JPH0222520A - Pitch extractor - Google Patents

Abstract

PURPOSE:To enable accurate pitch extraction which is small in the influence of higher harmonic components of high order by providing an A/D conversion part, a memory part, and an extreme value detection part. CONSTITUTION:An acoustic signal which is inputted to an HPF 1 is processed by high-pass filtering for removing the lower frequency component than a pitch frequency and then outputted to an LPF 2. Then this acoustic signal is processed by low-pass filtering for a sampling frequency so as to satisfy a sampling theorem. Then, the acoustic signal after passing through the HPF 1 and LPF 2 is A/D-converted as a band-limited acoustic signal by an A/D conversion part 3 at the sampling frequency into digital data. This data is outputted to a memory part 7, an extremal value detection part 8, and a polarity inversion detection part 9 in order. Thus, the high-accuracy pitch extraction which is small in the influence of high-order higher harmonic components.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to a pitch extracting device for acoustic signals such as various types of music, sounds, and voices.

Conventional technology In recent years, with the advancement of digital technology, various pitch extraction techniques have been proposed, including techniques for obtaining autocorrelation of time waveforms and cepstrum techniques, but these techniques are aimed at reducing the amount of calculation or high-speed processing. As a technique, a pitch extraction technique using polarity reversal detection is being considered.

The above-mentioned pitch extraction device will be described below with reference to the drawings.

FIG. 4 shows the configuration of a conventional pitch extraction device.

In FIG. 4, 1 is a bypass filter (HPF), 2
is a low-pass filter (LPF), 3 is an A/D converter that converts analog audio signals into digital data, and 4 is an A/D converter.
a polarity reversal detection section that detects polarity reversal of the digital data output from the D conversion section 3 and outputs a polarity reversal detection signal;
Reference numeral 5 denotes a control unit that outputs a timing instruction signal for instructing the start or end of timing of a pitch period according to the polarity reversal detection signal outputted from the polarity reversal detection unit 4; 6, a control unit that outputs a timing instruction signal that instructs to start or end timing of a pitch period; This is a timing section that measures time.

The operation of the pitch extraction mounting configured as above will be described below.

An acoustic signal input from a microphone or the like has its DC component removed by a bypass filter 1, and then passes through a low-pass filter 2 to undergo anti-aliasing filtering that satisfies the sampling theorem.
/D converter 3 performs 1[i-th analog-to-digital conversion and becomes digital data.

Next, the polarity reversal detection section 4 detects the polarity reversal of the digital data sequentially output from the A/D conversion section 3 and then outputs a polarity reversal detection signal to the control section 5.

The control unit 6 sends a timing instruction signal to the timing unit 6 to instruct the start or end of timing of the pitch period according to the polarity reversal detection signal.
Output to. When the control section 5 outputs a timing instruction signal instructing the start of timing, the timing section 6 starts counting the number of samples of digital data, that is, timing the pitch period.

Furthermore, when a timing instruction signal instructing the end of timing is output from the control unit 5, the timing unit 6 outputs the number of samples up to that point, that is, the timing result, to the control unit 5 as a pitch period, and then resets the timing result. By repeating the above operations according to the polarity reversal detection signal from the polarity reversal detection section 4, the pitch period of the digital data sequentially output from the A/D conversion section 3 is obtained.

Incidentally, in order to convert the pitch period Ni to the so-called pitch FiCHz, the sampling frequency Fs[Hz] can be used, and it can be obtained as F i [:Hz] = Fs/N 1 = (1).

Problems to be Solved by the Invention However, with the above configuration, as is clear from equation (1), not only does the accuracy of the pitch period obtained change depending on the sampling frequency, but also the sampling frequency must be changed to increase the accuracy. Therefore, there was a problem that high-speed arithmetic processing was required, resulting in an expensive device.

Furthermore, since only polarity inversion of digital data is used, there are cases where high-order harmonic components such as second overtones are erroneously extracted as pitch periods.

In view of the above problems, the present invention provides a highly accurate pitch extraction device that does not require high-speed arithmetic processing, has a simple configuration, and is less affected by high-order harmonic components.

, Means for Solving the Problems To achieve this object, the pitch extracting device of the present invention has the following features:
A converting band-limited acoustic signals into digital data
/D converter; a memory unit that temporarily stores at least two pieces of digital data output from the A/D converter; and a local maximum or minimum value of the digital data output from the A/D converter. An extreme value detection section that detects the value and outputs an extreme value detection signal, and a polarity reversal detection section that detects polarity reversal of digital data according to the extreme value detection signal and outputs a timing instruction signal that instructs the control section to time the pitch period. and a control section that starts or ends timing the pitch period in accordance with a timing instruction signal, and reads out and interpolates digital data whose polarity has been inverted from the memory section.

Effect: With this configuration, the band-limited acoustic signal is converted into digital data by the A/D converter. Digital data output from the A/D converter is sequentially sent to a memory section, an extreme value detector, and a polarity reversal detector.

At this time, the extreme value detection section outputs an extreme value detection signal to the polarity reversal detection section every time it detects a maximum value or a minimum value of the digital data that are sequentially sent out. The polarity reversal detection section detects the polarity reversal of the digital data that is sequentially sent out based on the extreme value detection signal, and outputs a timing instruction signal that instructs the timing of the pitch period to the control section, and then the digital data is processed by the control section. As the number of samples is counted, that is, the pitch period is timed, the control unit reads out at least two pieces of digital data with inverted polarity from the memory unit according to the sent clock instruction signal, calculates the zero crossing point by interpolation, and calculates the zero crossing point by interpolation. The sum of the results and the timing results is output as the pitch period.

By repeating the above operations, the sequentially detected pitch periods are output from the control section.

EXAMPLE Hereinafter, an example of the present invention will be described from the beginning with reference to the drawings.

FIG. 1 shows a block diagram of a pitch extraction device in one embodiment of the present invention.

In FIG. 1, 7 is a memory section, 8 is an extreme value detection section, and 10 is a memory section.
is the control section. Note that 1 is a bypass filter, 2 is a low-pass filter, and 3 is an A/D converter, which are the same as the conventional configuration.

The operation of the pitch extraction device configured as described above will be described below.

First, the acoustic signal input to the bypass filter (HPF) 1 is subjected to high-pass filtering that removes frequency components considerably lower than the pitch frequency (for example, below 50 [Hz]), and then filtered to a low-pass filter (HPF). (LP
F) Output to 2. The acoustic signal after high-pass filtering input to the low-pass filter 2 has a sampling frequency Fs so as to satisfy the so-called sampling theorem.
Low-pass filtering is applied to [:Hz] so that only Fs/2 [Hzl or less] is allowed to pass. In this embodiment, it is preferable to remove the influence of high frequency components by performing low pass filtering with a cutoff frequency of Fc<<Fs/2.

The acoustic signal that has passed through the bypass filter 1 and the low-pass filter 2 becomes digital data that has been A/D converted by the A/D converter 3 at a sampling frequency of Fs [Hz] as a band-limited acoustic signal. The digital data outputted from the A/D conversion section 3 is, for example, as shown in FIG. 3, and is sequentially outputted to the memory section 7, the extreme value detection section 8, and the polarity reversal detection section 9.

Here, the section shown in Fig. 3 d is the approximate pitch, and Fig. 3 f
It is generally known that the interval is easily influenced by formants. Therefore, the section f in FIG. 3 is extremely unstable because it changes variously depending on the phoneme, and is considered to be a major cause of erroneous pitch extraction.

Therefore, in order to detect only the section shown in FIG. After detecting the extreme value or minimum value of the input digital data, it outputs an extreme value detection signal to the polarity reversal detection section 9.

Next, when the extreme value detection signal is output, the polarity reversal detecting section 9 detects the reversal of the polarity of the digital data that is sequentially input, and outputs a timing instruction signal that instructs the control section 10 to time the pitch period.

At this time, if the control unit 10 has not already started counting the number of samples of digital data, that is, measuring the pitch period, it starts counting the number of samples of digital data that are sequentially input, and also inverts the polarity from the memory unit 7. For example, the two digital data yot and y11 as shown in FIG. 2(a) are read out, and the zero crossing point Not is calculated using an interpolation formula as shown in equation (2).

N01=Y11/(Y11+IYO11)・・・・・・
...(2) Alternatively, if the pitch cycle timing has already started, the counting of digital data is finished, and two digital data Y with inverted polarity are stored from the memory section 7, for example, as shown in FIG. 2(b).
O2 and Y12 are read out, and the zero crossing point N02 is calculated using an interpolation formula as shown in equation (3).

N02= lYO2+/ (Y12+lYO21)...
(3) Furthermore, when the control unit 10 finishes counting, it calculates the count value, that is, the number of samples N, No. as shown in FIG. 2(C).
After adding t, NO2, and NO2, the count value is reset to zero and the addition result is output as a pitch period.

Note that when the pitch period is expressed as a time value, the detected value may be multiplied by 1/Fs, and when the pitch frequency is expressed as a pitch frequency, Fs may be divided by the detected value.

In addition, in the above, regarding the interpolation formula in the control unit 10, we have described formula (2), that is, when the polarity of digital data is reversed from a positive value to a negative value, as the start of time measurement. It goes without saying that the same effect can be obtained even if the timing is started at the time when the polarity of .

Further, in this embodiment, the zero crossing points of two digital data whose polarities are inverted are obtained by linear interpolation, but function approximation such as curve interpolation may also be used.

As described above, according to this embodiment, the polarity reversal detecting section 9 detects the polarity reversal immediately after the extreme value of the digital data detected by the extreme value detecting section 8, and the control section 10 detects the polarity reversal of the two digital data. Since the zero crossing point is calculated using The maximum value or minimum value of the digital data converted by the /D converter is detected by the extreme value detector, and the polarity reversal immediately after the extreme value is detected is detected by the polarity reversal detector and synchronized with the output clock instruction signal. The control unit counts the number of samples of digital data that are sequentially inputted by the control unit to obtain the approximate period of the pitch, reads at least two or more digital data whose polarity is inverted at both ends of the approximate period from the memory, and interpolates the zero crossing point. Each calculation is performed by calculation, and the result of adding these results and the approximate period is output as the pitch period, so the influence of high-order harmonic components is small, and there is no variation in extraction accuracy depending on the sampling frequency. You can get the pitch.

In addition, since the zero crossing points between sample points are found, there is no need for a high sampling frequency as in the past to improve pitch extraction accuracy, so there is no need for high-speed arithmetic processing and it can be realized with inexpensive hardware. Become.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a pitch extraction device according to an embodiment of the present invention, FIGS. 2 and 3 are exemplary waveform diagrams of each data according to an embodiment of the present invention, and FIG. 4 is a block diagram of a pitch extraction device according to an embodiment of the present invention. FIG. 2 is a block diagram of an extraction device.

Claims (1)

[Claims] A converting band-limited acoustic signals into digital data
/D conversion section, a memory section that temporarily stores at least two or more data of the digital data output from the A/D conversion section, and a maximum of the digital data output from the A/D conversion section. an extreme value detection section that detects a value or a minimum value and outputs an extreme value detection signal; and an extreme value detection section that detects polarity reversal of the digital data according to the extreme value detection signal and outputs a timing instruction signal that instructs timing of a pitch period. a polarity reversal detection unit that detects the extreme value; and a control unit that starts or ends timing the pitch period according to the timing instruction signal, and reads digital data from the memory unit and interpolates it, and the control unit detects the extreme value. After the signal is output, add the approximate pitch period obtained by counting the number of samples of the digital data and the fractional pitch obtained by interrogating at least two or more data read from the memory section. A pitch extraction device featuring: