JPH08272390A - Interval conversion device - Google Patents

Abstract

PURPOSE: To provide an interval conversion device the distortion of whose output signal is less and of which the S/N ratio is excellent. CONSTITUTION: This device is provided with an interval input means 4 inputting a target interval, an interpolation coefficient control means 14 outputting an interpolation coefficient based on the inputted target interval and an interval conversion means 3 adjusting a data length of an input digital signal so as to convert interval to the target interval and converting interval for the input digital signal by interpolating with the interpolation coefficient outputted from the interpolation coefficient control means 14.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pitch converting apparatus for converting a pitch of an acoustic signal into an arbitrary pitch in AV (audio / visual) equipment.

[0002]

2. Description of the Related Art Conventionally, various devices have been proposed for arbitrarily converting pitches. A key controller used in karaoke equipment is a typical example. In these dedicated devices, pitch conversion is realized by processing such as changing the reproduction speed in analog processing and changing the sampling frequency in digital processing.

In recent years, due to the diversification of karaoke software, it has become necessary to easily convert the pitch not only by a dedicated player but also by various general devices. Since it is difficult to change the reproduction speed and the sampling frequency with these devices, the pitch conversion is realized by performing signal processing using a DSP or the like. This technique will be described below with reference to the drawings.

FIG. 25 is a block diagram showing the structure of a conventional pitch converting apparatus, and FIG. 26 is a block diagram showing the pitch converting section of the pitch converting apparatus. 25 and 26, 1 is an input signal, 2 is an A / D converter, 3 is a pitch converting means, 4 is a pitch inputting means, 5 is a D / A converter, 6 is an output signal, 11 is a data length adjustment It is a means.

The input acoustic signal 1 is converted into a digital signal by the A / D converter 2. If the input signal is sampled at the same sampling frequency as the A / D converter 2, the A / D converter 2 is not necessary. This digital signal is input to the pitch conversion means 3, and the target pitch to be changed is also input to the pitch input means 4. Pitch input means 4
Sends to the pitch converting means 3 a control signal having the inputted pitch. Then, the pitch converting means 3 receives the control signal and converts the pitch of the signal from the A / D converter 2. As shown in FIG. 26, the pitch converting means 3 is composed of a data length adjusting means 11.

Here, the conversion of the pitch is performed by the following procedure.

For example, a case of converting an input signal of frequency f _{1 into} an output signal of frequency f ₂ by pitch conversion will be described with reference to the drawings. In FIG. 27A, the frequency f
Converting an input signal of _{1 into} an output signal of frequency f ₂ is performed by converting each signal band to αf as shown in FIG.
₁ and αf ₂ , the signal band on the frequency axis is (f ₂ /
f ₁ ) Equivalent to multiplication. In other words, this is equivalent to multiplying the sampling frequency by (f ₁ / f ₂ ). When this is considered on the time axis, it becomes as shown in FIG. T ₁ and T ₂ in the figure are represented by T ₁ = 1 / f ₁ T ₂ = 1 / f ₂ .

In FIG. 27, as an example, (n + 1) T ₂ = n
The case where it becomes T ₁ is shown. This can be expressed as f ₂ / f ₁ = (n + 1) / n.

Considering on the time axis, in order to convert the input signal of the frequency f ₁ into the frequency f ₂ , it can be realized by reproducing the data of n samples of the input signal in a time corresponding to (n + 1) samples. . For that purpose, the data length adjusting means 11 may interpolate the data once every n samples. The data to be interpolated, considering the connection between the data,
Use the same value as the data at the point before interpolation. In the figure (f
_{The case where 2} / f ₁ ) = (n + 1) / n has been described,
The same method can be used regardless of the ratio of (f ₂ / f ₁ ) to 1 or more.

The example in which the frequency of the input signal is substantially increased has been described above. For the method of this inverse to the frequency low, the above-described method, by incorporating in place of the "method of interpolating data", "thinning out data" _{approach, 1> (f 2 / f} 1)> 0 Throat Even in such a case, pitch conversion can be realized.

FIG. 28 shows an example of the result of performing the above-mentioned pitch conversion processing. FIG. 28 shows a spectrum of an output signal (5.312 kHz) obtained by interpolating data every 16 input data when a sine wave of 5 kHz is input to the input signal.

Further, the output data obtained by the pitch conversion by the above method has a signal length changed as compared with the input data before the processing. Depending on the device, this signal length may need to be adjusted. A method of matching the input signal length and the output signal length in this case will be described with reference to FIGS.

FIG. 29 is a block diagram of a pitch converting apparatus having a conventional output signal length adjusting means. That is, FIG.
The output signal length adjusting means 16 is connected to the output of the pitch converting section 18 of the pitch converting device.

FIG. 30 is a diagram showing an adjusting method of the output signal length adjusting means in the conventional pitch converting apparatus. FIG. 30 shows an output signal length adjustment method in which the pitch conversion unit output signal is adjusted and the signal lengths are adjusted when the pitch conversion unit output signal becomes longer than the input signal. This is a signal length adjustment method for adjusting the pitch conversion unit output signal to adjust the signal length when the partial output signal length becomes shorter than the input signal length. 30 and 31, data A and B
Is the pitch converter input data, data A ', B', A ", B"
Is output data of the pitch conversion unit.

In FIG. 30, data A and B are input data of the pitch converting device in a predetermined time frame T, and their data lengths are L ₁ respectively. Outputs obtained by converting the pitch of the input data by the pitch conversion unit 18 become data A ′ and B ′, respectively, and the data lengths thereof are L ₂ (FIG. 22 (b)). In order to match the input signal length and the pitch conversion unit output data signal length, a part of the data A ′ and B ′ is output in an overlapping manner as in the output signal length adjusting means output of FIG. At this time, the data length of the overlapping portion of the data A ′ and B ′ is given by 2L ₂ −2L ₁ . That is, the data B ′ is overlapped from the beginning of the output signal length adjusting means output data in the figure from the point of L ₂ − (2L ₂ −2L ₁ ) = 2L ₁ −L _{2 to} the point of L ₂ .

At this time, if the data A'and B'are overlapped as they are, discontinuity occurs in the data between the overlapped portion and the non-overlapped portion, and the sound is distorted. Therefore, in a region where data is overlapped, data A ′ is faded out and data B ′ is faded in to perform crossfade processing, thereby reducing sound distortion.

Similarly, referring to FIG. 31, a description will be given of an output signal length adjusting method for adjusting the pitch conversion unit output signal length to adjust the signal length when the pitch conversion unit output signal length becomes shorter than the input signal length. I do.

In FIG. 31, data A and B are input data of the pitch converting device in a predetermined time frame T, and their data lengths are L ₁ respectively. Outputs obtained by converting the pitch of the input data by the pitch converting unit 18 are data A ″ and B ″, respectively, and the data lengths thereof are L ₃ shorter than L ₁ (FIG. 31 (b)).

In order to match the input signal length and the pitch conversion unit output signal length, data such as the data of the pitch conversion unit shown in the figure, in which data A ″ and B ″ are repeated twice, is created (FIG. 31).
(C)). Similar to the output signal length adjusting means output shown in FIG. 30, this data is subjected to crossfade processing at the portion where the data overlap, and the output signal length is adjusted to the input signal length while reducing the distortion of the output data.

The above-mentioned output signal length adjusting method is currently performed in a predetermined frame size unit. The data obtained by performing the pitch conversion and the data length adjustment in this manner is shown in FIG.
An output 6 obtained by converting the pitch by inputting it to the / A converter 5 is obtained.

[0021]

However, with the above-described conventional configuration, when the frequency of the input signal is low, pitch conversion with a high S / N ratio can be realized, but as the frequency of the input signal increases. Data error due to interpolation and thinning becomes large, and harmonic noise with respect to the input signal and noise easily occur in the band. As a result, as the frequency becomes higher, the distortion in the data length adjusting means becomes larger and the S / N ratio becomes worse.

When the signal length is adjusted so that the input and output signal lengths are equal to each other, the output signal length adjusting means performs crossfading for each frame, so that periodic noise in accordance with the frame size, Also, signal distortion may occur in the low frequency band, resulting in a poor S / N ratio.

It is an object of the present invention to solve the conventional problems as described above, and to provide a pitch converting device having less distortion and an excellent S / N ratio.

[0024]

According to the present invention of claim 1, a signal input means for inputting a digital signal, a pitch input means for inputting a target pitch at the time of pitch conversion for the input digital signal, and the inputted pitch. Based on the target pitch, an interpolation coefficient control means for outputting an interpolation coefficient, and a data length adjusting means for adjusting the data length of the input digital signal so as to convert the input digital signal into the pitch of the input target pitch, It is a pitch conversion device comprising an interpolation calculation means for interpolating the output of the data length adjusting means with an interpolation coefficient outputted from the interpolation coefficient control means, and a signal output means for outputting the interpolated signal.

According to the second aspect of the present invention, a signal input means for inputting a digital signal, a pitch input means for inputting a target pitch at the time of pitch conversion for the input digital signal, and interpolation based on the input target pitch. An interpolation coefficient control means for outputting a coefficient, an interpolation calculation means for interpolating an input digital signal with an interpolation coefficient output from the interpolation coefficient control means, and an output of the interpolation calculation means for a pitch of a target pitch inputted. It is a pitch converting device provided with a data length adjusting means for adjusting the data length of the output of the interpolation calculating means so as to perform conversion, and a signal outputting means for outputting the output signal of the data length adjusting means.

According to the present invention of claim 7, a pitch input means for inputting a target pitch at the time of pitch conversion for an input digital signal, and a pitch conversion for converting a pitch for the input digital signal based on the input target pitch. Means, a signal frequency component detecting means for detecting the frequency component of the input digital signal, and a signal length of the output of the pitch converting means to be equal to the signal length of the input digital signal according to the detected frequency component. It is a pitch converting apparatus provided with an output signal length adjusting means for performing a crossfade process that reduces distortion and noise.

According to the present invention of claim 8, a pitch input means for inputting a target pitch at the time of pitch conversion for an input digital signal, and a pitch conversion for converting a pitch for the input digital signal based on the input target pitch. Means, an output signal length adjusting means for performing crossfade processing so that the signal length of the output of the pitch converting means becomes the signal length of the input digital signal, and a signal for detecting the frequency component of the output of the output signal length adjusting means. The output signal length adjusting means is a pitch converting device for performing cross-fade processing such that at least distortion and noise are minimized in accordance with the frequency components detected by the signal frequency component detecting means. .

According to a ninth aspect of the present invention, a pitch input means for inputting a target pitch at the time of pitch conversion for an input digital signal, and a pitch conversion for converting a pitch for the input digital signal based on the input target pitch. Means, an output signal length adjusting means for performing crossfade processing so that the signal length of the output of the pitch converting means becomes the signal length of the input digital signal, and the frequency component and the input digital signal of the output of the output signal length adjusting means. And a signal frequency component detecting means for detecting the frequency component of the output signal length adjusting means, wherein the output signal length adjusting means, in accordance with the frequency component detected by the signal frequency component detecting means, performs cross-fade processing such that at least distortion and noise are minimized. It is a pitch converting device for performing.

Therefore, according to the present invention, when adjusting the output signal length by the data length adjusting means, the pitch converting apparatus with less distortion can be realized by interpolating the signal so as to suppress the harmonic noise and the in-band noise.

Further, for example, by providing the oversampling means, even if the frequency of the input signal becomes high, the harmonic components and the noise generated in the band can be suppressed, the distortion is reduced, and the S / N ratio is further reduced. It is possible to realize a pitch conversion device having excellent performance.

Crossfading processing is performed according to the input signal, the output signal, or the frequency characteristics of these signals.
For example, by adjusting the frame size of the output signal length adjusting means and the fade-in / fade-out time in crossfade, it is possible to realize a pitch converting device with less distortion and better S / N.

[0032]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below with reference to the drawings showing the embodiments thereof. (Embodiment 1) FIG. 1 is a block diagram showing a configuration of a pitch converting apparatus according to a first embodiment of the present invention, and FIG. 2 is a block diagram showing a configuration of a pitch converting unit in FIG. . Those having the same functions as those of the conventional example are indicated by the same numbers and symbols.

In FIGS. 1 and 2, 1 is an input signal and 2 is
Is an A / D converter, 3 is a pitch converting means, 4 is a pitch inputting means, 5 is a D / A converter, 6 is an output signal, 11 is a data length adjusting means, 12 is an interpolation coefficient calculating means, and 14 is an interpolation coefficient. It is a control means. Here, the connecting portion between the A / D converter 2 and the pitch converting means 3 is the signal input means, and the connecting portion between the pitch converting means 3 and the D / A converter 5 is the signal output means.

The input signal 1 is converted to a digital signal by the A / D converter 2 and then input to the pitch converting means 3.
When the input signal is digital, the input signal 1 is directly input to the pitch converting means 3.

On the other hand, information on which pitch the input signal is to be converted into, ie, the target pitch, is input to the pitch input means 4,
A control signal is transmitted to the data length adjusting means 11 and the interpolation coefficient control means 14 according to the inputted pitch. Upon receiving the signal from the pitch input means 4, the data length adjusting means 11 adjusts the data length so that the target pitch is reached. This adjustment of the data length is performed by interpolating one data item for every n input data items.
Or process by thinning out. As a result, the frequency of the output signal is multiplied by n / (n + 1) times by supplementing the data with the input signal, and is multiplied by n / (n-1) times by thinning out the data.

The interpolation coefficient calculating means 12 interpolates the output of the data length adjusting means 11 using the interpolation coefficient from the interpolation coefficient control means 14 (for example, a pitch conversion ratio corresponding to the target pitch can be considered). This interpolation method is shown below.

FIG. 3 shows an example of smoothly adjusting n pieces of data to (n + 1) pieces. FIG. 3A shows an input data string 21 of the data length adjusting means 11, which is a pulse string sampled at the sampling frequency f. FIG. 3 (b)
Is the output data train 22 of the data length adjusting means 11, and is a pulse train sampled at the sampling frequency f like the input data train. The input data string 21 is assumed to be n pieces of data T _{1 to} T _n . That is, the data length adjusting means 11 replenishes one data with this number. Therefore, the output data string 22 has one more data than the input data 21, and the supplemented data T _{n + 1} has the same size as T _n . In order to reduce the distortion in the conventional device in FIG. 3B, the n data strings 2 in FIG.
1 envelope and (n + 1) number of data strings 22 in FIG.
The envelopes of should be similar. Therefore, in FIG. 3B, the interpolation coefficient is set so as to have a shape in which FIG. 3A is extended by (n + 1) / n times on the time axis.

In order to set the interpolation coefficient, the (n + 1) output data strings 22 in FIG. 3B are converted into n in FIG.
The sampling cycle is narrowed and overlapped with the interval between the input data strings 21. This is shown in FIG. In FIG. 3C, the output data string 22 is shown by a dotted line in order to distinguish the data. The sampling cycle of the output data string 22 of FIG. 3C is n / (n + 1) times the sampling cycle of the input data string 21 of FIG.

In FIG. 3C, interpolation is performed so that the envelopes of the data strings are equal and the amplitude of the output data string 22 is on the line connecting the input data strings 21. In this embodiment, in order to simplify the data processing, the input data strings 21 are connected by straight lines. The present invention is not limited to this, and curves obtained by various interpolations may be used.

An enlarged view of the portion between the first two points of the input data string 21 in FIG. 3C is shown in FIG. 3D. In FIG. 3D, the input data are indicated by T ₁ and T ₂ , and the output data after interpolation processing are indicated by S ₁ and S ₂ . The position of the output waveform S _{2 is at} a position where the input data T ₁ and T ₂ are divided into (n: 1) based on the difference in sampling period. When this is the k-th position, the output data S ₁ is (n) between T ₁ and T _2.
+ 1-k: k). However, (1 ≤ k
≦ n + 1).

In FIG. 3D, the output waveform S ₂ is adjusted so that the magnitude is at the position of the point 23. Since a straight line is formed between the input data T ₁ and T ₂ , a straight triangle L is formed above the straight line L in FIG. 3D, and the triangle 24 composed of the point 23 and the output data S ₁ is the input data. It is similar to the triangle composed of T ₂ . Using this, output data S
₂ is

[0042]

[Equation 1]

It can be expressed as For the kth data,

[0044]

[Equation 2]

It is expressed as follows. The coefficient part of T _k in this equation becomes the interpolation coefficient. The interpolation coefficient can be expressed by using n and k from (Equation 2). This interpolation coefficient is provided by the interpolation coefficient control means 14. By setting the interpolation coefficient in this way, the envelopes of the input data sequence 21 and the output data sequence 22 become similar, and pitch conversion with less distortion than in the conventional method can be realized.

The output of the pitch converting means 3 is input to the D / A converter 5, and the pitch conversion result is output as an output signal 6.

A pitch conversion using the pitch converting apparatus according to the present embodiment and a comparison with the conventional method will be described below. The conditions for comparison are that the pitch converter input signal is a 5 kHz sine wave, the sampling frequency of the A / D converter 2 is 44.1 kHz, and one data is added to 17 data, that is, the output frequency is 5 * 17/18 = 4.72kHz. In the present embodiment, the comparison is performed under the above conditions,
Even when the sampling frequency and the data interpolation interval are changed, the same effect as in the present embodiment can be obtained. Although noise decreases as the frequency of the input signal decreases and noise increases as the frequency increases, 5 kHz is selected as a representative value in the present embodiment.

In FIGS. 4 and 5, the conventional method is shown by a dotted line, and the result of the pitch converting apparatus of the present embodiment is shown by a solid line. As a result of the simulation, in FIG. 4, the spectrum of the pitch-converted signal component is equal, but in the conventional method, many harmonic components and in-band noise are generated in addition to the pitch-converted output, and the signal dynamic range is about 17 [dB]. Is.
On the other hand, the pitch converting apparatus according to the present embodiment can reduce the noise level by about 15 [dB] as compared with the conventional apparatus. As a result, the signal dynamic range is approximately 35 [d
B] could be improved.

Further, FIG. 5 shows characteristics when the device of this embodiment is constituted by an actual circuit. Also in FIG. 5, almost the same result as the simulation result (FIG. 4) is shown, and it can be seen that the harmonic components and the in-band noise are reduced. In the signal dynamic range, the conventional device is 15 [d
B], on the other hand, in the pitch conversion device of the present embodiment,
It has been improved to about 35 [dB].

From the above results, by using the pitch converting apparatus according to the present embodiment, it is possible to suppress harmonic components and in-band noise more than the conventional apparatus, and it is possible to realize pitch conversion with less distortion.

In this embodiment, 1 is set for every n input data.
Although the method of replenishing one data has been described with an example, a method of thinning out one data for every n input data can be performed by using the same method.
In that case, (Equation 2) is

[0052]

(Equation 3)

By setting the interpolation coefficient as shown in (Equation 3), harmonic noise,
In-band noise can be reduced (however, if the method of the second embodiment described later is used for the method of thinning out data, a pitch conversion device having a better S / N ratio can be configured. it can).

By using the above method, pitch conversion with less distortion can be realized. By interpolating one data for every n input data, the pitch can be lowered, and by decreasing n, the pitch drop can be increased. It is also possible to raise the pitch by interpolating one data for n input data and then thinning out the data. Similarly, the pitch can be increased by thinning out one data item from the n input data items, and the pitch increase can be increased by decreasing n. It is also possible to reduce the pitch by thinning out one data for n input data and then interpolating the data.

In the present embodiment, the pitch conversion is realized by interpolating or thinning out one piece of data for n pieces of input data, but the pitch is changed by interpolating or thinning m pieces of data for n pieces of input data. You can also do things. But m
By increasing S, the distortion of the data increases and S
The / N ratio becomes worse. (Embodiment 2) Hereinafter, a second embodiment of the present invention will be described with reference to the drawings.

The block diagram showing the configuration of the pitch converting apparatus in the second embodiment of the present invention is the same as the configuration of FIG. 1, and FIG. 6 is a block diagram showing the configuration of the pitch converting unit in FIG. is there. Those having the same functions as those of the conventional example are indicated by the same numbers and symbols.

1 and 6, 1 is an input signal, 2 is
Is an A / D converter, 3 is a pitch converting means, 4 is a pitch inputting means, 5 is a D / A converter, 6 is an output signal, 11 is a data length adjusting means, 12 is an interpolation coefficient calculating means, and 14 is an interpolation coefficient. It is a control means.

The input signal 1 is converted into a digital signal by the A / D converter 2 and then input to the pitch converting means 3.
When the input signal is digital, the input signal 1 is directly input to the pitch converting means 3.

On the other hand, the target pitch is input to the pitch input means 4, and a control signal is transmitted to the data length adjusting means 11 and the interpolation coefficient control means 14 according to the input pitch. Upon receiving the signal from the pitch input means 4, the interpolation coefficient calculation means 12 uses the interpolation coefficient from the interpolation coefficient control means 14 to generate an A / D signal.
The output of the converter 2 is interpolated. This interpolation coefficient calculation means 12
Is output to the data length adjusting means 11 to adjust the data length. The interpolation method and the data length adjustment method are shown below.

FIG. 7 shows an example in which n pieces of data are smoothly adjusted to (n-1) pieces. FIG. 7A shows an input data train 21 of the data length adjusting means 11, which is a train of n pulses of T _{1 to} T _n . When converting n pieces of data into (n-1) pieces of data by using the same method as in the first embodiment, the output data row 30 is the data row S of the dotted line in FIG. 7B. Adjust so that _{1 to} S _n-1 . As a result, the frequency of the output data can be n / (n-1) times that of the input.

In FIG. 7B, the output data string 30
Is interpolated so as to be on the line connecting the input data strings 21. In this embodiment, in order to simplify the data processing, the input data strings 21 are connected by straight lines. still,
The present invention is not limited to this, and curves obtained by various interpolations may be used.

By considering the same as in the first embodiment, the output data string 30 is expressed by (Equation 3). In this equation, k means the kth data of the output data. The coefficient part of T _k in this equation becomes the interpolation coefficient. The interpolation coefficient can be expressed by using n and k from (Equation 3). This interpolation coefficient is provided by the interpolation coefficient control means 14. By setting the interpolation coefficient in this way, the input data string 2
1 and the envelope of the output data string 30 are similar, and pitch conversion with less distortion than in the conventional method can be realized.

The output of the pitch converting means 3 is input to the D / A converter 5, and the pitch conversion result is output as an output signal 6.

A pitch conversion using the pitch converting apparatus according to the present embodiment and a comparison of the conventional method will be described below. The conditions for comparison are that the input signal of the pitch converter is a 5 kHz sine wave, the sampling frequency of the A / D converter 2 is 44.1 kHz, and one data is reduced for 17 data, that is, the output frequency is It is 5 * 17/16 = 5.31kHz. In the present embodiment, the comparison is performed under the above conditions. However, even if the sampling frequency and the data interpolation interval are changed, the same effect as that of the present embodiment can be obtained. When the frequency of the input signal becomes low, the noise decreases, and when it becomes high, the noise increases. However, in the present embodiment, the typical value is 5 kHz.
Will be selected.

In FIG. 8 and FIG. 9, the conventional method is shown by a dotted line, and the result of the pitch converting apparatus of this embodiment is shown by a solid line. As a result of the simulation, in FIG. 8, the spectrum of the pitch-converted signal component is equal, but in the conventional method, many harmonic components and in-band noise are generated in addition to the pitch-converted output, and the signal dynamic range is about 17 [dB]. ]. On the other hand, the pitch converting apparatus according to the present embodiment can reduce the noise level by about 15 [dB] as compared with the conventional apparatus. As a result, the signal dynamic range could be improved to about 35 [dB].

Further, FIG. 9 shows characteristics when the device of this embodiment is constituted by an actual circuit. 9 shows almost the same result as the simulation result (FIG. 8), and it can be seen that the harmonic components and the in-band noise are reduced. In the signal dynamic range, the conventional device is 15 [d
B], on the other hand, in the pitch conversion device of the present embodiment,
It has been improved to about 35 [dB].

From the above results, by using the pitch converting apparatus according to the present embodiment, harmonic components and in-band noise can be suppressed more than in the conventional apparatus, and pitch conversion with less distortion can be realized.

In the present embodiment, the method of thinning out one data for every n input data has been described by way of example. However, by using a similar method, every n input data can be obtained. A method of supplementing one data can also be performed. In that case, (Equation 3) becomes (Equation 2). In the data length adjusting means 11, the frequency is changed to n by changing the data length from n to (n + 1).
It can be multiplied by / (n + 1). By setting the interpolation coefficient as described above, it is possible to reduce harmonic noise and in-band noise as compared with the conventional method (however, regarding the method of interpolating the data, the method of the first embodiment described above is used. It is possible to construct a pitch conversion device having a better S / N ratio by using it).

By using the above method, pitch conversion with less distortion can be realized. By interpolating one data for every n input data, the pitch can be lowered, and by decreasing n, the pitch drop can be increased. Also, 1 for every n input data
After interpolating each data, by thinning out the data,
You can also raise the pitch. Similarly, the pitch can be increased by thinning out one data item from the n input data items, and the pitch increase can be increased by decreasing n. In addition, by thinning out one data from the n input data and interpolating the data,
You can also lower the pitch. (Embodiment 3) Hereinafter,
A third embodiment of the present invention will be described with reference to the drawings.

FIG. 10 is a block diagram showing the structure of the pitch converting apparatus according to the third embodiment of the present invention.
FIG. 1 is a block diagram showing the configuration of the pitch converting section in FIG. Those having the same functions as those of the conventional example are indicated by the same numbers and symbols.

10 and 11, 1 is an input signal, 2 is an A / D converter, 3 is a pitch converting means, 4 is a pitch inputting means, 5 is a D / A converter, 6 is an output signal, and 7 is Oversampling means, 8 is first filter means, 9 is second
Filter means, 10 is down-sampling means, 11
Is a data length adjusting means, 12 is an interpolation coefficient calculating means, and 14 is an interpolation coefficient controlling means. Here, the oversampling means 7 and the first filter means 8 are included in the signal input means, and the downsampling means 10 and the second filter means 9 are included in the signal output means.

In this embodiment, a method of replenishing n pieces of input data with one piece of data will be described with reference to the drawings and the method described in the first embodiment.

The input signal 1 is converted into a digital signal by the A / D converter 2 and then input to the oversampling means 7. Input signal 1 if the input signal is digital
Is directly input to the oversampling means 7.

The oversampling means 7 oversamples the input signal at a sampling frequency f _s2 that is at least twice the sampling frequency f _s1 of the A / D converter 2. In the present embodiment, the same interpolation data as the previous data is used as the interpolation data between the sampling frequencies f _s1 , but the same effect can be obtained by inserting 0 or performing linear interpolation. By performing oversampling, it is apparently equivalent to the input signal frequency being multiplied by (f _s1 / f _s2 ). Because of f _s1 <f _s2 , the signal to be processed has characteristics similar to those of the frequency reduced by a factor of (f _s1 / f _s2 ).
In the pitch converting apparatus of the present invention, noise decreases as the frequency of the input signal decreases as described above. Therefore, the S / N ratio is improved by oversampling, and the S / N ratio increases as the oversampling frequency increases. Is improved.

The output of the oversampling means 7 is input to the first filter means 8. The frequency characteristic of the first filter means 8 is a low-pass characteristic that allows a frequency region equal to or lower than half the frequency of f _s1 to pass. As a result, at the time of pitch conversion, the occurrence of aliasing is prevented and the data interpolated by the oversampling means 7 are smoothly connected.

On the other hand, the target pitch is input to the pitch inputting means 4, and the data length adjusting means 11, depending on the input pitch.
And the control signal is transmitted to the interpolation coefficient control means 14 in the pitch converting section. In response to the output of the first filter means 8, the data length adjusting means 11 adjusts the data length based on the control signal from the pitch input means 4. The interpolation coefficient calculation means 12 uses the interpolation coefficient from the interpolation coefficient control means 14 to interpolate the output of the data length adjustment means 11. The data length adjusting method and the interpolation method are the same as the method shown in the first embodiment.

The signal processed by the data length adjusting means 11 is input to the second filter means 9. The frequency characteristic of the second filter means 9, like the filter means 8, is a low-pass characteristic that allows passage of a frequency region equal to or lower than half the sampling frequency f _s1 of the A / D converter 2. This prevents aliasing that occurs when the sampling frequency of the signal after the pitch conversion by the pitch conversion means 3 is returned to the sampling frequency of the A / D converter 2.

The output of the second filter means 9 is input to the down sampling means 10. Here, the data oversampled to the sampling frequency f _s2 is converted into A / D
Convert to the sampling frequency of the converter 2, f _s1 .
In this embodiment, as means for downsampling,
A method of outputting data once every (f _s2 / f _s1 ) and ignoring the rest of the data is used, but the present invention is not limited to this, and various other downsampling methods may be used. Good.

The signal down-sampled by the down-sampling means 10 is input to the D / A converter 5, and D / A converter 5
An output signal 6 converted into an analog signal is obtained from the A converter 5.

As described above, by oversampling the input signal, the pitch conversion means 3 apparently
The frequency of the input signal can be lowered and the S / N ratio can be improved. The simulation results of this pitch converter and the measurement results by the actual circuit configuration are shown below.

The conditions for simulation and actual measurement are as follows:
Input signal of pitch converter is 5kHz sine wave, A / D converter 2
The sampling frequency is 44.1kHz, the oversampling frequency is 88.2kHz, and one data is added to 17 data, that is, the output frequency is 5 * 17/18 = 4.72kHz.
In the present embodiment, the above conditions are used, but even when the sampling frequency and the data interpolation interval are changed,
The same effect as the present embodiment can be obtained. Although the noise decreases as the frequency of the input signal decreases and the noise increases as the frequency increases, 5 kHz is selected as the representative value in the present embodiment. Further, by increasing the oversampling frequency, there is the same effect as lowering the input signal,
The S / N ratio can be further improved.

The result of the simulation is shown in FIG.
In the simulation and actual measurement results, the result using the method of the first embodiment is shown by a dotted line, and the result of the pitch conversion device of the present embodiment is shown by a solid line. In FIG.
In the method of the first embodiment, in addition to the pitch conversion output,
Many harmonic components and in-band noise are generated, and the signal dynamic range is about 35 [dB]. On the other hand, in the pitch conversion device according to the present embodiment, compared with the conventional device,
The noise level can be lowered by about 15 [dB], and the in-band noise is below the display range. As a result, the signal dynamic range could be improved to about 50 [dB].

FIG. 13 shows the characteristics when the device of this embodiment is constructed by an actual circuit. Also in FIG. 13, almost the same result as the simulation result (FIG. 12) is shown, and it can be seen that the harmonic components and the in-band noise are reduced. The signal dynamic range is 33 [dB] in the first embodiment, but is improved to about 48 [dB] in the pitch conversion device of the present embodiment.

As described above, the method of replenishing one piece of data with n pieces of input data has been described. However, the method of thinning out one piece of data with n pieces of input data is the same as in the first embodiment. This can be realized by using the configuration method, and the S / N ratio can be improved as compared with the pitch conversion device of the configuration of the first embodiment (however, regarding the method of thinning out the data, the fourth embodiment described later will be described). It is possible to construct a pitch conversion device having a better S / N ratio by using the method of form).

From the above results, by using the pitch converting apparatus using oversampling according to the present embodiment, the harmonic component and the in-band noise can be reduced by the conventional apparatus and the configuration shown in the first embodiment. It is possible to suppress the pitch and realize pitch conversion with less distortion. In addition, by increasing the oversampling frequency, S /
The N ratio can be improved. (Embodiment 4) Hereinafter, a fourth embodiment of the present invention will be described with reference to the drawings.

A block diagram showing the configuration of the pitch converting apparatus in the fourth embodiment of the present invention is the same as FIG. 10, and FIG. 14 is a block diagram showing the configuration of the pitch converting unit in FIG. Those having the same functions as those of the conventional example are indicated by the same numbers and symbols.

In FIGS. 10 and 14, 1 is an input signal, 2 is an A / D converter, 3 is a pitch converting means, 4 is a pitch inputting means, 5 is a D / A converter, 6 is an output signal, and 7 is Oversampling means, 8 is first filter means, 9 is second
Filter means, 10 is down-sampling means, 11
Is a data length adjusting means, 12 is an interpolation coefficient calculating means, and 14 is an interpolation coefficient controlling means.

In the present embodiment, n input signals are represented by (n-
1) A method for smoothly adjusting the number of output signals will be described with reference to the drawings and the second embodiment.

The input signal 1 is converted into a digital signal by the A / D converter 2 and then input to the oversampling means 7. Input signal 1 if the input signal is digital
Is directly input to the oversampling means 7.

The oversampling means 7 oversamples the input signal at a sampling frequency f _s2 that is at least twice the sampling frequency f _s1 of the A / D converter 2. In the present embodiment, the same interpolation data as the previous data is used as the interpolation data between the sampling frequencies f _s1 ,
The same effect can be obtained by inserting 0 or performing linear interpolation.

By performing oversampling, it is apparently equal to the input signal frequency multiplied by (f _s1 / f _s2 ). Because of f _s1 <f _s2 , the signal to be processed has characteristics similar to those of the frequency reduced by a factor of (f _s1 / f _s2 ). In the pitch converting apparatus of the present embodiment, noise decreases as the frequency of the input signal becomes lower as described above, so the S / N ratio is improved by oversampling.
The higher the oversampling frequency, the more S / N
The ratio is improved.

The output of the oversampling means 7 is input to the first filter means 8. The frequency characteristic of the first filter means 8 is a low-pass characteristic that allows a frequency region equal to or lower than half the frequency of f _s1 to pass. As a result, at the time of pitch conversion, the occurrence of aliasing is prevented and the data interpolated by the oversampling means 7 are smoothly connected.

On the other hand, the target pitch is input to the pitch input means 4, and a control signal is transmitted to the data length adjusting means 11 and the interpolation coefficient control means 14 in the pitch converting section according to the input pitch. The interpolation coefficient calculation means 12 interpolates the output of the A / D converter 2 using the interpolation coefficient from the interpolation coefficient control means 14 which receives the signal from the pitch input means 4. The output of the interpolation coefficient calculating means 12 is used as the data length adjusting means 1
Input 1 and adjust the data length. The interpolation method and the data length adjustment method are the same as the method described in the second embodiment.

The signal thus processed by the interpolation coefficient calculating means 12 and adjusted in data length is input to the second filter means 9. The frequency characteristic of the second filter means 9 is a low-pass characteristic that allows a frequency region equal to or lower than half the sampling frequency f _s1 of the A / D converter 2 to pass, as in the case of the first filter means 8. As a result, when the sampling frequency of the signal after the pitch conversion by the pitch converting means 3 is returned to the sampling frequency of the A / D converter 2, generation of alias is prevented.

The output of the second filter means 9 is input to the downsampling means 10. Here, the data oversampled to the sampling frequency f _s2 is converted into A / D
Convert to the sampling frequency of the converter 2, f _s1 .
In this embodiment, as means for downsampling,
A method of outputting data once every (f _s2 / f _s1 ) and ignoring the rest of the data is used, but the present embodiment is not limited to this, and various other downsampling methods are used. May be.

The signal down-sampled by the down-sampling means 10 is input to the D / A converter 5, and D / A converter 5
An output signal 6 converted into an analog signal by the A converter 5 is obtained.

As described above, by oversampling the input signal, the pitch conversion means 3 apparently
The frequency of the input signal can be lowered and the S / N ratio can be improved. The simulation results of this pitch converter and the measurement results by the actual circuit configuration are shown below.

The conditions for simulation and actual measurement are as follows:
Input signal of pitch converter is 5kHz sine wave, A / D converter 2
The sampling frequency is 44.1kHz, the oversampling frequency is 88.2kHz, and 1 data is reduced from 17 data, that is, the output frequency is 5 * 17/16 = 5.31 kHz.
In the present embodiment, the above conditions are used, but even when the sampling frequency and the data interpolation interval are changed,
The same effect as the present embodiment can be obtained. Although the noise decreases as the frequency of the input signal decreases and the noise increases as the frequency increases, 5 kHz is selected as the representative value in the present embodiment. Further, increasing the oversampling frequency has the same effect as lowering the input signal, and the S / N ratio can be further improved.

The result of the simulation is shown in FIG.
In the simulation and actual measurement results, the results using the method of the second embodiment are shown by dotted lines, and the results of the pitch conversion device of this embodiment are shown by solid lines. In FIG. 15, according to the method of the second embodiment, in addition to the pitch conversion output,
Many harmonic components and in-band noise are generated, and the signal dynamic range is about 35 [dB]. On the other hand, in the pitch conversion device according to the present embodiment, compared with the conventional device,
The noise level can be lowered by about 15 [dB], and the in-band noise is below the display range. As a result, the signal dynamic range could be improved to about 50 [dB].

FIG. 16 shows the characteristics when the device of this embodiment is constructed by an actual circuit. Also in FIG. 16, almost the same result as the simulation result (FIG. 15) is shown, and it can be seen that the harmonic components and the in-band noise are reduced. The signal dynamic range is 33 [dB] in the second embodiment, but is improved to about 48 [dB] in the pitch conversion device of the present embodiment.

Although the method of thinning out one data item from every n input data items has been described above, the method of replenishing one data item to every n input data items is the same as that of the second embodiment. Can be realized by using the method, and the S / N ratio can be improved as compared with the pitch converting apparatus having the configuration of the second embodiment (however, regarding the method of supplementing the data, the third embodiment described above is used). It is better to use the morphological method
It is possible to configure a pitch converting device having a better S / N ratio).

From the above results, by using the pitch converting apparatus using oversampling according to the present embodiment, it is possible to obtain higher harmonic components and in-band than the conventional apparatus and the configurations shown in the first and second embodiments. Noise can be suppressed, and pitch conversion with little distortion can be realized. Also, by increasing the oversampling frequency,
Further, the S / N ratio can be improved. (Fifth Embodiment) Hereinafter, a fifth embodiment of the present invention will be described with reference to the drawings.

FIG. 18 is a block diagram showing the structure of the pitch converting apparatus according to the fifth embodiment of the present invention.
9 is a block diagram showing the configuration of the output signal length adjusting means in FIG. Those having the same functions as those of the conventional example are indicated by the same numbers and symbols.

18 and 19, 1 is an input signal, 2 is an A / D converter, 15 is a signal frequency component detecting means, 16 is an output signal length adjusting means, 19 is an output signal length adjusting means coefficient memory, 5 is a D / A converter, 6 is an output signal,
Reference numeral 18 denotes a pitch converting section, and the pitch converting section 18 is the first
Alternatively, it is similar to the pitch converting section in the second embodiment, and has a structure including a pitch input means 4, a data length adjusting means 11, an interpolation coefficient calculating means 12, and an interpolation coefficient controlling means 14. Alternatively, it is a configuration including the pitch converting section, the oversampling means 7, the first filter means 8, the second filter means 9, and the downsampling means 10 in the third or fourth embodiment.

In the present embodiment, there is a method of adjusting the signal (tone conversion unit output) obtained by smoothly adjusting n input signals to (n + 1) output signals by the tone conversion unit 18 to the same signal length as the input signal. Will be described with reference to the drawings.

In FIG. 18, the input signal 1 is converted into a digital signal by the A / D converter 2 and then the pitch conversion section 1
8 and the signal frequency component detecting means 15. When the input signal is digital, the input signal 1 is directly input to the pitch converting section 18 and the signal frequency component detecting means 15.

The signal frequency component detecting means 15 detects the frequency component of the inputted input signal and outputs the control signal to the output signal length adjusting means 16. As a method of detecting the signal frequency component, any signal frequency detecting method such as fast Fourier transform may be used.

The pitch converting section 18 receives the input signal and converts the pitch using any of the methods of the first to fourth embodiments. In the present embodiment, the pitch converting section outputs the pitch conversion means to (n
The output data at point +1) is output to the output signal length adjusting means 16.

In the output signal length adjusting means 16, in order to adjust the signal length, the pitch conversion section output is divided at every predetermined time T. The output data unit during this time T is one frame. In the case of the present embodiment, the number of data pieces of the output signal of the pitch converting section 18 in one frame is (n + 1) / n for one frame of the input signal. In order to make this output signal length equal to the input signal length, the signal length is adjusted by using a method similar to the conventional method shown in FIG.

In the conventional output signal length adjusting means 16 (see FIG. 29), a certain predetermined frame length and a crossfade of a predetermined fade-in / fade-out time are used to adjust the signal length. In the method, when the input signal has a pitch cycle like a voice or a sound of a wind instrument, the frequency component of the input signal may cause distortion such that a periodic sound of the pitch cycle sounds like swaying. Further, since the frame length is fixed, periodic noise is generated due to the discontinuous signal generated at each frame period.

In order to prevent these signal distortions and noises, in the present invention, the signal frequency component detecting means 15 detects the largest signal frequency component in the input signal and the low frequency distortion component of 100 Hz or less, and outputs the output signal length. The control signal is output to the adjusting means 16. Based on the control signal, the output signal length adjusting means 16 changes the frame length and the fade-in / fade-out time in accordance with the frequency component and distortion component of the input signal. For the frame length and the fade-in / fade-out time to be changed, values obtained by performing preliminary experiments in advance and adjusting so that distortion and noise are minimized are used. This value is stored in advance in the coefficient memory 19, and is transferred to the output signal length adjusting means 16 based on the control signal of the signal frequency component detecting means 15. The output signal length adjusting means 16 adjusts the fade-in / fade-out time of the crossfade according to this value to adjust the signal length. FIG. 17 shows an example in which the fade-in / fade-out time of crossfade is adjusted.

The simulation results of this embodiment are shown in FIG. The simulation conditions are as follows: the input signal of the pitch converter is a 5kHz sine wave, the sampling frequency of the A / D converter 2 is 44.1kHz, and one data is added to 17 data, that is, the output frequency is 5 * 17/18 = It is 4.72kHz. In this embodiment, 5 kHz is selected as the representative value. A solid line in FIG. 20 is an example of the frequency characteristic of the pitch converting device according to the present invention, and a broken line is an example of the frequency characteristic of the conventional device in the case of the same input signal. It can be seen from FIG. 20 that the pitch conversion device according to the present embodiment has less in-band noise than the conventional device. This can reduce signal distortion and in-band noise.

As described above, the pitch converter output signal length (n + 1)
Was described as the method of making the input signal length n equal to the input signal length n, but the method of making the pitch conversion unit output signal length (n-1) the input signal length n is similar to the conventional output signal length adjustment method of FIG. It can be realized by using this configuration method.

From the above results, by using the pitch converting device using the output signal length adjusting means 16 according to the present embodiment, it is possible to suppress the harmonic component and the in-band noise, and the distortion is less than that of the conventional device. Pitch conversion can be realized. (Sixth Embodiment) Hereinafter, a sixth embodiment of the present invention will be described with reference to the drawings.

FIG. 21 is a block diagram showing the structure of the pitch converting apparatus according to the sixth embodiment of the present invention.
The block diagram showing the configuration of the output signal length adjusting means 16 in No. 1 is the same as that in FIG. Those having the same functions as those of the conventional example are indicated by the same numbers and symbols.

21 and 19, 1 is an input signal, 2 is an A / D converter, 15 is a signal frequency component detecting means, 16 is an output signal length adjusting means, and 19 is a coefficient memory for the output signal length adjusting means. 5, 5 is a D / A converter, 6 is an output signal, 18 is a pitch converting section, and the pitch converting section 18 is similar to the pitch converting section in the first or second embodiment, and the pitch inputting means 4, a data length adjusting means 11, an interpolation coefficient calculating means 12, and an interpolation coefficient controlling means 14. Alternatively, it is a configuration including the pitch converting section, the oversampling means 7, the first filter means 8, the second filter means 9, and the downsampling means 10 in the third or fourth embodiment.

In the present embodiment, a method of adjusting the signal (tone conversion unit output) obtained by smoothly adjusting n input signals to (n + 1) output signals by the tone conversion unit 18 to the same signal length as the input signal is described. Will be described with reference to the drawings.

In FIG. 21, the input signal 1 is converted into a digital signal by the A / D converter 2, and then the pitch conversion section 1
8 is input. When the input signal is digital, the input signal 1 is directly input to the pitch converting section 18.

The pitch converting section 18 receives the input signal and converts the pitch by using any one of the methods of the first to fourth embodiments. In the present embodiment, the pitch converting section outputs the pitch conversion means to (n
The output data of the +1) point is output to the output signal length adjusting means 16. Further, the output of the output signal length adjusting means 16 is output to the signal frequency component detecting means 15 and the D / A converter 5.

The signal frequency component detecting means 15 detects the frequency component of the input output signal of the output signal length adjusting means 16 and outputs a control signal to the output signal length adjusting means 16. The detection method of the signal frequency component is fast Fourier transform, etc.
Any signal frequency detection method may be used.

In the output signal length adjusting means 16, in order to adjust the signal length, the pitch conversion section output is divided at every predetermined time T. The output data unit during this time T is one frame. In the case of the present embodiment, the number of data pieces of the output signal of the pitch converting section 18 in one frame is (n + 1) / n for one frame of the input signal. In order to make this output signal length equal to the input signal length, the signal length is adjusted by using a method similar to the conventional method shown in FIG.

In the conventional output signal length adjusting means 16 (see FIG. 29), a certain predetermined frame length and a crossfade of a predetermined fade-in / fade-out time are used to adjust the signal length. In the method, when the input signal has a pitch cycle like a voice or a sound of a wind instrument, the frequency component of the input signal may cause distortion such that a periodic sound of the pitch cycle sounds like swaying. Further, since the frame length is fixed, periodic noise is generated due to the discontinuous signal generated at each frame period.

In order to prevent these signal distortion and noise, in the present invention, in the signal frequency component detecting means 15, the largest signal frequency component in the output signal of the output signal length adjusting means 16 and the low frequency distortion component of 100 Hz or less are obtained. Is detected and a control signal is output to the output signal length adjusting means 16. Based on the control signal, the output signal length adjusting means 16 changes the frame length and the fade-in / fade-out time corresponding to the frequency component and the distortion component of the output signal. For the frame length and the fade-in / fade-out time to be changed, values obtained by performing preliminary experiments in advance and adjusting so that distortion and noise are minimized are used. This value is stored in advance in the coefficient memory 19, and is transferred to the output signal length adjusting means 16 based on the control signal of the signal frequency component detecting means 15. The output signal length adjusting means 16 adjusts the fade-in / fade-out time of the crossfade according to this value to adjust the signal length. FIG. 17 shows an example in which the fade-in / fade-out time of crossfade is adjusted.

FIG. 22 shows the simulation result of this embodiment. The simulation condition is that the input signal of the pitch converter is a 5kHz sine wave, the sampling frequency of the A / D converter 2 is 44.1kHz, and the data is increased by 1 to 17 data, that is, the output frequency is 5 * 17/18 = It is 4.72 kHz.
In this embodiment, 5 kHz is selected as the representative value.
The solid line in FIG. 22 is an example of the frequency characteristic of the pitch converting device according to the present invention, and the broken line is an example of the frequency characteristic of the conventional device in the case of the same input signal. It can be seen from FIG. 22 that the pitch conversion device according to the present embodiment has less in-band noise than the conventional device. This can reduce signal distortion and in-band noise.

As described above, the pitch converter output signal length (n + 1)
Was described as the method of making the input signal length n equal to the input signal length n, but the method of making the pitch conversion unit output signal length (n-1) the input signal length n is similar to the conventional output signal length adjustment method of FIG. It can be realized by using this configuration method.

From the above results, by using the pitch converting device using the output signal length adjusting means 16 according to the present embodiment, it is possible to suppress the harmonic component and the in-band noise and reduce the distortion as compared with the conventional device. Pitch conversion can be realized. (Embodiment 7) Hereinafter, a seventh embodiment of the present invention will be described with reference to the drawings.

FIG. 23 is a block diagram showing the structure of the pitch converting apparatus according to the seventh embodiment of the present invention.
The block diagram showing the configuration of the output signal length adjusting means 16 in 3 is the same as that in FIG. Those having the same functions as those of the conventional example are indicated by the same numbers and symbols.

23 and 19, 1 is an input signal, 2 is an A / D converter, 15 is a signal frequency component detecting means, 16 is an output signal length adjusting means, and 19 is a coefficient memory for the output signal length adjusting means. 5, 5 is a D / A converter, 6 is an output signal, 18 is a pitch conversion section, and the pitch conversion section 18 is similar to the pitch conversion section in the first or second embodiment, and the pitch input means 4, a data length adjusting means 11, an interpolation coefficient calculating means 12, and an interpolation coefficient controlling means 14. Alternatively, the configuration includes the pitch converting section, the oversampling means 7, the first filter means 8, the second filter means 9, and the downsampling means 10 in the third or fourth embodiment.

In the present embodiment, a method for adjusting the signal (tone conversion unit output) obtained by smoothly adjusting n input signals to (n + 1) output signals by the tone conversion unit 18 to the same signal length as the input signal is described. Will be described with reference to the drawings.

In FIG. 23, the input signal 1 is converted into a digital signal by the A / D converter 2, and then the pitch conversion section 1
8 and the signal frequency component detecting means 15. When the input signal is digital, the input signal 1 is directly input to the pitch converting section 18 and the signal frequency component detecting means 15.

The pitch converting section 18 receives the input signal and converts the pitch using any of the methods of the first to fourth embodiments. In the present embodiment, the pitch converting section outputs the pitch conversion means to (n
The output data of the +1) point is output to the output signal length adjusting means 16. The output of the output signal length adjusting means 16 is output to the D / A converter 5 and the signal frequency component detecting means 15.

The signal frequency component detecting means 15 detects the signal frequency component of the difference between the input signal inputted and the output of the output signal length adjusting means 16, and outputs the control signal to the output signal length adjusting means 16. As a method of detecting the signal frequency component, any signal frequency detecting method such as fast Fourier transform may be used.

In the output signal length adjusting means 16, in order to adjust the signal length, the pitch conversion section output is divided at every predetermined time T. The output data unit during this time T is one frame. In the case of the present embodiment, the number of data pieces of the output signal of the pitch converting section 18 in one frame is (n + 1) / n for one frame of the input signal. In order to make this output signal length equal to the input signal length, the signal length is adjusted by using a method similar to the conventional method shown in FIG.

In the conventional output signal length adjusting means 16 (see FIG. 29), a certain predetermined frame length and a crossfade of a predetermined fade-in / fade-out time are used to adjust the signal length. In the method, when the input signal has a pitch cycle like a voice or a sound of a wind instrument, the frequency component of the input signal may cause distortion such that a periodic sound of the pitch cycle sounds like swaying. Further, since the frame length is fixed, periodic noise is generated due to the discontinuous signal generated at each frame period.

In order to prevent these signal distortions and noises, in the present invention, the signal frequency component detecting means 15 has the largest signal frequency component of the difference component between the input signal and the output signal of the output signal length adjusting means 16 and 100 Hz or less. A low-frequency distortion component is detected and a control signal is output to the output signal length adjusting means 16. Based on the control signal, the output signal length adjusting means 16 changes the frame length and the fade-in / fade-out time in accordance with the frequency component and distortion component of the input signal. For the frame length and the fade-in / fade-out time to be changed, values obtained by performing preliminary experiments in advance and adjusting so that distortion and noise are minimized are used. This value is stored in advance in the coefficient memory 19, and is transferred to the output signal length adjusting means 16 based on the control signal of the signal frequency component detecting means 15. The output signal length adjusting means 16 adjusts the fade-in / fade-out time of the crossfade according to this value to adjust the signal length. FIG. 17 shows an example in which the fade-in / fade-out time of crossfade is adjusted.

The simulation result of this embodiment is shown in FIG. The simulation condition is that the input signal of the pitch converter is a 5kHz sine wave, the sampling frequency of the A / D converter 2 is 44.1kHz, and the data is increased by 1 to 17 data, that is, the output frequency is 5 * 17/18 = It is 4.72 kHz.
In this embodiment, 5 kHz is selected as the representative value.
The solid line in FIG. 24 is an example of the frequency characteristic of the pitch converting device according to the present invention, and the broken line is an example of the frequency characteristic of the conventional device in the case of the same input signal. It can be seen from FIG. 24 that the pitch conversion device according to the present embodiment has less in-band noise than the conventional device. This can reduce signal distortion and in-band noise.

As described above, the pitch converter output signal length (n + 1)
Was described as the method of making the input signal length n equal to the input signal length n, but the method of making the pitch conversion unit output signal length (n-1) the input signal length n is similar to the conventional output signal length adjustment method of FIG. It can be realized by using this configuration method.

From the above results, by using the pitch converting device using the output signal length adjusting means 16 according to the present embodiment, it is possible to suppress the harmonic component and the in-band noise and to reduce the distortion as compared with the conventional device. Pitch conversion can be realized.

As described above, the pitch converting apparatus of the present invention is
When changing the data length of a signal, it is possible to realize a pitch conversion device with less distortion by using interpolation that smoothly connects the signals. Also, using oversampling when converting the pitch of the above method, apparently,
By lowering the frequency of the input signal, it becomes possible to realize a pitch converting device having a better S / N ratio. Further, when adjusting the output signal length during the pitch conversion of the above method, by adjusting the signal length adjustment frame length and the cross-fade fade-in / fade-out time by the input and pitch conversion unit output signals, S / It is possible to realize a pitch converting device having an excellent N ratio.

[0140]

As is apparent from the above description, the present invention is based on the target pitch input to the pitch input means.
The interpolation coefficient control means for outputting the interpolation coefficient, the data length adjusting means for adjusting the data length of the input digital signal so as to convert the pitch to the target pitch, and the output of the data length adjusting means When it is provided with an interpolating calculation means for interpolating with the interpolating coefficient output from, there is an advantage that distortion can be reduced.

Further, there is an advantage that an output signal having a further excellent S / N ratio can be obtained by apparently lowering the frequency of the input signal by using oversampling at the time of pitch conversion.

Further, the signal frequency component detecting means for detecting the frequency component of the input digital signal and the signal length of the output of the pitch converting means are set to the signal length of the input digital signal in accordance with the detected frequency component. When at least the output signal length adjusting means for performing the crossfade processing that minimizes distortion and noise is provided, there is an advantage that periodic noise and signal distortion can be reduced and the S / N ratio can be improved. .

[Brief description of drawings]

FIG. 1 is a block diagram of a pitch converting device according to a first embodiment of this invention.

FIG. 2 is a block diagram showing a configuration of a pitch converting unit according to the first embodiment of the present invention.

FIG. 3 is a diagram for explaining a pitch conversion method according to the first embodiment of the present invention.

FIG. 4 is a diagram showing a frequency characteristic example of a pitch conversion output in a simulation of the pitch conversion device according to the first embodiment of the present invention and a conventional device.

FIG. 5 is a diagram showing a frequency characteristic example of a pitch conversion output in an actual circuit configuration of the pitch conversion device of the first embodiment of the present invention and a conventional device.

FIG. 6 is a block diagram showing a configuration of a pitch converting unit according to a second embodiment of the present invention.

FIG. 7 is a diagram showing a pitch conversion method according to a second embodiment of the present invention.

FIG. 8 is a diagram showing a frequency characteristic example of a pitch conversion output in a simulation of the pitch conversion device according to the second embodiment of the present invention and a conventional device.

FIG. 9 is a diagram showing a frequency characteristic example of a pitch conversion output in an actual circuit configuration of the pitch conversion device of the second embodiment of the present invention and a conventional device.

FIG. 10 is a block diagram of a pitch converting apparatus according to a third embodiment of the present invention.

FIG. 11 is a block diagram showing a configuration of a pitch converting section according to a third embodiment of the present invention.

FIG. 12 is a diagram showing a frequency characteristic example of a pitch conversion output in a simulation of the pitch conversion device of the third exemplary embodiment of the present invention and the pitch conversion device of the first exemplary embodiment.

FIG. 13 is a diagram showing frequency characteristic examples of pitch conversion output in actual circuit configurations of the pitch conversion device of the third exemplary embodiment of the present invention and the pitch conversion device of the first exemplary embodiment.

FIG. 14 is a block diagram showing a configuration of a pitch converting section according to a fourth embodiment of the present invention.

FIG. 15 is a diagram showing frequency characteristic examples of pitch conversion output in simulation of the pitch conversion device of the fourth exemplary embodiment of the present invention and the pitch conversion device of the second exemplary embodiment.

FIG. 16 is a diagram showing an example of frequency characteristics of pitch conversion output in actual circuit configurations of the pitch conversion device of the fourth embodiment of the present invention and the pitch conversion device of the second embodiment.

FIG. 17 is a diagram showing an example in which the fade-in / fade-out time of crossfade is adjusted by an output signal length adjustment method that multiplies the output signal length by n / (n + 1).

FIG. 18 is a block diagram showing a configuration of a fifth exemplary embodiment of the present invention.

FIG. 19 is a block diagram showing a configuration of output signal length adjusting means according to fifth, sixth and seventh embodiments of the present invention.

FIG. 20 is a diagram showing a frequency characteristic example of an output signal in a simulation of the pitch conversion device of the fifth embodiment of the present invention and a conventional pitch conversion device.

FIG. 21 is a block diagram showing a configuration of a sixth exemplary embodiment of the present invention.

FIG. 22 is a diagram showing an example of frequency characteristics of an output signal in a simulation of the pitch conversion device according to the sixth embodiment of the present invention and a conventional pitch conversion device.

FIG. 23 is a block diagram showing a configuration of a seventh exemplary embodiment of the present invention.

FIG. 24 is a diagram showing an example of frequency characteristics of an output signal in a simulation of a pitch conversion device according to a seventh embodiment of the present invention and a conventional pitch conversion device.

FIG. 25 is a block diagram of a conventional pitch converting device.

FIG. 26 is a block diagram showing a configuration of a pitch converting section of a conventional pitch converting device.

FIG. 27 is a diagram showing a pitch conversion method in a conventional pitch conversion device.

FIG. 28 is a diagram showing an example of input and output frequency characteristics of a conventional pitch converting apparatus.

FIG. 29 is a block diagram of a conventional pitch converting apparatus provided with an output signal length adjusting means for making the input signal length equal to the output signal length.

FIG. 30 is a diagram showing an output signal length adjusting method for multiplying the conventional output signal length by n / (n + 1).

FIG. 31 is a diagram showing a conventional output signal length adjusting method for multiplying the output signal length by n / (n−1).

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Input signal 2 A / D converter 3 Pitch conversion means 4 Pitch input means 5 D / A converter 6 Output signal 7 First filter means 8 Oversampling means 9 Second filter means 10 Downsampling means 11 Data length adjustment Means 12 Interpolation coefficient calculation means 14 Interpolation coefficient control means 15 Signal frequency component detection means 16 Output signal length adjustment means 18 Pitch conversion section 19 Coefficient memory

Claims (10)

[Claims] 1. A signal input means for inputting a digital signal, a pitch input means for inputting a target pitch at the time of pitch conversion for the input digital signal, and an interpolation coefficient for outputting an interpolation coefficient based on the input target pitch. Control means,
Data length adjusting means for adjusting the data length of the input digital signal so that the input digital signal is converted into a pitch of the input target pitch, and an output of the data length adjusting means is adjusted by the interpolation coefficient control. A pitch conversion device comprising: an interpolation calculation means for interpolating with an interpolation coefficient output from the means; and a signal output means for outputting the interpolated signal. 2. A signal input means for inputting a digital signal, a pitch input means for inputting a target pitch at the time of pitch conversion for the input digital signal, and an interpolation coefficient for outputting an interpolation coefficient based on the input target pitch. Control means,
Interpolation calculation means for interpolating the input digital signal with the interpolation coefficient output from the interpolation coefficient control means; and the interpolation so as to convert the output of the interpolation calculation means into a pitch of the input target pitch. A pitch converting apparatus comprising: a data length adjusting means for adjusting a data length of an output of the computing means; and a signal outputting means for outputting an output signal of the data length adjusting means. 3. The pitch conversion device according to claim 1, wherein the interpolation coefficient is a pitch conversion ratio corresponding to the target pitch. 4. The signal input means includes oversampling means for oversampling the input digital signal, and a low frequency band for passing a frequency lower than half the sampling frequency of the input digital signal in the output of the oversampling means. A first filter unit having a pass characteristic, wherein the signal output unit passes a frequency lower than a half of a sampling frequency of the input digital signal among signals input to the signal output unit. 4. A second filter means having a pass characteristic and down-sampling means for down-sampling an output signal from the second filter means.
The described pitch conversion device. 5. The data length adjusting means reduces the output data length with respect to the input data by one piece of data for the number of pieces of data determined by the target pitch input to the pitch input means. The pitch conversion device according to claim 1, wherein the process is performed so as not to output. 6. The data length adjusting means sets one piece of data for the number of pieces of data determined by the target pitch input to the pitch input means so as to increase the output data length of the input data. The pitch converting apparatus according to any one of claims 1 to 4, wherein processing is performed so as to replenish. 7. A pitch input means for inputting a target pitch at the time of pitch conversion for an input digital signal, a pitch conversion means for converting a pitch for the input digital signal based on the input target pitch, and the input. A signal frequency component detecting means for detecting a frequency component of the digital signal and at least the most distortion according to the detected frequency component so that the signal length of the output of the pitch converting means becomes the signal length of the input digital signal. And an output signal length adjusting means for performing a crossfade process that reduces noise. 8. A pitch input means for inputting a target pitch at the time of pitch conversion for an input digital signal, a pitch converting means for pitch-converting the input digital signal based on the input target pitch, and the pitch. Output signal length adjusting means for performing crossfade processing so that the signal length of the output of the converting means becomes the signal length of the input digital signal, and signal frequency component detecting means for detecting the frequency component of the output of the output signal length adjusting means. And a pitch characterized in that the output signal length adjusting means performs the crossfade processing such that at least distortion and noise are minimized, in accordance with the frequency component detected by the signal frequency component detecting means. Converter. 9. A pitch input means for inputting a target pitch at the time of pitch conversion for an input digital signal, a pitch converting means for converting a pitch for the input digital signal based on the input target pitch, and the pitch. Output signal length adjusting means for performing crossfade processing so that the signal length of the output of the converting means becomes the signal length of the input digital signal, and the frequency component of the output of the output signal length adjusting means and the frequency component of the input digital signal. And a signal frequency component detecting means for detecting, wherein the output signal length adjusting means, in accordance with the frequency component detected by the signal frequency component detecting means, performs the cross-fade process such that at least distortion and noise are minimized. A pitch conversion device characterized by performing. 10. A memory for storing predetermined condition data that minimizes distortion and noise in the crossfade processing, corresponding to a frequency component of an input digital signal, and the output signal length adjusting means, 10. The pitch conversion according to claim 7, 8 or 9, wherein the cross-fade processing is performed based on the frequency component detected by the signal frequency component detection means by using the condition data stored in the memory. apparatus.