JP2003015695A - Device for expanding audio frequency band - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve reproduced sound quality by expanding a reproduction frequency band to digital audio signals the high frequency reproducing limit of which is decided by the sampling theorem. SOLUTION: Digital audio signals are given to an over-sampling type LPF(low-pass filter) 102 for over-sampling and frequency band limitation. Moreover, a band-pass filter 103, a rectifier circuit 104 and a high-pass filter 105 are used for generating higher harmonics higher than the frequency band of the input signals. A level control circuit 107 performs level control on the basis of the spectral intensity of the high frequency band components of the input signals detected with a spectrum analysis circuit 106, and an adder circuit 108 adds the higher harmonics to the input signal. Next, the over-sampling means 109 performs over-sampling at a still higher sampling frequency, and outputs audio signals of an expanded frequency band through an LPF 110.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an audio band expanding device capable of reproducing a reproduced sound of an audio signal in an audio device, particularly a reproduced sound quality of a high frequency range and reproducing an audio signal which is comfortable for human ears. In particular, the present invention relates to an audio band extension technique for processing digital audio data in the digital domain.

[0002]

2. Description of the Related Art In nature, there are sounds having a frequency band of 100 kHz or more. Further, although it has been said that the human upper limit frequency of audibility is 20 kHz, recent research has revealed that it perceives sound at 20 kHz or higher. However, in digital audio, the upper limit of high frequency reproduction is determined by the sampling theorem. In the case of a compact disc (hereinafter referred to as a CD), 22.05 kHz is the upper limit. In the case of a CD, band components above the high-frequency reproduction limit may be deleted, which may cause an unnatural hearing. For this reason, a technique has been proposed in which a component above the upper limit of high frequency reproduction is added.

Japanese Patent Application No. 11-36685 discloses a technique for adding a signal having a spectrum with a frequency exceeding the upper limit of the reproduction frequency band or the upper limit of the audible frequency band to a digital audio signal. Audio band extension method and apparatus.

FIG. 8 is a block diagram of a conventional audio band extending apparatus having the above-mentioned function. This audio band extending apparatus includes an input terminal 801, a digital oversampling low pass filter (oversampling LPF) 802, a band pass filter 803, a rectifier circuit 804, a high pass filter 805, a spectrum analysis circuit 806, a level control circuit 807, and an addition circuit. 808 and an output terminal 809 are included.

The operation of the conventional audio band extending apparatus having such a configuration will be described. Input terminal 801
A digital audio signal is input through. If this signal is reproduced from a CD, it is a signal having a sampling frequency fs0 = 44.1 kHz and a word length of 16 bits. FIG. 9A is an example of a signal spectrum at the input terminal 801.

The oversampling type LPF 802 is
The sampling frequency fs0 of the signal input through the input terminal 801 is multiplied by p (p is a number of 2 or more), and an unnecessary band, that is, a band of 1/2 or more of the sampling frequency fs0 of the input signal is attenuated by 60 dB or more. Let Figure 9
(B) is an example of an output spectrum of the oversampling type LPF 802. In this example, p = 4, that is, the sampling frequency f of the oversampling type LPF 802.
s2 is set to 4fs0.

Next, the bandpass filter 803 limits the output band of the oversampling type LPF 802.
That is, the bandpass filter 803 has fs0 / 4 to fs.
A signal having a band of 0/2 is output. Then, the rectifier circuit 804 rectifies the output of the bandpass filter 803 by a half wave or both waves to generate a harmonic wave of the input signal. Then, the high-pass filter 805 cuts off the low-frequency component of the output of the rectifier circuit 804 and outputs a signal having a spectrum of fs0 / 2 or more.

On the other hand, the spectrum analysis circuit 806 detects the spectrum intensity of the high frequency component of the output of the oversampling type LPF 802. Here, fs0 / 4 to fs
A spectral intensity of 0/2 is detected. Then, the level control circuit 807 controls the output level of the high pass filter 805 according to the output of the spectrum analysis circuit 806.
Here, when the spectrum intensity in the input signal fs0 / 4 to fs0 / 2 is large, the high-pass filter 805
Output level is increased, and if the spectrum intensity is low, the output level is decreased. FIG. 9C shows an example of the output spectrum of the level control circuit 807.

The adder circuit 808 outputs the output of the oversampling type LPF 802 and the level control circuit 807.
And the output of the above are added, and the addition signal is output via the output terminal 809. FIG. 9D shows the signal spectrum at the output terminal 809.

As described above, the audio band is expanded by generating a harmonic having a spectrum equal to or more than the band of the input signal and adding a harmonic component to the input signal according to the high frequency spectrum intensity of the input signal. can do.

[0011]

However, the above-mentioned conventional configuration has a problem that the amount of processing thereafter increases in accordance with the oversampling ratio p of the oversampling type LPF 802. That is, assuming that the processing amount when ap is 2 is aMIPS and the p becomes 4 as in the above-described conventional example, the spectrum analysis circuit 80 of FIG.
6. The processing amount of the band pass filter 803, the rectifier circuit 804, the high pass filter 805, and the level control circuit 807 also increases to about 2a MIPS.

The present invention has been made in view of such conventional problems, and it is an object of the present invention to provide an audio band expanding apparatus capable of suppressing an increase in processing amount even when the oversampling ratio p is large. And

[0013]

According to a first aspect of the present invention, an audio signal sampled at a frequency fs0 is input, oversampling is performed at a frequency fs1 (fs1> fs0), and fs0 / n (n is an integer). ) Inputting the output signal of the first oversampling means for passing only the following band A component and the first oversampling means, the audio signal of the band A is transmitted for a predetermined time T.
A non-linear circuit that receives the output signal of the first over-sampling means and the spectrum analysis means that analyzes the spectrum in the band A by dividing the output signal into a distorted output signal and outputs an audio signal including a harmonic component. Means, a high-pass filter for extracting a component of the band B located higher than the band A from the output signal of the non-linear means, and an output level of the high-pass filter according to the analysis result of the spectrum analyzing means. A level control means for controlling and outputting the output signal, an output signal addition means for adding the output of the high pass filter controlled by the level control means and the output of the first oversampling means, and the output signal addition A second audio signal for oversampling the audio signal output from the means at a frequency fs2 (fs2> fs1). It is characterized in that it comprises the over-sampling means.

According to a second aspect of the present invention, the audio signal sampled at the frequency fs0 is input, and the frequency fs is input.
First oversampling means for oversampling at s1 (fs1> fs0) and passing only a component of band A of fs0 / n (n is an integer) or less, and an output signal of the first oversampling means are input. Then, the audio signal of the band A is divided at every predetermined time T, the spectrum analysis means for analyzing the spectrum in the band A, and the dither generation for generating the dither in the band B located higher than the band A. Means, level control means for controlling and outputting the output level of the dither generation means according to the analysis result of the spectrum analysis means, output of the dither generation means controlled by the level control means, and the first Output signal adding means for adding the output of the oversampling means, and an audio signal output from the output signal adding means. Te, frequency fs2 (fs2> f
second oversampling means for performing oversampling in s1).

According to the invention of claim 3 of the present application, the audio signal sampled at the frequency fs0 is inputted, and the frequency fs is inputted.
First oversampling means for oversampling at s1 (fs1> fs0) and passing only a component of band A of fs0 / n (n is an integer) or less, and an output signal of the first oversampling means are input. Then, the audio signal in the band A is divided at every predetermined time T, the spectrum analysis means for analyzing the spectrum in the band A and the output signal of the first oversampling means are input to distort the input signal. Non-linear means for outputting an audio signal containing a harmonic component, a high-pass filter for taking out a component of band B located higher than the band A from the output signal of the non-linear means, and a position higher than the band A Dither generation means for generating dither in the band B for
High-frequency signal adding means for adding the output of the high-pass filter and the output of the dither generating means, and the output level of the high-frequency signal adding means is controlled and output according to the analysis result of the spectrum analyzing means. Output from the output signal addition means, and output signal addition means for adding the output of the high frequency signal addition means controlled by the level control means and the output of the first oversampling means. Frequency fs2
Second oversampling with (fs2> fs1)
And oversampling means of.

According to a fourth aspect of the present invention, in the audio band extending apparatus according to any one of the first to third aspects, the sampling frequency fs1 of the first oversampling means is the sampling frequency fs0 of the input audio signal. It is characterized in that it has a value that is at least twice that of the above, and removes aliasing components due to oversampling processing and outputs.

The invention of claim 5 of the present application is the same as claim 1 or 3
In the audio band extending apparatus, the operation word length and its output in the non-linear means and the high pass filter are set to be larger than the word length of the input audio signal.

The invention of claim 6 of the present application is the invention of claim 2 or 3.
In the audio band extending apparatus, the arithmetic word length and its output in the dither generation means are set to be larger than the word length of the input audio signal.

The invention of claim 7 of the present application is the invention of claim 1 or 3.
In the audio band extending apparatus, the operation word length and its output in the non-linear means and the high pass filter are larger than the word length of the input audio signal, and the output signal adding means is the LSB of the word length of the input audio signal.
A band extension signal having an amplitude substantially the same as the width is added to the output of the first oversampling means.

The invention of claim 8 of the present application is the invention of claim 2 or 3.
In the audio band extending apparatus, the operation word length and its output in the dither generation means are larger than the word length of the input audio signal, and the output signal addition means has an amplitude substantially equal to the LSB width of the word length of the input audio signal. The band extension signal of is added to the output of the first oversampling means.

The invention of claim 9 of the present application is the same as claim 1 or 3
In the audio band extending device, the non-linear means performs non-linear processing on either the half-clip means for extracting only one side with respect to the reference level of the signal changing to the positive or negative side or the absolute value converting means for returning the amplitude of the one side to the opposite side. It is characterized by being used as.

The invention of claim 10 of the present application is the same as claims 1 to 3.
In the audio band extending apparatus according to any one of the items 1 to 3, the spectrum analysis means analyzes whether the input audio signal is a single spectrum signal or a plurality of spectrum signals.

According to the invention of claim 11 of the present application, in the audio band extending apparatus of claim 2 or 3, the dither generating means is a diamond dither generating means having a probability distribution of a triangular distribution within a predetermined amplitude. It is a feature.

According to a twelfth aspect of the present invention, in the audio band extending apparatus according to the eleventh aspect, the diamond dither generation means outputs two independent PN sequence generation means PA and PN sequence generation means PB. Is added.

According to a thirteenth aspect of the present invention, in the audio band extending apparatus according to the eleventh aspect, the diamond dither generation means is one PN sequence generation means PP which is different from each other and does not overlap with each other.
It is characterized by adding and.

According to a fourteenth aspect of the present invention, in the audio band extending apparatus according to the second or third aspect, the dither generating means is a bell type dither generating means in which the probability density has a bell-shaped distribution within a predetermined amplitude. It is characterized by that.

According to a fifteenth aspect of the present invention, in the audio band extending apparatus according to the fourteenth aspect, the bell-type dither generation means is three independent PN sequence generation means PA,
It is characterized in that the respective outputs of the PN sequence generating means PB and the PN sequence generating means PC are added.

According to a sixteenth aspect of the present invention, in the audio band extending apparatus according to the fourteenth aspect, the bell-type dither generating means is one PN sequence generating means PP, which are different from each other and do not overlap with each other. And a partial element PPC are added.

According to a seventeenth aspect of the present invention, in the audio band extending apparatus according to the second or third aspect, the dither generation means outputs a dither in which the spectrum distribution of the band B has a 1 / f characteristic. It is what

According to a eighteenth aspect of the present invention, in the audio band extending apparatus according to the second or third aspect, the level control means determines that only a single spectrum exists based on the analysis result of the spectrum analysis means. Sets the output level of the dither generation means to zero, and when it is determined that there are a plurality of spectra, outputs so that the spectrum of the dither of the band B is continuous with the highest band of the band A. It is characterized by controlling the level.

According to a nineteenth aspect of the invention of the present application, in the audio band extending apparatus according to the second or third aspect, when the level control means determines that a plurality of spectra are present based on the analysis result of the spectrum analysis means. Is the band A and the band B based on the slope of the spectral envelope.
It is characterized in that the dither level is controlled so that the dither continues between the two.

According to a twentieth aspect of the present invention, in the audio band extending apparatus according to the second or third aspect, the level control means determines that the signal in the band A is zero based on the analysis result of the spectrum analysis means. In this case, the dither level is muted to zero.

According to a twenty-first aspect of the invention of the present application, in the audio band extending apparatus according to the second or third aspect, the level control means smooths the analysis output of the spectrum analysis means by using a predetermined smoothing filter to obtain a dither level. It is characterized by slowing down the fluctuation of.

According to a twenty-second aspect of the present invention, in the audio band extending apparatus according to the twenty-first aspect, the smoothing filter controls an attack characteristic and a release characteristic, respectively.

[0035]

BEST MODE FOR CARRYING OUT THE INVENTION An audio band expanding apparatus according to each embodiment of the present invention will be described below with reference to the drawings. (Embodiment 1) FIG. 1 is a block diagram of an audio band extending apparatus in Embodiment 1 of the present invention. This audio band extending apparatus includes an input terminal 101, an oversampling type digital low pass filter (oversampling type LPF) 102, and a band pass filter 10.
3, a rectifier circuit 104, a high-pass filter 105, a spectrum analysis circuit 106, a level control circuit 107, an addition circuit 108, an oversampling means 109, a low-pass filter (LPF) 110, and an output terminal 111.

The oversampling type LPF 102 is
An audio signal sampled at a frequency fs0 is input, oversampling is performed at a frequency fs1 (fs1> fs0), and a band A equal to or less than fs0 / n (n is an integer) is input.
This is the first oversampling means for passing only the component of.

The spectrum analysis circuit 106 inputs the output signal of the oversampling type LPF 102 and outputs the band A
Is a spectrum analysis unit that analyzes the spectrum in the band A by dividing the audio signal of the above into predetermined time intervals T.

Bandpass filter 103 and rectifier circuit 1
Reference numeral 04 has a function of non-linear means. That is, the output signal of the oversampling type LPF 102 is input, the output signal is distorted, and an audio signal including a harmonic component is output. The high-pass filter 105 is a high-pass filter that extracts a component of band B located higher than band A from the output signal of the non-linear means.

The level control circuit 107 receives the high pass filter 10 according to the analysis result of the spectrum analysis circuit 106.
5 is a level control means for controlling and outputting the output level of No. 5. The adder circuit 108 is an output signal adder that adds the output of the high-pass filter 105 controlled by the level control circuit 107 and the output of the oversampling type LPF 102.

The oversampling means 109 is a second oversampling means for oversampling the audio signal output from the adder circuit 108 at a frequency fs2 (fs2> fs1).

The operation of the audio band expanding apparatus having such a configuration will be described. A digital audio signal is input through the input terminal 101. If this signal is reproduced from a CD, it is a signal having a sampling frequency fs0 = 44.1 kHz and a word length of 16 bits. An example of this spectrum is shown in FIG. According to the Nyquist sampling theorem, the spectrum is bilaterally symmetrical with fs0 / 2 as the central axis.

The oversampling type LPF 102 is
The sampling frequency of the signal input via the input terminal 101 is multiplied by p (p is a positive number), and an unnecessary band is attenuated. Here, p = 2, which is different from p = 4 in the conventional example. The oversampling type LPF 102 has a band of ½ or more of the sampling frequency of the input signal,
It is assumed that the LPF function attenuates 60 dB or more.

Next, the bandpass filter 103 limits the output band of the oversampling type LPF 102.
If the sampling frequency of the input signal is fs0, the bandpass filter 103 outputs a signal having a band of fs0 / 4 to fs0 / 2. Then, the rectifier circuit 104 generates a harmonic component of the input signal by rectifying the output of the bandpass filter 103 by half wave or both waves. The high-pass filter 105 blocks low-frequency components of the output of the rectifier circuit 104 and outputs a signal having a spectrum of fs0 / 2 or more.

On the other hand, the spectrum analysis circuit 106 detects the spectrum intensity of the high frequency component of the output of the oversampling type LPF 102. Here, fs0 / 4 to fs
A spectral intensity of 0/2 is detected. Then, the level control circuit 107 controls the output level of the high pass filter 105 according to the output (analysis result) of the spectrum analysis circuit 106. Here, fs0 / 4 to fs0 of the input signal
When the spectrum intensity at / 2 is high, the output level of the high-pass filter 105 is increased, and when the spectrum intensity is low, the output level is decreased.

The adder circuit 108 outputs the output of the oversampling type LPF 102 and the level control circuit 107.
And the output of. An example of the signal spectrum added here is shown in FIG. A1 and A2 are output components of the oversampling type LPF 102, and B1
And B2 are components output from the level control circuit 107.

The oversampling means 109 further oversamples the output signal of the adder circuit 108 at the frequency fs2 = 2fs1. FIG. 2C shows an example of the output signal spectrum of the oversampling means 109. This is an example of double oversampling with respect to fs1. In FIG. 2C, the band from 0 to fs1 / 2 is the output signal itself of the adder circuit 108,
The band from fs1 / 2 to fs2 / 2 is a loopback of the band from 0 to fs1 / 2. Here, the spectrum at fs1 / 2 has continuous values. In this way, fs1
A harmonic component having continuous spectrum intensity can be obtained even in a band of / 2 or more.

In the low pass filter 110, fs1 / 2
The above components are attenuated. As a result, a signal whose spectrum intensity increases again at fs1 / 2 or more is converted into fs1 /
Shape to a natural spectrum that decreases from 2 to fs2 / 2. The output signal of the low-pass filter 110 is output via the output terminal 111. Low pass filter 110
2 (d) shows an example of the spectrum of the output signal of FIG. Here, the image component centering on fs2 is also shown.

In this way, a harmonic having a spectrum above the band of the input signal is generated, and the generated harmonic component is added to the input signal according to the high-frequency spectrum intensity of the input signal, and then oversampled. Then, Fig. 2
The audio band can be extended as indicated by the portions B1, C1, and C2 in (d).

As described above, the audio band extending apparatus according to the present embodiment has a bandpass filter, a rectifying circuit,
A high-pass filter is used to generate harmonics above the input signal band, and the spectrum analysis circuit and level control circuit control the level according to the high-frequency spectrum intensity of the input signal to add it to the input signal and then oversample. I am trying to output it.

Therefore, after performing harmonic generation that requires a large amount of processing at the frequency fs1, oversampling to the frequency fs2 is performed, so that the fs can be directly measured as in the conventional example.
As compared with the method of generating harmonics in No. 2, it is possible to generate harmonics with a smaller processing amount. For example, fs1 = 2 × fs
0, fs2 = 4 × fs0, the method of the present embodiment can extend the bandwidth with a processing amount of about ½ that of the conventional method.

Further, since the signal processing is digital processing, there is no variation in the components of the circuit or variation in performance due to temperature characteristics. Further, the sound quality does not deteriorate every time the audio signal passes through the circuit. Furthermore,
Even if the accuracy of the filter is pursued, the circuit scale does not become larger than that of the analog circuit configuration. Therefore, it is possible to realize an audio band extending device which does not lead to an increase in cost.

In this embodiment, after the band of the input signal is limited by the bandpass filter 103, the rectifying circuit 10
4 generated harmonics. However, it is needless to say that even if the rectifier circuit 104 generates harmonics without passing through the bandpass filter 103, the highpass filter 105 cuts off low frequencies, and thus the same effect can be obtained.

(Embodiment 2) Next, an audio band extending apparatus according to Embodiment 2 of the present invention will be described.
FIG. 3 is a block diagram of the audio band extending apparatus in the second embodiment. This audio band expanding device includes an input terminal 301, an oversampling type LPF 302,
It includes a dither generation circuit 303, a high pass filter 304, a spectrum analysis circuit 305, a level control circuit 306, an addition circuit 307, an oversampling means 308, a low pass filter 309, and an output terminal 310.

The blocks having the same functions as those in the first embodiment are given the same names, and their functional descriptions are omitted.
The dither generation circuit 303 is a dither generation unit that generates dither in the band B located higher than the band A. The level control circuit 306 is a level control unit that controls and outputs the output level of the dither generation circuit 303 output via the high pass filter 304 according to the analysis result of the spectrum analysis circuit 305.

The operation of the audio band expanding apparatus having such a configuration will be described. A digital audio signal is input through the input terminal 301. If this signal is reproduced from a CD, it is a signal having a sampling frequency fs0 = 44.1 kHz and a word length of 16 bits.

The oversampling type LPF 302 is
The sampling frequency of the signal input via the input terminal 301 is multiplied by p (p is a positive number), and an unnecessary band is attenuated. Here, oversampling type LPF 302
Suppose that the band of ½ or more of the sampling frequency of the input signal is attenuated by 60 dB or more.

Next, the dither generation circuit 303 outputs 0 to pfs.
White noise in which the spectral intensity is uniformly distributed in the band of / 2 is generated as dither. Then, the high-pass filter 304 band-limits the output of the dither generation circuit 303 and outputs noise having a frequency band of fs0 / 2 or more.

On the other hand, the spectrum analysis circuit 305 detects the spectrum intensity of the high frequency component of the output of the oversampling type LPF 302. Here, fs0 / 4 to fs
A spectral intensity of 0/2 is detected. Then, the level control circuit 306 controls the output level of the high pass filter 304 according to the output of the spectrum analysis circuit 305. Here, the output level of the high-pass filter 304 is increased when the spectrum intensity of the input signal fs0 / 4 to fs0 / 2 is large, and the output level is reduced when the spectrum intensity is small.

The adder circuit 307 outputs the output of the oversampling type LPF 302 and the level control circuit 306.
And the output of. Oversampling means 308
Further oversamples the output signal of the adder circuit 307 at the frequency fs2. The low-pass filter 309 is
F with respect to the output signal of the oversampling means 308
Attenuates components of s1 / 2 or more. As a result, fs1
A signal whose spectral intensity increases again at / 2 or more,
Shapes to a natural spectrum that decreases from fs1 / 2 to fs2 / 2. The output signal of the low-pass filter 309 is output via the output terminal 310.

In this way, a dither having a spectrum larger than the band of the input signal is generated, the dither is added to the input signal according to the high-frequency spectrum intensity of the input signal, and then oversampling is performed to extend the audio band. can do.

As described above, the audio band expanding apparatus according to this embodiment uses the dither generating circuit and the high-pass filter to generate high frequency dither over the band of the input signal, and the spectrum analyzing circuit and the level control circuit generate the input signal. The level is controlled according to the intensity of the high-frequency spectrum of the signal, added to the input signal, further oversampled, and output.

Therefore, as compared with the conventional method in which the harmonics are directly generated at the frequency fs2 by oversampling to the frequency fs2 after the harmonics which require a large amount of processing are generated at the frequency fs1, It is possible to generate harmonics with a small amount of processing.

Further, since the signal processing is digital processing, variations in parts constituting the circuit and performance variations due to temperature characteristics do not occur. Further, the sound quality does not deteriorate every time the audio signal passes through the circuit. Furthermore,
Even if the accuracy of the filter is pursued, the circuit scale does not become larger than that of the analog circuit configuration. Therefore, it is possible to realize an audio band extending device which does not lead to an increase in cost.

(Embodiment 3) Next, an audio band extending apparatus according to Embodiment 3 of the present invention will be described.
FIG. 4 is a block diagram of an audio band extending apparatus according to the third embodiment. This audio band expanding device includes an input terminal 401, an oversampling type LPF 402,
Band pass filter 403, rectifier circuit 404, high pass filter 405, dither generation circuit 406, high pass filter 407, adder circuit 408, spectrum analysis circuit 40.
9, a level control circuit 410, an adder circuit 411, an oversampling means 412, a low pass filter 413, and an output terminal 414.

The blocks having the same functions as those in the first or second embodiment are given the same names, and their functional descriptions are omitted. The dither generation circuit 406 is a dither generation unit that generates dither in the band B located higher than the band A.
The addition circuit 408 outputs the output signal of the high pass filter 405 and the dither generation circuit 4 via the high pass filter 407.
It is a high-frequency signal adding means for adding the signal output from 06. The level control circuit 410 is a level control unit that controls the output level of the addition circuit 408 according to the analysis result of the spectrum analysis circuit 409.

The operation of the audio band expanding apparatus having such a configuration will be described. A digital audio signal is input through the input terminal 401. If this signal is reproduced from a CD, it is a signal having a sampling frequency fs0 = 44.1 kHz and a word length of 16 bits.

The oversampling type LPF 402 is
The sampling frequency of the audio signal input via the input terminal 401 is multiplied by p (p is a positive number), and an unnecessary band is attenuated. Here, oversampling type L
The PF 402 attenuates a band of ½ or more of the sampling frequency of the input signal by 60 dB or more.

Next, the bandpass filter 403 limits the output band of the oversampling type LPF 402.
If the sampling frequency of the input signal is fs0, the bandpass filter 403 outputs a signal having a band of fs0 / 4 to fs0 / 2. Then, the rectifier circuit 404 rectifies the output of the bandpass filter 403 by a half wave or both waves to generate a harmonic wave of the input signal. The high pass filter 405 cuts off the low frequency component of the output of the rectifier circuit 404 and outputs a signal having a spectrum of fs0 / 2 or more.

Next, the dither generation circuit 406 outputs 0 to pfs.
The white noise in which the spectrum intensity is uniformly distributed in the band of / 2 is generated. Then, the high pass filter 40
7 band-limits the output of the dither generation circuit 406, and fs0
Outputs noise having a frequency band of / 2 or more. The adder circuit 408 adds the output of the high pass filter 405 and the output of the high pass filter 407.

On the other hand, the spectrum analysis circuit 409 detects the spectrum intensity of the high frequency component of the output of the oversampling type LPF 402. Here, fs0 / 4 to fs
A spectral intensity of 0/2 is detected. Level control circuit 4
Reference numeral 10 controls the output level of the addition circuit 408 according to the output of the spectrum analysis circuit 409. Here, when the spectrum intensity of the input signal fs0 / 4 to fs0 / 2 is large, the output level of the adding circuit 408 is increased,
When the spectrum intensity is low, the output level is low.

Then, the adder circuit 411 outputs the output of the oversampling type LPF 402 and the level control circuit 410.
Add to the output of. Oversampling means 412
Further oversamples the output signal of the adder circuit 411 at the frequency fs2. The low-pass filter 413 is f
Attenuates components of s1 / 2 or more. As a result, fs1
A signal whose spectral intensity increases again at / 2 or more is f
Shapes to a natural spectrum that decreases from s1 / 2 to fs2 / 2. The output signal of the low-pass filter 413 is output via the output terminal 414.

In this way, harmonics and dither having a spectrum above the band of the input signal are generated, this harmonic component is added to the input signal according to the intensity of the high frequency spectrum of the input signal, and the By sampling,
The audio band can be extended.

As described above, the audio band expanding apparatus according to the present embodiment uses the bandpass filter, the rectifier circuit, and the highpass filter to generate harmonics higher than the band of the input signal, and the dither generation circuit and the highpass filter input the harmonics. Generate high-frequency dither above the signal band, control the level according to the high-frequency spectrum intensity of the input signal with the spectrum analysis circuit and level control circuit, add to the input signal, and then oversample and output. There is.

Therefore, compared with the conventional method of directly generating the harmonics at the frequency fs2 by performing oversampling at the frequency fs2 after performing the harmonic generation requiring a large amount of processing at the frequency fs1 Higher harmonics can be generated with a throughput.

Since the signal processing is digital processing, there are no variations in the components of the circuit and variations in performance due to temperature characteristics. Further, the sound quality does not deteriorate every time the audio signal passes through the circuit. Further, even if the accuracy of the filter is pursued, the circuit scale does not become large as compared with the analog circuit configuration. Therefore, it is possible to realize an audio band extending device which does not lead to an increase in cost.

In this embodiment, the bandpass filter 403 limits the band of the input signal and then the rectifier circuit 40 is used.
4 generated harmonics. However, it is needless to say that even if the rectifier circuit 404 generates harmonics without passing through the bandpass filter 403, the highpass filter 405 blocks the low frequency band, and the same effect can be obtained.

(Embodiment 4) Next, an audio band extending apparatus according to Embodiment 4 of the present invention will be described.
FIG. 5 is a block diagram of an audio band extending apparatus according to the fourth embodiment. This audio band expanding device includes an input terminal 501, an oversampling type LPF 502,
Independent PN sequence noise generators 504, 505, 5
06 and adder circuit 507 dither generation circuit 5
03, high-pass filter 508, spectrum analysis circuit 5
09, a level control circuit 510, an adding circuit 511, an oversampling means 512, a low pass filter 513,
The output terminal 514 is included.

The blocks having the same functions as those in the first to third embodiments are given the same names, and their functional descriptions are omitted. The dither generation circuit 503 may be a diamond dither generation unit in which the probability density has a triangular distribution within a predetermined amplitude. The diamond dither generation means in this case adds the outputs of the two PN series generation means PA and the PN series generation means PB which are independent of each other.
Further, the diamond dither generation means is a partial element PP of one PN sequence generation means PP which is different from each other and does not overlap.
It is also possible to add A and the subelement PPB.

The dither generation circuit 503 shown in FIG. 5 is a bell-type dither generation means in which the probability density has a bell-shaped distribution within a predetermined amplitude. The bell-type dither generation means has three independent PN series generation means PA (PN series noise generator 504) output, PN series generation means PB (PN series noise generator 505) output, and PN series generation means PC.
The output of the (PN series noise generator 506) is added. Further, the bell-type dither generation means is different from each other in one PN sequence generation means PP and does not overlap with each other.
It is also possible to add PA, partial element PPB, and partial element PPC.

The operation of the audio band expanding apparatus having such a configuration will be described. A digital audio signal is input through the input terminal 501. If this signal is reproduced from a CD, it is a signal having a sampling frequency fs0 = 44.1 kHz and a word length of 16 bits.

The oversampling type LPF 502 is
The sampling frequency of the audio signal input via the input terminal 501 is multiplied by p (p is a positive number), and an unnecessary band is attenuated. Here, oversampling type L
The PF 502 attenuates a band of ½ or more of the sampling frequency of the input signal by 60 dB or more.

Next, the PN series noise generating circuits 504, 505 and 506 which are independent of each other generate white noise in which the spectrum intensity is uniformly distributed in the band of 0 to pfs / 2.

Then, the adding circuit 507 adds the outputs of the PN series noise generating circuits 504, 505 and 506.
The noise after the addition has a bell shape in which the probability distribution of the spectrum is close to a Gaussian distribution and approaches a natural sound. Then, the high-pass filter 508 band-limits the output of the dither generation circuit 503 and outputs noise having a frequency band of fs0 / 2 or more.

On the other hand, the spectrum analysis circuit 509 detects the spectrum intensity of the high frequency component of the output of the oversampling type LPF 502. Here, fs0 / 4 to fs
A spectral intensity of 0/2 is detected.

Then, the level control circuit 510 responds to the output of the spectrum analysis circuit 509 by the high-pass filter 5
The output level of 08 is controlled. Here, f of the input signal
When the spectrum intensity in s0 / 4 to fs0 / 2 is large, the output level of the high pass filter 508 is increased, and when the spectrum intensity is small, the output level is decreased.

The adder circuit 511 outputs the output of the oversampling type LPF 502 and the level control circuit 510.
And the output of. Oversampling means 512
Further oversamples the output signal of the adder circuit 511 at the frequency fs2. The low-pass filter 513 is
Attenuates components above fs1 / 2. As a result, fs
A signal in which the spectral intensity increases again at ½ or more is shaped into a natural spectrum that decreases from fs1 / 2 toward fs2 / 2. The output signal of the low-pass filter 513 is output via the output terminal 514.

In this way, a Gaussian distribution type dither having a spectrum above the band of the input signal is generated, the dither generated according to the high frequency spectrum intensity of the input signal is added to the input signal, and further oversampling is performed. With, the audio band can be extended.

As described above, the audio band expanding apparatus of the present embodiment uses the dither generating circuit and the high-pass filter to generate the Gaussian distribution type high frequency dither over the band of the input signal, and uses the spectrum analyzing circuit and the level control circuit. The level is controlled according to the high-frequency spectrum intensity of the input signal, added to the input signal, further oversampled, and output.

As described above, as compared with the conventional method in which the harmonics are directly generated at the frequency fs2 by oversampling to the frequency fs2 after the harmonics which require a large amount of processing are generated at the frequency fs1, It is possible to generate harmonics with a small amount of processing.

Since the added high frequency component is a Gaussian distribution type dither, it is close to the distribution in the natural world, and natural sound quality in which a specific sound is not emphasized can be obtained even if the audio band is expanded. Further, since the signal processing is digital processing, there is no variation in the components of the circuit or variation in performance due to temperature characteristics. Further, the sound quality does not deteriorate every time the audio signal passes through the circuit. Further, even if the accuracy of the filter is pursued, the circuit scale does not become large as compared with the analog circuit configuration. Therefore, it is possible to realize an audio band extending device which does not lead to an increase in cost.

Incidentally, the dither generation circuit 503 of the present embodiment.
Needless to say, when applied to the dither generation circuit 506 of the third embodiment, an effect equal to or higher than that of the present embodiment can be obtained.

(Fifth Embodiment) Next, an audio band extending apparatus according to a fifth embodiment of the present invention will be described.
FIG. 6 is a block diagram of an audio band extending apparatus in the fifth embodiment. This audio band expanding device includes an input terminal 601, an oversampling type LPF 602,
Dither generation circuit 603, high-pass filter 604, 1 /
An f characteristic filter 605, a spectrum analysis circuit 606, a level control circuit 607, an addition circuit 608, an oversampling means 609, a low pass filter 610, and an output terminal 611 are included.

The operation of the audio band expanding apparatus having such a configuration will be described. A digital audio signal is input through the input terminal 601. If this signal is reproduced from a CD, it is a signal having a sampling frequency fs0 = 44.1 kHz and a word length of 16 bits.

The oversampling type LPF 602 is
The sampling frequency of the signal input via the input terminal 601 is multiplied by p (p is a positive number), and an unnecessary band is attenuated. Here, oversampling type LPF 602
Suppose that the band of ½ or more of the sampling frequency of the input signal is attenuated by 60 dB or more.

Next, the dither generation circuit 603 outputs 0-pfs.
The white noise in which the spectrum intensity is uniformly distributed in the band of / 2 is generated. Then, the high pass filter 60
4 band-limits the output of the dither generation circuit 603, fs0
Outputs noise having a frequency band of / 2 or more. 1 / f
The characteristic filter 605 band-limits the output of the high-pass filter 604 and outputs noise so as to have a 1 / f characteristic.

On the other hand, the spectrum analysis circuit 606 detects the spectrum intensity of the high frequency component of the output of the oversampling type low pass filter 602. Here, fs0
The spectrum intensity of / 4 to fs0 / 2 is detected.

Then, the level control circuit 607 controls the output level of the 1 / f characteristic filter 605 according to the output of the spectrum analysis circuit 606. Here, the output level of the 1 / f characteristic filter 605 is increased when the spectrum intensity of the input signal fs0 / 4 to fs0 / 2 is large, and is decreased when the spectrum intensity is small. Then, the adding circuit 608 adds the output of the oversampling type LPF 602 and the output of the level control circuit 607.

The oversampling means 609 further oversamples the input signal at the frequency fs2. The low-pass filter 610 attenuates the components of fs1 / 2 or more. As a result, a signal whose spectral intensity increases again at fs1 / 2 or higher is output from fs1 / 2 to fs1 / 2.
Shape to a natural spectrum that decreases toward 2/2.
The output signal of the low-pass filter 610 is output via the output terminal 611.

In this way, the dither having the 1 / f characteristic having the spectrum above the band of the input signal is generated, the generated dither is added to the input signal according to the high frequency spectrum intensity of the input signal, and the oversampling is performed. by doing,
The audio band can be extended.

As described above, the audio band extending apparatus of this embodiment has the dither generating circuit, the high pass filter,
A 1 / f characteristic filter is used to generate high-frequency dither over the band of the input signal, the spectrum analysis circuit and the level control circuit control the level according to the high-frequency spectrum intensity of the input signal, and the sum is added to the input signal. I am trying to output after oversampling.

As described above, as compared with the conventional method in which the harmonics are directly generated at the frequency fs2 by oversampling to the frequency fs2 after the harmonics which require a large amount of processing are generated at the frequency fs1, It is possible to generate harmonics with a small amount of processing.

Further, since the dither having the 1 / f characteristic is used, it becomes close to the sound in the natural world, and even if the audio band is expanded, a natural sound quality in which a specific sound is not emphasized can be obtained. Further, since the signal processing is digital processing, there is no variation in the components of the circuit or variation in performance due to temperature characteristics. Further, the sound quality does not deteriorate every time the audio signal passes through the circuit. Further, even if the accuracy of the filter is pursued, the circuit scale does not become large as compared with the analog circuit configuration. Therefore, it is possible to realize an audio band extending device which does not lead to an increase in cost.

(Sixth Embodiment) Next, an audio band extending apparatus according to a sixth embodiment of the present invention will be described.
FIG. 7 is a block diagram of an audio band extending apparatus in the sixth embodiment. This audio band expanding device includes an input terminal 701, an oversampling type LPF 702,
Band pass filter 703, rectifier circuit 704, high pass filter 705, dither generation circuit 706, high pass filter 707, addition circuit 708, spectrum analysis circuit 70.
9, a level control circuit 710, a switch 711, an addition circuit 712, an oversampling means 713, a low pass filter 714, and an output terminal 715.

The operation of the audio band expanding apparatus configured as described above will be described. A digital audio signal is input through the input terminal 701. If this signal is reproduced from a CD, it is a signal having a sampling frequency fs0 = 44.1 kHz and a word length of 16 bits.

The oversampling type LPF 702 is
The sampling frequency of the signal input through the input terminal 701 is multiplied by p (p is a positive number), and an unnecessary band is attenuated. Here, oversampling type LPF 702
Suppose that the band of ½ or more of the sampling frequency of the input signal is attenuated by 60 dB or more.

Next, the bandpass filter 703 limits the output band of the oversampling type LPF 702.
If the sampling frequency of the input signal is fs0, the bandpass filter 703 outputs fs0 / 4 to fs0 / 2.
A signal having a band of is output. Then, the rectifier circuit 70
Reference numeral 4 rectifies the output of the bandpass filter 703 by a half wave or both waves to generate a harmonic wave of the input signal. Then, the high-pass filter 705 cuts off the low frequency component of the output of the rectifier circuit 704 and outputs a signal having a spectrum of fs0 / 2 or more.

Next, the dither generation circuit 706 outputs 0 to pfs.
The white noise in which the spectrum intensity is uniformly distributed in the band of / 2 is generated. Then, the high pass filter 70
7 band-limits the output of the dither generation circuit 706, and fs0
Outputs noise having a frequency band of / 2 or more. The adder circuit 708 adds the output of the high pass filter 705 and the output of the high pass filter 707.

On the other hand, the spectrum analysis circuit 709 detects the spectrum intensity of the high frequency component of the output of the oversampling type LPF 702. Here, the spectrum intensities of 0 to fs1 / 4 and fs0 / 4 to fs0 / 2 are detected.
Then, the level control circuit 710 controls the output level of the addition circuit 708 according to the output of the spectrum analysis circuit 709. Here, fs0 / 4 to fs0 / 2 of the input signal
If the spectrum intensity at is large, the adding circuit 708
The output level is increased, and when the spectrum intensity is low, the output level is decreased.

The switch 711 is the spectrum analysis circuit 70.
It turns on and off according to the output of 9. Here, the spectrum intensity at 0 to fs / 4 of the input signal and fs0 / 4 to
Comparing the spectral intensities at fs0 / 2, 0-f
The spectral intensity of s1 / 4 is equal to or higher than a predetermined level and fs
If the spectrum intensity of 0/4 to fs0 / 2 is below a predetermined level, or if the spectrum intensity of 0 to fs / 4 is below a predetermined level and the spectrum intensity of fs0 / 4 to fs0 / 2 is above a predetermined level, the switch 711. Off,
Otherwise, the switch 711 is turned on.

Then, the adder circuit 712 adds the output of the oversampling type LPF 702 and the output of the switch 711. In the oversampling means 713,
Further oversample the input signal. The low-pass filter 714 attenuates the components of fs1 / 2 or more. As a result, a signal whose spectral intensity is increased again at fs1 / 2 or higher is shaped into a natural spectrum that decreases from fs1 / 2 toward fs2 / 2. The output signal of the low-pass filter 714 is output via the output terminal 715.

As described above, a harmonic and a dither having a spectrum above the band of the input signal are generated, a harmonic component is added to the input signal according to the high frequency spectrum intensity of the input signal, and further oversampling is performed. With, the audio band can be extended. Further, when a signal having a single spectrum such as a sine wave is input, the switch 711 is turned off so that the signal does not deteriorate.

As described above, the audio band extending apparatus of the present embodiment uses the bandpass filter, the rectifier circuit, and the highpass filter to generate harmonics higher than the band of the input signal, and the dither generation circuit and the highpass filter. Generate a high frequency dither above the input signal band by using the spectrum analysis circuit and level control circuit to mute the level according to the high frequency spectrum intensity of the input signal and add it to the input signal, then oversample and output. I am trying to do it.

Therefore, as compared with the conventional method of directly generating harmonics at frequency fs2 by performing oversampling at frequency fs2 after performing harmonic generation requiring a large amount of processing at frequency fs1, the number of harmonics is smaller. Higher harmonics can be generated with a throughput.

Therefore, when a sine wave having a single spectrum is input, the audio band expanding function is stopped so that signal deterioration does not occur. In addition, since the signal processing is digital processing, performance variation does not occur due to variations in the components forming the circuit and temperature characteristics. Further, the sound quality does not deteriorate every time the audio signal passes through the circuit. Further, even if the accuracy of the filter is pursued, the circuit scale does not become large as compared with the analog circuit configuration. Therefore, it is possible to realize an audio band extending device which does not lead to an increase in cost.

In this embodiment, the bandpass filter 703 limits the band of the input signal and then the rectifier circuit 704 is used.
Although the harmonics are generated in step 1, the harmonics may be generated in the rectifier circuit 704 without passing through the bandpass filter 703. In this case, needless to say, the same effect can be obtained because the high-pass filter 705 blocks low frequencies.

In the above embodiment, a smoothing filter may be provided in the input section of the level control circuit to temporally smooth the analysis result obtained from the spectrum analysis circuit. This makes it possible to slow down the fluctuation of the output level of the level control circuit. The smoothing filter may be one that can control the attack characteristic and the release characteristic, respectively.

Further, the non-linear means by the bandpass filter and the rectifier circuit, the dither generation circuit, and the high-pass filter make the operation word length larger than the word length of the oversampling type LPF, and the LSB of the word length of the oversampling type LPF. You may process so that the signal component of substantially the same amplitude may be added to the width. For example, oversampling type LPF
If the word length of the audio signal output from is 16 bits, noise (dither signal) of 4 bits is added to the LSB of this signal so that it has almost the same amplitude as the 16th bit of the least significant bit. To do so.

[0118]

As described above, according to the invention as described above, a harmonic wave which requires a large amount of processing is generated at the frequency fs1 and then oversampled at the frequency fs2, so that the harmonic wave is directly generated at the frequency fs2 as in the conventional case. Compared with the method of generating, there is an effect that a harmonic component can be generated with a smaller processing amount.

Furthermore, when an audio signal having a single spectrum is input, the signal deterioration can be suppressed by stopping the extension function of the audio band. Further, since the signal processing is digital processing, it is possible to obtain an effect that variations in parts constituting the circuit and variations in performance due to temperature characteristics do not occur.

Further, it is possible to obtain the effect that the sound quality is not deteriorated every time the audio signal passes through the circuit. Furthermore, even if the accuracy of each filter is pursued, the circuit scale does not become large as compared with the analog circuit configuration. Therefore, a device which does not lead to an increase in cost can be obtained.

Further, since the Gaussian distribution type dither of 1 / f characteristic close to natural sound is used, even if the audio band is extended,
It is possible to obtain the effect of providing natural sound quality without emphasizing a specific band.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an audio band extending device according to a first embodiment of the present invention.

FIG. 2 is a frequency spectrum diagram showing the operation of the audio band extending apparatus according to the first embodiment.

FIG. 3 is a block diagram showing a configuration of an audio band extending device according to a second embodiment of the present invention.

FIG. 4 is a block diagram showing a configuration of an audio band extending device according to a third embodiment of the present invention.

FIG. 5 is a block diagram showing a configuration of an audio band extending device according to a fourth embodiment of the present invention.

FIG. 6 is a block diagram showing a configuration of an audio band expanding device according to a fifth embodiment of the present invention.

FIG. 7 is a block diagram showing a configuration of an audio band expanding device according to a sixth embodiment of the present invention.

FIG. 8 is a block diagram showing a configuration of a conventional audio band expansion device.

FIG. 9 is a frequency spectrum diagram showing the operation of a conventional audio band extending apparatus.

[Explanation of symbols]

101, 301, 401, 501, 601, 701 Input terminals 102, 302, 402, 502, 602, 702 Oversampling type digital low pass filter (oversampling type LPF) 103, 403, 703 Band pass filter 104, 404, 704 Rectification Circuits 105, 304, 405, 407, 508, 604, 7
05,707 High-pass filter 106,305,409,509,606,709 Spectrum analysis circuit 107,306,410,510,607,710 Level control circuit 108,307,408,411,511,608,7
08,712 Adder circuits 109,308,412,512,609,713 Oversampling means 110,309,413,513,610,714 Low pass filters 111,310,414,514,611 Output terminals 303,406,503,603 , 706 Dither generation circuit 504, 505, 506 PN series noise generator 605 1 / f characteristic filter 711 switch

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Kazuya Iwata             1006 Kadoma, Kadoma-shi, Osaka Matsushita Electric             Sangyo Co., Ltd. F-term (reference) 5D020 CE03                 5D045 DA20                 5J064 AA01 BA06 BB07 BC08 BC12                       BC18 BD02

Claims (22)

[Claims] 1. An audio signal sampled at a frequency fs0 is input, and a frequency fs1 (fs1> fs0) is input.
The first oversampling means for passing only the component of the band A equal to or less than fs0 / n (n is an integer), and the output signal of the first oversampling means are input to the audio signal of the band A. A signal is divided into every predetermined time T, the spectrum analysis means for analyzing the spectrum in the band A, and the output signal of the first oversampling means are input, and the output signal is distorted to include a harmonic component. Non-linear means for outputting an audio signal, a high-pass filter for extracting a component of band B located higher than the band A from the output signal of the non-linear means, and the high-pass filter according to the analysis result of the spectrum analysis means. Level control means for controlling and outputting the output level of the high-pass filter; and the high-pass filter controlled by the level control means. Output signal adding means for adding the output of the filter and the output of the first oversampling means, and oversampling the audio signal output from the output signal adding means at a frequency fs2 (fs2> fs1). Second oversampling means,
An audio band expansion device comprising: 2. An audio signal sampled at a frequency fs0 is input, and a frequency fs1 (fs1> fs0) is input.
The first oversampling means for passing only the component of the band A equal to or less than fs0 / n (n is an integer), and the output signal of the first oversampling means are input to the audio signal of the band A. A spectrum analysis unit that divides a signal at predetermined time intervals T to analyze a spectrum in the band A; a dither generation unit that generates a dither in a band B located higher than the band A; and the spectrum analysis. Level control means for controlling and outputting the output level of the dither generation means according to the analysis result of the means, output of the dither generation means and output of the first oversampling means controlled by the level control means. And an output signal adding means for adding and the frequency fs2 for the audio signal output from the output signal adding means. A second over-sampling means for performing oversampling at fs2> fs1),
An audio band expansion device comprising: 3. An audio signal sampled at a frequency fs0 is input, and a frequency fs1 (fs1> fs0) is input.
The first oversampling means for passing only the component of the band A equal to or less than fs0 / n (n is an integer), and the output signal of the first oversampling means are input to the audio signal of the band A. A signal that divides the signal at predetermined time intervals T to analyze the spectrum in the band A, and the output signal of the first oversampling means are input, and the input signal is distorted to produce an audio signal containing a harmonic component. A non-linear means for outputting a signal; a high-pass filter for extracting a component of a band B located above the band A from the output signal of the non-linear means; and a dither inside the band B located above the band A. A high-pass signal adding means for adding an output of the high-pass filter and an output of the dither generating means; Level control means for controlling and outputting the output level of the high frequency signal adding means according to the analysis result of the vector analyzing means, output of the high frequency signal adding means controlled by the level control means, and the first Output signal adding means for adding the output of the oversampling means, and second oversampling means for performing oversampling at a frequency fs2 (fs2> fs1) on the audio signal output from the output signal adding means. ,
An audio band expansion device comprising: 4. The first oversampling means has a sampling frequency fs1 that is at least twice the sampling frequency fs0 of the input audio signal, and removes and outputs a folding component due to oversampling processing. The audio band expansion device according to any one of claims 1 to 3, characterized in that: 5. The operation word length and the output thereof in the non-linear means and the high-pass filter are set to be larger than the word length of the input audio signal.
The audio band expansion device described. 6. The audio band extending apparatus according to claim 2, wherein the operation word length and the output thereof in the dither generation means are made larger than the word length of the input audio signal. 7. The operation word length and its output in the non-linear means and the high-pass filter are larger than the word length of the input audio signal, and the output signal adding means sets the LSB width of the word length of the input audio signal. The audio band expanding apparatus according to claim 1 or 3, wherein a band expanding signal having substantially the same amplitude is added to the output of the first oversampling means. 8. The operation word length and its output in the dither generation means are larger than the word length of the input audio signal, and the output signal addition means has a band of an amplitude substantially the same as the LSB width of the word length of the input audio signal. 4. The audio band extending apparatus according to claim 2, wherein the extension signal is added to the output of the first oversampling means. 9. The non-linear means uses, as non-linear processing, either half-clip means for extracting only one side with respect to a reference level of a signal that changes between positive and negative sides, or absolute value conversion means for returning the amplitude of one side to the opposite side. The audio band expansion device according to claim 1 or 3, characterized in that 10. The audio band according to claim 1, wherein the spectrum analysis means analyzes whether the input audio signal is a signal having a single spectrum or a signal having a plurality of spectra. Expansion device. 11. The dither generation means is diamond dither generation means having a triangular distribution of probability density within a predetermined amplitude.
The audio band expansion device described. 12. The diamond dither generation means outputs two PN sequence generation means PA independent from each other and a PN.
12. The audio band expanding device according to claim 11, wherein the audio band expanding device adds the output of the sequence generating means PB. 13. The audio according to claim 11, wherein the diamond dither generation means is for adding the different subelements PPA and PPB of one PN sequence generation means PP which are different from each other and do not overlap each other. Bandwidth extension device. 14. The audio band extending apparatus according to claim 2, wherein the dither generation unit is a bell-type dither generation unit in which the probability density has a bell-shaped distribution within a predetermined amplitude. 15. The bell-type dither generation means adds the respective outputs of three PN sequence generation means PA, PN sequence generation means PB, and PN sequence generation means PC which are independent of each other. Item 14. The audio band extending device according to Item 14. 16. The bell-type dither generation means comprises: a sub-element PPA, a sub-element PPB, and a sub-element PPC of one PN sequence generation means PP which are different from each other and do not overlap.
15. The audio band expanding device according to claim 14, wherein the audio band expanding device and the audio signal are added together. 17. The dither generation means outputs a dither in which the spectral distribution of the band B has a 1 / f characteristic.
The audio band expansion device described. 18. The level control unit sets the output level of the dither generation unit to zero and determines that a plurality of spectra exist when the analysis result of the spectrum analysis unit determines that only a single spectrum exists. If the determination is made, the output level is controlled so that the spectrum height of the dither of the band B is continuous with the highest spectrum of the band A. apparatus. 19. The level control means, when it is determined from the analysis result of the spectrum analysis means that a plurality of spectra are present, between the band A and the band B based on the slope of the spectrum envelope. 4. The audio band extending apparatus according to claim 2, wherein the dither level is controlled so that the dither continues. 20. The level control means mutes the dither level to zero when the signal of the band A is determined to be zero from the analysis result of the spectrum analysis means. 3. The audio band expansion device described in 3. 21. The audio according to claim 2, wherein the level control means smoothes the analysis output of the spectrum analysis means by using a predetermined smoothing filter to slow down the fluctuation of the dither level. Bandwidth extension device. 22. The audio band extending apparatus according to claim 21, wherein the smoothing filter controls an attack characteristic and a release characteristic, respectively.