JPH06300793A - Fundamental frequency detecting method for ac signal - Google Patents

Abstract

PURPOSE:To detect a fundamental frequency by a relatively simple circuit in a short time by selecting one by one from a low frequency side of divided bands of each hierarchy as a passband of an input AC signal, comparing a level of a passed signal with a reference level, and deciding whether the fundamental frequency exists in the passband or not. CONSTITUTION:An AC signal to be input is ADD-converted by an A/D converter 1, and supplied to a variable characteristic IIR filter 2. The filter 2 exhibits band pass characteristics to be decided with BPF coefficient data to be input. A signal passed through the filter 2 is supplied to one input terminal of a level comparator 5 through a detector 3 and an LPF 4. The comparator 5 compares data from a criterion level data generator 6 with input signal data, and supplies a logic '1' signal if the voltage level exceeds the criterion level. Further, a CPU 7 supplies the BPF coefficient data for deciding its passband characteristics to the filter 2, and simultaneously sends a criterion level set command to the generator 6.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of detecting a fundamental frequency of an AC signal such as an audio signal.

[0002]

2. Description of the Related Art As a fundamental frequency extraction device for a voice signal,
For example, connecting many BPFs in parallel and
There is a method in which the pass band for the input signal is switched in order to compare the output level with a predetermined reference level and the frequency corresponding to the BPF when the output level becomes maximum is used as the detection fundamental frequency.

[0003]

In such a conventional fundamental frequency extraction device, the number of BPFs corresponding to the required resolution is required, and the number of changeover switches equal to the number of BPFs is also required. Became complicated and costly. Therefore, a large number of B's are used by using a characteristic variable filter as disclosed in Japanese Patent Laid-Open No. 61-26089.
It may be possible to substitute the switching connection of the PF, but it is not practical because the time required for determining the passing signal component for each switching times the number of times of switching is required for detecting the fundamental frequency.

SUMMARY OF THE INVENTION An object of the present invention is to provide a method of detecting a fundamental frequency of an AC signal, which has a relatively simple circuit configuration and can detect the fundamental frequency in a short time.

[0005]

A fundamental frequency detecting method of an alternating current signal according to the present invention is a method of detecting a fundamental frequency of an input alternating current signal, in which a detection frequency range is k (k is an integer of 2 or more) Is divided into a plurality of layers for each n (n is 1 ≦
(an integer of n ≦ k), a first step of setting a sub-layer split band and making a plurality of (n + 1) -th sub-layer split bands correspond to each of the n-th sub-layer split bands; Each divided band is designated in order from the low frequency side, and this is designated as a pass band for the input AC signal, and the designated n-th hierarchical layer divided band when the level of the pass component of the input signal passing through the pass band reaches a predetermined level is n. The second step of setting the next detection band and (n +
1) Designation (n + 1) -th layer division when the level of the pass signal component reaches a predetermined level by designating one division band in order from the low frequency side only for the next layer division band and using this as a pass band for the input signal The third step of setting the band to the (n + 1) th order detection band, and the third step of setting the (n + 1) th order detection band as a new nth order detection band, unless the new nth order reaches a predetermined upper limit order. It is characterized by comprising a fourth step to be repeated.

[0006]

According to the fundamental frequency detecting method of the alternating current signal of the present invention, one is selected from the lower frequency side among the divided bands for each layer, and the level of the passing signal is used as the pass band of the input alternating current signal. As compared with the level, it is determined whether the fundamental frequency is within the pass band.
Therefore, the divided band in which the fundamental frequency exists can be detected in a relatively short time.

[0007]

FIG. 1 shows a fundamental frequency detecting apparatus for carrying out the AC signal fundamental frequency detecting method according to the present invention. In this device, an AC signal such as an audio signal supplied to an input terminal is converted into a 16-bit PCM digital signal by the A / D converter 1 to have a variable characteristic IIR.
(Infinite Impulse Response) It is supplied to the filter 2. The variable characteristic IIR filter exhibits a bandpass characteristic determined by the input BPF coefficient data. The variable characteristic IIR filter is a so-called DSP (Digital Signal Pro).
cessor). The signal that has passed through the variable characteristic IIR filter 2 is a digital detector 3 and a digital LP.
It is supplied to one input terminal of the level comparator 5 via F4. The level comparator 5 compares the discrimination reference level data from the discrimination reference level data generation circuit 6 with the input signal data, and when the voltage level represented by the input signal data exceeds the discrimination reference level, a logical "1" signal. CP
Supply to U7. The CPU 7 reads various programs such as a subroutine program, which will be described later, from the ROM 8, executes the programs, writes the data generated by the execution of these programs to the RAM 9, and reads the written data as necessary. The CPU 7 accepts the manual input data or the operation command from the keyboard 10, while drivingly controlling the output means such as the printer 11 as necessary.

Further, the CPU 7 supplies the BPF coefficient data for defining the pass band characteristic to the variable characteristic IIR filter 2 in the pass band setting subroutine described later, and at the same time, issues the discrimination reference level setting command corresponding to the pass band characteristic. It is supplied to the discrimination reference level data generation circuit 6. The flowcharts of FIGS. 2 and 3 show the pass band setting subroutine described above. First, in this flowchart, each of the passbands to be set is represented by B (N1,
N2, N3, N4). Where N
1 to N4 represent the band numbers of the respective layers when the entire target frequency band is divided into first to fourth layers as shown in the table of FIG. That is, N1 is 1
And can take the values of 2. N2 is a value of 1 and 2, N
3 can take the values 1, 2 and 3, and N4 can take the values 1 and 2. In this way, for example, the pass band No. 5
Is represented as B (1,1,3,1),
Each pass band No. 1 to 24 are B (N1, N2, N3,
N4) is represented by the general formula.

The arrow in FIG. 4 schematically shows the flow of detection described below, the upward arrow is the flow when it is detected, and the downward arrow is the flow when it is not detected. . Also, the dashed arrow indicates an optional flow. In this subroutine, first, the value of N1 is set to 1, and N2 to N4 are set to 0 (step S
1).

Here, it is assumed that B (N1,0,0,0) represents the band of the primary hierarchy. That is, B (1,0,
0,0) indicates the No. 1 band of the primary hierarchy in FIG. 4, and B (2,0,0,0) indicates the No. 2 band of the primary hierarchy. Then, the BPF coefficient data is changed to the variable characteristic IIR filter 2 so that the variable characteristic IIR filter 2 has the pass band characteristic corresponding to B (N1,0,0,0).
On the other hand, the command is supplied to the discrimination reference level data generating circuit 6 to set the discrimination reference level corresponding thereto (step S2). Then, it is determined whether the output of the level comparator 5 is logic "1" (step S3). If the output of the level comparator is logic "0", N1 is set to N
As a value of 1 + 1 (2 in this case) (step S4), N
It is determined whether 1 exceeds the maximum value (step S
5). In the example of FIG. 4, N1 has a maximum value of 2, and
Since it does not exceed 2, return to step 2. If N1 = 3 at this time, the maximum value of N1 is 2
Thus, the CPU operation returns to the main routine (not shown). The main routine is, for example, a routine that is repeatedly executed by the CPU in synchronization with a clock pulse, and the subroutine of FIGS. 2 and 3 is executed by interrupting the main routine.

When it is determined in step S3 that the output of the level comparator is logic "1", the band number N2 of the secondary hierarchy is set to 1 (step S6). Then B
The pass band characteristic corresponding to (N1, N2, 0, 0) and the discrimination reference level corresponding thereto are set, and each data is supplied to the variable IIR filter 2 and the discrimination reference level data generating circuit 6 (step S7). ). Then, it is determined whether or not the output of the level comparator 5 has become a logic "1" (step S8). If the output of the level comparator 5 is logic "0", N2 is set to N2 + 1 (step S
9) Then, it is determined whether or not the value of N2 is equal to the maximum value (2 in the case of the example of the table of FIG. 4) (step S10).
Unless N2 reaches the maximum value, the process returns to step S7. If N2 reaches the maximum value, the fundamental frequency of the input signal exists in the band of the maximum value of N2 for the second hierarchical layer. Since this is clear, the process directly shifts to the determination of the next order, that is, the third order hierarchy. Of course, when it is determined in step S8 that the logic output of the level comparator is "1", the process proceeds to the determination of the third order hierarchy.

In determining the third hierarchical level, N3 is set to 1 (step S11). And B
The BPF coefficient data corresponding to (N1, N2, N3, 0) is supplied to the variable IIR filter 2, and the discrimination reference level data is supplied to the discrimination reference level data generation circuit 6 (step S12). Next, it is determined whether or not the output of the level comparator 5 has become logic "1" (step S13).
If this determination step 13 determines the band B determined by the value
If it can be detected that it is within (N1, N2, N3, 0), the process proceeds to the determination of the next quaternary layer. If the logical "1" cannot be determined in this determination step, N3 is set to N3 + 1 (step S14), and N3 is set.
Determines whether has reached the maximum value (step S1)
5). Then, unless N3 reaches the maximum value, the process returns to step S12, and when N3 reaches the maximum value, the process shifts to the determination of the next quaternary layer.

In determining the fourth hierarchical level, first, N4
Is set to 1 (step S16). And B (N1, N
2, N3, N4), the pass band characteristic corresponding to the variable characteristic I
While supplying the BPF coefficient data to the IR filter 2, the discrimination reference level data of the discrimination reference level data generating circuit 6 is set (step S17). Next, it is judged whether or not the output of the level comparator 5 is logic "1" (step S18).

If it is determined in step S18 that the output of the level comparator 5 is "1", the fundamental frequency of the input signal corresponds to the BPF coefficient data to the filters N1 to N4 at that time. If the control signal can be supplied to the outside, it can be used as the fundamental frequency detector 20 shown in FIG. The fundamental frequency detector 20 supplies data representing the fundamental frequency band of the input signal to the fundamental frequency signal generator 21, while also supplying the second overtone, fourth overtone and eighth overtone signal generators 22, 23, 24. The signals from the signal generators 21 to 24 are supplied to the adder 29 via the variable gain amplifiers 25 to 28, and the signals added there are output as a musical tone of a desired tone color although the tone is equal to the pitch of the input signal. It will be done. It should be noted that sounds lower than the fundamental frequency or sounds that are not overtone may be combined.

In the above embodiment, the n-th layer is a quaternary layer, but it is naturally conceivable that n is a value larger than 4. In the above embodiment, the determination step corresponding to the broken line shown in FIG. 4 is omitted as the detection procedure. As described above, when a signal component is detected in a divided band of a certain layer, the band of the last one of the bands obtained by dividing the divided band in the next layer is further divided. This is based on the assumption that when no signal component is detected until immediately before, there should be a signal component in the remaining last band.

However, in an actual usage environment, it is possible to make an erroneous determination due to the influence of noise exhibiting a specific frequency distribution, the influence of the level fluctuation of the input signal, or the like. To counter this, the flow can be modified. For example, if the flows of to are enabled and the flows of and are disabled in the second hierarchical band of FIG. 4, when the fundamental frequency is in the band B (2,0,0,0), the band B is In the case of (1, 0, 0, 0), even if it is erroneously determined that there is a signal due to noise, an opportunity for re-determination may occur.

In this case, if it is determined that there is no layer in the primary layer and then the program is returned to enter the secondary layer as described above, the signal level in the primary layer is low. 2 even if not detected
Since there is a possibility that it can be detected in the next layer, the accuracy is further improved. Furthermore, the above-mentioned procedure can be applied to other layers, but it may be appropriately selected in consideration of the increase in detection time.

Further, a desired frequency can be designated by the keyboard 10 and applied to the analysis of the fundamental frequency. From the key input section, for example, minimum frequency 1000Hz, maximum frequency 22
If you specify linear division of 70Hz and resolution 10Hz, CPU
At 128, the total number of divisions is 128. Then this is 2 ⁿ × 3
Obtain n and m when expressed in the form of ^m (n and m are 0 or positive integers). In this example, n = 7 and m = 0.
That is, the number of layers in this case is 7. Next, the upper and lower limits of each divided band are obtained. In this example, the lowest is 995 to 1005 Hz, 1005 to 1015 Hz, ..., 2265 to 2275 Hz.

The upper and lower limits of each divided layer are also obtained. For example, the first division is 2 divisions, so 995 to 1635Hz and 1635
~ 2275Hz. Next, the BP corresponding to each of these bands
The F coefficient data and the discrimination reference data are obtained, and the fundamental frequency is detected by the above procedure.

[0020]

As is apparent from the above, the fundamental frequency detection method for an AC signal according to the present invention divides the band to be detected into a plurality of layers, and sequentially descends from the lower layer to the higher layer. The configuration is such that the band to which the input signal fundamental frequency belongs for each layer is determined from the low frequency side. Since the number of determinations is reduced, the time required for detection is shortened.

[Brief description of drawings]

FIG. 1 is a block diagram showing a fundamental frequency detection device for an input AC signal according to the present invention.

FIG. 2 is a flowchart showing a subroutine operation of a CPU of the fundamental frequency detection device for an input AC signal of FIG.

FIG. 3 is a flowchart showing a subroutine operation of a CPU of the fundamental frequency detection device for the input AC signal of FIG.

FIG. 4 is a table showing how the target frequency band is divided into layers.

5 is a block diagram showing an application example of the apparatus of FIG.

[Explanation of symbols for main parts]

 21 Basic Frequency Signal Generator 22 2nd Overtone Signal Generator 23 4th Overtone Signal Generator 24 8th Overtone Signal Generator 25-28 Variable Gain Amplifier 29 Adder

Claims (2)

[Claims] 1. A method for detecting a fundamental frequency of an input AC signal, wherein a detection frequency range is divided into a plurality of k (k is an integer of 2 or more) next layers and each n (n is 1 ≦ n). ≦ k) a first step of setting a sub-layer sub-band and setting each of the n-th sub-layer sub-bands to correspond to a plurality of (n + 1) -th sub-layer sub-bands; Each of the bands is designated in order from the low frequency side, which is used as a pass band for the input AC signal, and the designated n-th hierarchical division band when the level of the pass component of the input signal passing through the pass band reaches a predetermined level A second step of setting a detection band, and only one (n + 1) -th layer hierarchical division band corresponding to the n-th detection band is designated in order from the low frequency side by one division band, which is used as a pass band for the input signal. Ingredient Designation when the level reaches a predetermined level (n +
1) Third in which the next hierarchical division band is the (n + 1) th detection band
And a fourth step in which the (n + 1) th order detection band is set as a new nth order detection band and the third step is repeated unless the new nth order reaches a predetermined upper limit order. Method of detecting fundamental frequency of alternating current signal. 2. In the second and third steps, when the pass signal component level does not reach the predetermined upper limit level for the division bands other than the remaining one division band of the division bands of the nth or (n + 1) th hierarchical layer. The method of claim 1, wherein the remaining one divided band is used as a detection band.