the comparison, if the two match, there is no need to re-produce the sound, so the process returns to the main routine (FIG. 3). To the sound source 5 (step S607), and thereafter, the process proceeds to step S608.

In step S608, a sounding instruction is performed in accordance with the variable PIT indicating the note number, and the value of the variable PIT is substituted for the variable ON NOW.
It is shown that a tone having a pitch corresponding to is being produced (step S608). Then, the process returns to the main routine (FIG. 3).

The above processing is repeated until the zero cross point (Z1 in FIG. 2) from the negative side to the positive side is reached.
The determination in step S305 of FIG. 3 is YES, and the status flag STATUS FLAG is set to 1 (step S306).
Here, the count value COUNT 1 of the time of the section shown in FIG.
The initial value 1 is set to the count value COUNT 3 and the time or time, and the variable SUM 1 is set to the current value to obtain the accumulation of the A / D conversion value of the amplitude value of the input waveform being counted up in COUNT 1. Register re, which is the A / D conversion value of
The value of g 1 is written.

Thereafter, the same operation as described above is performed, and the process returns to step S304. Where STATUS FLAG = 1
Then, the process moves to step S310. In this state, in order to count the time exemplified in the section of FIG.
Incrementing the counter COUNT 1), also the third to count the time instantiated in the segment.
Of the variable COUNT 3 (hereinafter, referred to as a third counter COUNT 3) as the counter of the second counter is started.

First, in step S310, in order to detect a zero cross point from the positive side to the negative side of the input waveform, that is, a zero cross point Z2 which is the end point of the section in FIG. Is the value of the variable ADMEM where the value is written and the register A where the current A / D conversion value is written
It is determined whether the value of g 1 is negative.

If the determination is NO, the first counter
A register in which the value of COUNT 1 is increased by 1 and the A / D conversion value of the new input waveform is stored in the variable SUM 1
The value of reg1 is accumulated (step S311). In addition, the third
Counter COUNT 3 is incremented by 1 (step
S313).

Thereafter, steps S310 → S311 → S313 → S307 →
The loop processing of S308 → S309 → S302 to S304 → S310... Is repeated, and the first counter COUNT
The value of the first and third counter COUNT 3 is incremented by one.

When the determination in step S310 is YES and the zero-cross point Z2 from the positive side to the negative side is reached, the status flag is set to 2 (step S312). Then, the process proceeds to step S313, and after the value of COUNT 3 is increased by 1, the process proceeds to step S307 → S308 → S309, and step S302.
Return to

After the above-described loop processing is performed, FIG.
The counter value corresponding to the time of the section is the first counter
A, which is held in COUNT 1 and corresponds to the area of waveform a
The accumulated value of the / D conversion value is stored in the variable SUM 1.

Next, since the status flag is set to 2 in step S312, the process proceeds from step S302 to S304 this time to step S401 → S402 in FIG. In step S402, in order to detect a zero crossing point (Z3 in FIG. 4) from the negative side to the positive side, the same processing as in step S305 in FIG. 3, that is, the value of the variable ADMEM (the previous A / D conversion value ) Is negative,
It is determined whether the value of the register reg1 (current A / D conversion value) is positive.

While the determination is NO, S402 → S313
(Fig. 3) → S307 to S309 → S302 to S304 → S401 (Fig. 4) → S4
The loop processing of 02... Is repeated, and the count value of the third counter COUNT 3 is incremented by one at each repetition (S313).

When the determination in step S402 is YES and the zero-cross point Z3 from the negative side to the positive side is reached, the status flag is set to 3 and the value of COUNT 3 at this time (corresponding to the time in FIG. 2). ) Is stored in the variable PITCOUNTER (step S403). The value of this PITCOUNTER is held as a candidate for the cycle corresponding to the pitch to be extracted, that is, as a value that has a possibility of becoming the count value of the section in FIG. Further, the value of the second counter COUNT2 is set to 1.

Thereafter, steps S403 → S313 (FIG. 3) to S3
The process proceeds from 09 → S302 to S304 → S401 (FIG. 4). This time STATUS F
Since LAG = 3, the process proceeds from step S401 to step S404. Here, in order to detect a zero cross point from the positive side to the negative side of the input waveform, that is, a zero cross point Z4 corresponding to the end point of the section in FIG. 2, the value of the variable ADEM (1
It is determined whether the previous A / D conversion value is positive and the value of the register reg1 (current A / D conversion value) is negative (step S404).

While this determination is NO, step S404 → S4
05-> S313 (Fig. 3)-S309->S302-S304-> S401 (Fig. 4)->
The loop processing of S404 is repeated, and each time the processing of step S405 is performed in each repetition, the second counter COUN
The value 1 is added to T2, and the A / D conversion value stored in the register reg1 is accumulated in the variable SUM2. Thus, this time, the time of the section of the input waveform in FIG. 2 and the area of the waveform b (accumulated value of the A / D conversion value) are obtained. In addition, the value of the third counter COUNT 3 is also incremented by one for each repetition described above (S313).

If the determination in step S404 is YES, it means that the zero crossing point Z4 has been detected, and the status flag is changed to 4 here (step S406). Thereafter, the process proceeds to step S313 in FIG. 3, and after COUNT 3 is increased by 1, the process proceeds to steps S307 → S308 → S309 to S401 (FIG. 4).

This time, since STATUS FLAG = 4, the process proceeds from step S401 to step S407. In this state, the sections and are sequentially compared with the section in FIG. 2, and when a section having a time substantially equal to the section is found, the areas of the waveforms a and d corresponding to each of the sections are further added. The values of SUM1 and SUM2, which are the accumulated values of the A / D converted values of the corresponding input waveforms, are compared, and by judging that they are substantially equal, the time of the division of FIG. A processing operation determined as a pitch cycle is performed.

First, the first counter obtained so far
It is determined whether or not the respective values of COUNT 1 and the second counter COUNT 2 are substantially equal, that is, whether or not the absolute value of the difference between the respective values of COUNT 1 and COUNT 2 is within an allowable error range. (Step S407). This tolerance range is
It depends on the sampling frequency and pitch frequency of the input signal, but is determined empirically.

In this case, since COUNT 1 and COUNT 2 each hold the count value of the section in FIG. 2, the determination in step S407 is NO, and the process proceeds to step S408. Here, if noise or the like is mixed in the input audio signal, the determination of step S407 may result in an abnormally long state of NO for a long time, and the tone generation instruction may not be performed.

To prevent the occurrence of such an abnormal state,
The value of a third counter COUNT 3 which counts the time to be taken as the pitch period of the audio signal is constantly monitored, and a predetermined limit value whose value is much larger than the pitch period of the lowest tone of the audio signal. It is determined whether or not LIMIT has been exceeded (step S408).

This limit value is also similar to the above-mentioned allowable error range.
Determined empirically. For example, set the sampling frequency to 20KH
Assuming that z and the lowest extracted pitch (pitch) are 110.5 Hz (corresponding to the pitch name: A1), 20000 値 110.5 = 180.995...

If the determination in step S408 is YES, that is, if the state is abnormal, the flow advances to step S314 in FIG. 3 to instruct the sound source 5 in FIG. 2 to mute all sounds being generated. , The variable ON NOW is cleared. Thereafter, the process returns to step S301, and the above-described pitch extraction processing operation is performed again.

If the determination in step S408 is NO, the process proceeds to step S409, where the value of the second counter COUNT 2 is cleared, and the status flag is returned to 2. And
The process for obtaining the next zero crossing is started again. At this time, the value held in the variable PIT COUNTER becomes invalid, but the third counter COUNT 3 is still counting up, and the first counter COUNT 1 is also shown in FIG.
The count value of the section is held.

As described above, the status flag 2
After returning to the state, the status flag advances to the state of 3 and 4 as in the previous case, and the second counter COUNT
2 holds the count value of the section in FIG. 2, and the variable PIT COUNTER holds the count value (of FIG. 2) from the start of the section to the zero crossing point Z5 of the section. Also, the variable
The accumulated value of the A / D converted value written in SUM 2 is cleared, and each A / D converted value reg 1 of waveform c is sequentially accumulated in SUM 2.

However, the value of COUNT 1 (corresponding to the time of the section) and the value of COUNT 2 (corresponding to the time of the section) do not match, so the determination in step S407 is NO. Therefore, the process proceeds to step S409 via step S408, and the status flag is re-executed from the state of 2.

Then, when the status flag returns to the state of 4 again, COUNT 2 holds the count value of the section of FIG. 2, while PIT COUNTER holds the time count value of FIG. doing. Then, in step S407, YES is determined for the first time,
Proceed to S410.

In step S410, the accumulated value of the A / D conversion value corresponding to the area of the waveform a held in the variable SUM 1 and the area of the waveform d held in the variable SUM 2 are further obtained. It is determined whether or not the accumulated values of the A / D conversion values substantially match.

If this determination is NO, it is determined that the waveform does not correspond to the correct pitch period, and the flow advances to step S408 to execute the process again from the state of STATUS FLAG = 2 and search for a new waveform. Is done.

On the other hand, if the determination in step S410 is YES, it is determined that a waveform corresponding to a correct pitch period has been obtained, and the flow advances to step S411. The correct value of the pitch cycle of the input waveform at this time is the value of the third counter COUNT 3 (FIG. 2) held in PIT COUNTER. Then, the note number corresponding to this value is read from the table in the ROM 9 of FIG. 1, and is moved to the variable PIT (step S410).

Next, the status flag, all the counters, and the contents of the variables SUM 1 and SUM 2 are cleared and reset to the initial state (step S412). Thereafter, the process proceeds to the note on / off processing of steps S307 to S309 in FIG. 3, in which a new tone generation instruction of a musical tone having a note number corresponding to the pitch (pitch) of the input audio signal is performed. .

That is, in the flow of FIG. 6 corresponding to step S309, after the determination in step S604 is YES, if it is determined in step S605 that there is no currently sounding sound (YES), as described above, In step S608, a sound generation instruction is performed according to the variable PIT indicating the note number. If it is determined in step S605 that there is a currently sounding sound (NO), as described above, in step S606, the variable ON NOW indicating the sounding note number and the variable PIT indicating the extracted note number are used. Each value is compared. As a result of the comparison, if the two match, there is no need to re-sound, so the process returns to the main routine (FIG. 3). If the two do not match, the note number to be sounded has changed. The mute is instructed to the sound source 5 (step S607), and then in step S608, a new sounding instruction is performed according to the variable PIT in which the new note number is stored.

After the above sound generation instruction, step S307 in FIG.
The process proceeds to S302 to S304, and the same operation as before is repeated from the state of STATUS FLAG = 0. In the embodiment described above, the waveform portion on the positive side (for example, waveforms a, b, c, d... In FIG. 2) with respect to the reference level of the amplitude value 0 of the waveform.
The pitch was extracted by comparing the time of each section with the accumulated value of the A / D conversion value. However, the present invention is not limited to this. Actions can be taken.

Further, the above-described pitch extraction operation is performed in parallel with respect to each of the waveforms on the positive side and the negative side with respect to the reference level, and the value of each third counter COUNT 3 allows a predetermined error. If they match, for example, two COUNs
If the average value of the values of T 3 is obtained as the pitch cycle of the input waveform, more accurate pitch extraction is possible.

In the note on / off processing of FIG. 6, if the pitch of the musical tone being produced changes,
In this case, the sound is re-produced after the sound is muted. In this case, for example, if the envelope does not change so much, the sound source 5 of FIG. It is also possible to give instructions.

[0076]

According to the present invention, by extracting the time points at which the intervals between the zero-cross points are substantially equal and the peaks of the waveform are determined to be similar, the pitch of the input waveform can be reduced irrespective of the influence of harmonic components. It is possible to extract stably. In this case, in particular, not only the interval between the zero-cross points but also the accumulated amplitude value corresponding to the area of the peak of the waveform is used as the determination element, thereby enabling more accurate pitch extraction.

As a result, it is possible to reliably extract the pitch indicating the pitch from a voice waveform such as a singing voice or a musical tone including harmonics.

[Brief description of the drawings]

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 the embodiment.

FIG. 4 is an operation flowchart (part 2) of the present embodiment.

FIG. 5 is an operation flowchart relating to envelope detection according to the present embodiment.

FIG. 6 is an operation flowchart relating to note on / off according to the present embodiment.

[Explanation of symbols]

 Reference Signs List 1 capacitor 2 A / D converter 3 CPU 4 RAM 5 sound source 6 D / A converter 7 amplifier 8 speaker 9 ROM

──────────────────────────────────────────────────続 き Continued on the front page (58) Fields surveyed (Int. Cl. ⁷ , DB name) G10L 11/00-21/06 JICST file (JOIS)

Claims (2)

(57) [Claims] An interval from a zero-cross point at which the input waveform zero-crosses from a negative or positive first polarity to another second polarity to a zero-cross point at which the input waveform zero-crosses from the second polarity to the first polarity. To measure the first interval as the first interval
Measuring the accumulated value of the absolute value of each amplitude of the input waveform as the first amplitude accumulated value while the first interval measuring means measures the first interval. After completion of the measurement by the first amplitude accumulated value measuring means and the first interval measuring means, the zero-crossing point at which the input waveform zero-crosses from the first polarity to the second polarity is used. A second interval measuring means for sequentially measuring an interval of each section from the second polarity to the zero-cross point at which the first polarity crosses to the first polarity as a second interval, and the second interval measuring means comprising: A second amplitude accumulated value measuring means for sequentially measuring the accumulated value of the absolute value of each amplitude of the input waveform as a second amplitude accumulated value in each section in which the measurement of the interval is sequentially performed; Zero at which measurement by the first interval measuring means is started A third interval measuring means for sequentially measuring the distance from Ross point, the said second measuring means each second
At each time point when the measurement of the interval is started, an interval storage unit that stores the interval measured by the third interval measuring unit while sequentially updating the interval information as cycle information, and the second unit that is measured by the first interval measuring unit. The first interval and the second intervals sequentially measured by the second interval measuring means coincide with each other within a predetermined allowable error range, and the first amplitude measured by the first amplitude accumulated value measuring means. Determining means for sequentially determining whether or not the accumulated amplitude value and each of the second accumulated amplitude values sequentially measured by the second accumulated amplitude value measuring means coincide with each other within a predetermined allowable error range; When the determination unit detects the coincidence, the period information stored in the time storage unit is output as pitch information of the input waveform. By repeating the above Pitch extraction apparatus characterized by comprising: a repetition control means for sequentially extracting pitch information from the force waveform, a. 2. The method according to claim 1, wherein the time interval measured by the third interval measuring means exceeds an interval corresponding to a preset pitch cycle of the lowest note at the time when the determining means does not detect the coincidence. And if it is determined that it has exceeded,
2. The pitch extracting device according to claim 1, further comprising a limiter that restarts each of the processes from the measurement process by the first interval measuring device.