JP2001143384A - Device and method for degital signal processing - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce unnecessary bit assignment in encoding by reducing the quantization noise of a high-frequency band according to the spectrum of an inputted digital signal. SOLUTION: An audio compressing circuit 6 selects, by using a filter coefficient selection part 71, a variable digital low-pass filter 65 having optimum characteristics for a digital audio signal from a filter coefficient storage memory 72 according to the frequency band of the peak of the digital audio signal found by a frequency conversion part 66 and a peak calculation part 70. The digital audio signal passes through the variable digital low-pass filter 65 to have its high-frequency quantization noise reduced and then bit assignment is performed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital signal processing apparatus and a digital signal processing method for processing a digital signal which is noise-shaped with a specific word length, and more particularly to a digital signal used in digital dubbing. The present invention relates to a processing device and a digital signal processing method.

[0002]

2. Description of the Related Art There has been known a noise shaping technique for changing a frequency characteristic of quantization noise in a digital signal in accordance with human auditory characteristics. Noise shaping technology is a technology that reduces high-frequency noise at high sensitivity and lowers high-frequency noise at low sensitivity to reduce quantization noise and improve auditory characteristics. is there.

FIG. 5 is a block diagram illustrating primary noise shaping. FIG. 6 is a graph showing characteristics of the primary noise shaping shown in FIG.

When the circuit shown in FIG. 5 is expressed by an equation, it is as follows.

[0005]

(Equation 1)

[0006] In Formula 1, the x _n input digital signal, y _n is the output digital signal, Q is the tentative quantizer 10
1 shows the output.

[0007] For example, a 20-bit DSP
When implemented using a (Digital Signal Processor), y _n-1 -Q in Equation 1 becomes the lower 4 bits when the upper 16 bits of the output data before one sampling are provisionally quantized by the provisional quantizer 101. . The main quantization by the main quantizer 102 is performed by an external circuit of the DSP.

When the quantization bit length is 16-bit linear PCM (Pulse Code Modulation), the quantization noise level becomes -96 dB (uniform with respect to frequency) (shown by a broken line in FIG. 6). ). On the other hand, the quantization noise level after the primary noise shaping has a differential characteristic as shown in FIG. That is, the quantization noise level in the low frequency band is lower than before the processing, but the quantization noise level in the high frequency band is higher than before the processing.

At present, according to the unified standard, CD (Compact
The word length of the digital audio signal in Disk) is 16
Specified in bits. Among such CDs, a digital audio signal of a CD recorded by using the noise shaping technique as described above is converted to an MD (Mini Disp.
Conventionally, when digital dubbing is performed on a recording medium such as k), a directly input 16-bit digital audio signal is frequency-converted and encoded.

Hereinafter, a general method for digitally dubbing a CD recorded on a CD recorded with reduced quantization noise in a low frequency band to a MD using a noise shaping technique will be briefly described.

An audio signal supplied from the CD to the MD recording / reproducing apparatus as a digital optical input is converted into a digital electric signal, and then clock extraction and data detection corresponding to a sampling frequency of 44.1 kHz are performed. When the sound source is a DAT or CS broadcast, or a personal computer, the sampling frequency is 48 kHz,
Data detection corresponding to 2 kHz is performed, and this data is
Further, by the sampling rate converter, 44.1
It is converted to kHz data.

Next, the data at 44.1 kHz is converted to the ATR
The data is compressed by an AC (Adaptive Transform Acoust-ic Coding) method and subjected to signal processing such as addition of parity by an encoder.

Next, a laser beam is applied to the magneto-optical disk constituting the MD by using an optical pickup, and at the same time, the magnetization corresponding to the digital audio signal processed by the encoder signal is changed by using a magnetic head. Formed on a magneto-optical disk. In this way, digital audio data is recorded at a predetermined position on the magneto-optical disk.

[0014]

However, in the above conventional configuration, frequency conversion and encoding are directly performed on an input digital audio signal. Therefore, the quantization noise in the high frequency band increased by the noise shaping processing is also a target of encoding bit allocation. That is, since the power of the quantization noise in the high frequency band after the noise shaping processing is large, bits are assigned to the noise portion. For this reason, there is a problem that the allocating of the coded bits is wasted.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-described problems, and has been made in view of the above-mentioned problems. It is an object of the present invention to provide a digital signal processing device and a digital signal processing method capable of reducing unnecessary bit allocation at the time.

[0016]

In order to solve the above-mentioned problems, a digital signal processing apparatus according to the present invention is a digital signal processing apparatus used when digitally dubbing a digital signal subjected to noise shaping processing. And a digital low-pass filter that passes a low-frequency band component of which the quantization noise is reduced by the noise shaping process in the digital signal.

The noise shaping process is a process for transforming the frequency characteristics of quantization noise in a digital signal in accordance with human auditory characteristics. Specifically, this is a technique for reducing the quantization noise in a high-sensitivity middle-low frequency band and conversely increasing the quantization noise in a low-sensitivity high frequency band to improve the hearing characteristics. Digital dubbing means re-recording a digital signal recorded on an existing recording medium on another recording medium.

The quantization noise in the low frequency band is reduced by the noise shaping process, and the quantization noise in the high frequency band is increased in the digital signal recorded on a certain recording medium. When re-recording (digital dubbing) such a digital signal on another recording medium, the present invention is configured to pass the input digital signal through a digital low-pass filter. As a result, the digital signal in the high frequency band in which the quantization noise has been increased is removed. That is, the dubbing-side device can remove quantization noise in the high frequency band of the digital signal to be dubbed.

For example, when the configuration of the present invention is provided in an MD recording apparatus, when a digital audio signal is dubbed from a CD recorded using a noise shaping technique to an MD, noise is reduced and sound quality is degraded. Can be prevented.

Further, in the dubbing-side device,
Since the high frequency band component of the digital signal can be removed, even when performing frequency conversion, encoding, or the like, quantization noise in a high frequency band that is not an original signal component is not targeted. In other words, for the quantization noise in the high frequency band increased by the noise shaping processing,
Useless bit allocation can be prevented.

Thus, when digitally dubbing a digital signal subjected to noise shaping processing, quantization noise in a high frequency band can be suitably reduced and unnecessary bit allocation at the time of encoding can be reduced. .

Further, in order to solve the above-mentioned problems, a digital signal processing apparatus according to the present invention comprises a frequency conversion means for converting the digital signal into a spectrum of a plurality of frequency bands and a plurality of digital low-pass filter characteristics. It is preferable to further include a filter characteristic storing unit that performs the filtering and a filter characteristic selecting unit that selects a digital low-pass filter characteristic adapted to the digital signal from the filter characteristic storing unit according to the spectrum.

According to the above arrangement, the input digital signal is converted into a spectrum of a plurality of frequency bands by the frequency conversion means, and according to the spectrum, the filter characteristic is selected from the filter characteristic storage means by the filter characteristic storage means. A digital low-pass filter characteristic adapted to a digital signal can be selected. Since a plurality of digital low-pass filter characteristics are provided in advance in the filter characteristic storage means, an optimal digital low-pass filter can be selected according to the quantization noise characteristic of the input digital signal.

Thus, the quantization noise of the input digital signal can be suitably reduced.

Further, in order to solve the above-mentioned problem, the digital signal processing device of the present invention passes the digital signal through the digital low-pass filter according to whether or not the input digital signal has been subjected to noise shaping processing. It is preferable that it is possible to select whether or not to perform.

If the configuration is such that all the input digital signals pass through the digital low-pass filter, the digital low-pass filter can be passed even when a digital signal that has not been subjected to noise shaping is input. Become. In this case, the cutoff frequency of the digital low-pass filter is 22.05 kHz.
If it is lower than z, the gain in the high frequency band decreases, and the frequency characteristics on the dubbing side do not become flat.

On the other hand, according to the configuration of the present invention as described above, whether or not to pass a digital low-pass filter is selected according to whether or not an input digital signal has been subjected to noise shaping processing. it can. For example, if the input digital signal has not been subjected to noise shaping processing, it is not necessary to pass through the digital low-pass filter, so that it is set so as not to pass through the digital low-pass filter.

Thus, it is possible to perform processing according to the input digital signal.

In order to solve the above-mentioned problems, a digital signal processing method according to the present invention is a digital signal processing method used when digitally dubbing a digital signal subjected to noise shaping processing. The digital low-pass filter, which passes a low-frequency band component in which quantization noise has been reduced by noise shaping processing among the signals, includes a step of passing the digital signal.

The noise shaping process and the digital dubbing have been described above.
The quantization noise in the low frequency band is reduced by the noise shaping processing, and the quantization noise in the high frequency band is increased in the digital signal recorded on a certain recording medium. When re-recording (digital dubbing) such a digital signal on another recording medium, in the method of the present invention, the input digital signal is passed through a digital low-pass filter. As a result, the digital signal in the high frequency band in which the quantization noise has been increased is removed. That is, the dubbing-side device can remove the quantization noise in the high frequency band of the digital signal to be dubbed.

For example, using the method of the present invention, a CD recorded using the noise shaping technique
When dubbing a digital audio signal to an MD,
Noise is reduced, and deterioration of sound quality can be prevented.

Further, in the dubbing-side device,
Since high-frequency band components of a digital signal can be removed, when performing frequency conversion, encoding, or the like on the digital input signal, even quantization noise in a high-frequency band that is not an original signal component is targeted. Disappears. That is, it is possible to prevent useless bit allocation for the quantization noise in the high frequency band increased by the noise shaping processing.

Thus, when digitally dubbing a digital signal subjected to noise shaping processing, quantization noise in a high frequency band can be suitably reduced, and unnecessary bit allocation at the time of encoding can be reduced. .

Further, in order to solve the above-mentioned problem, the digital signal processing method of the present invention converts the digital signal into a spectrum of a plurality of frequency bands before passing the digital signal through a digital low-pass filter. Preferably, the method further includes the step of converting and the step of using the spectrum to select a digital low-pass filter characteristic adapted to the digital signal.

According to the above method, the input digital signal can be converted into a spectrum of a plurality of frequency bands, and a digital low-pass filter characteristic adapted to the digital signal can be selected according to the spectrum. Therefore, an optimal digital low-pass filter can be selected according to the quantization noise characteristics of the input digital signal.

Thus, the quantization noise of the input digital signal can be suitably reduced.

Further, in order to solve the above-mentioned problem, the digital signal processing method of the present invention passes the digital signal through the digital low-pass filter according to whether or not the input digital signal has been subjected to noise shaping processing. It is preferable to be able to select whether or not to perform.

If a method is used in which all the input digital signals pass through a digital low-pass filter, the digital low-pass filter can be passed even when a digital signal that has not been subjected to noise shaping is input. Become. In this case, the cutoff frequency of the digital low-pass filter is 22.05 kHz.
If it is lower than z, the gain in the high frequency band decreases, and the frequency characteristics on the dubbing side do not become flat.

On the other hand, according to the method of the present invention as described above, whether or not to pass a digital low-pass filter is selected according to whether or not the input digital signal has been subjected to noise shaping processing. it can. For example, if the input digital signal has not been subjected to noise shaping, it is not necessary to pass through the digital low-pass filter, so that the digital signal is set so as not to pass through the digital low-pass filter.

Thus, it becomes possible to perform processing according to the input digital signal.

[0041]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below with reference to FIGS.

FIG. 2 is a block diagram showing a configuration of an MD (Mini Disk) recording / reproducing apparatus provided with an audio compression circuit 6 functioning as a digital signal processing apparatus. The following shows a CD (Compact Disk) recorded with noise shaping to reduce quantization noise in the low frequency band.
To perform digital dubbing on an MD, and M
The method of reproducing D will be described with reference to FIG.

The audio signal to be recorded on the MD, which has been subjected to the noise shaping process, is supplied from a terminal 2 as a digital optical input. The digital optical input is converted into a digital electric signal by the photoelectric element 3,
It is supplied to a digital PLL (Phase Locked Loop) circuit 4. The digital PLL circuit 4 extracts a clock from the digital electric signal and detects a digital audio signal corresponding to a CD sampling frequency of 44.1 kHz. Further, the frequency conversion circuit 5
And converted into a digital audio signal of 44.1 kHz.

Next, the 44.1 kHz digital audio signal is converted by the audio compression circuit 6 into ATRAC.
(Adaptive Transform Acoustic Coding). The compressed digital audio signal is subjected to signal processing such as addition of parity by an encoder in an encoder / decoder signal processing circuit 8 via a shock proof memory controller 7 for controlling a shock proof memory 9 described later. The shockproof memory 9 absorbs the difference between the transfer rate of the digital audio signal output from the audio compression circuit 6 and the transfer rate of the digital audio signal input to the encoder / decoder signal processing circuit 8, and reproduces the data. It is provided to interpolate the middle stage of the signal due to disturbance such as vibration at the time and protect the digital audio signal.

Next, the optical pickup 12 irradiates the magneto-optical disk 1 constituting the MD with laser light. At the same time, the magnetic head 11 is moved to the laser beam irradiation position on the magneto-optical disk 1 by the head drive circuit 10. Then, the magnetic head 11 forms magnetization on the magneto-optical disk 1 corresponding to the digital audio signal processed by the encoder / decoder signal processing circuit 8. Thus, digital audio data is recorded at a predetermined position on the magneto-optical disk 1.

The feed motor 13 is connected to the optical pickup 1
2 is moved in a direction perpendicular to the tracks of the magneto-optical disk 1. The spindle motor 14 rotates the magneto-optical disk 1. The driver circuit 15 includes the feed motor 13 and the spindle motor 14.
And a circuit for supplying power to drive a drive device (not shown) for driving the objective lens of the optical pickup 12. Further, the servo circuit 16 feeds back each device driven by the driver circuit 15 so that the operation of causing the light emitted from the optical pickup 12 to follow the target track of the magneto-optical disk 1 is performed accurately. Control.

Next, at the time of MD reproduction, the optical pickup 12 irradiates the magneto-optical disk 1 with laser light, reads the reflected light, and records the RF signal (modulated digital audio signal) recorded on the magneto-optical disk 1 Is detected. The RF signal thus detected is amplified by the RF amplifier 17 and sent to the encoder / decoder signal processing circuit 8,
Decoding processing and the like by a decoder are performed. Next, processes reverse to those at the time of recording, such as a shock proof process by the shock proof memory 9 and the shock proof memory controller 7 and a decompression process of a reproduced signal by the audio decompression circuit 18 are performed. Thereafter, the signal is converted into an analog signal by the DA converter 19, and a reproduction output of the audio signal is derived from the output terminal 20.

All the above processes are centrally managed by the system control microcomputer 21. Note that 22
Is an input device used for input operation to the system control microcomputer 21, and 23 is a power supply.

Next, digital signal processing in the audio compression circuit 6 will be described in detail with reference to FIG.
FIG. 2 is a block diagram showing an encoding process (compression of a digital audio signal) performed by the ATRAC system in the audio compression circuit 6.

The voice compression circuit 6 includes a variable digital low-pass filter (digital low-pass filter) 65.
(In the figure, it is described as a variable digital LPF 65.) A frequency conversion unit (frequency conversion unit) 66, a bit allocation unit 69, a peak calculation unit (filter characteristic selection unit) 70, a filter coefficient selection unit (filter characteristic selection unit) ) 71, and a filter coefficient storage memory (filter characteristic storage means) 72.

In the voice compression circuit 6, the switch 63 is connected to the switch 64a in advance, and the switch 67 is connected to the switch 68a in advance. The audio compression circuit 6 converts the digital audio signal sampled at 44.1 kHz by the frequency conversion circuit 5 into an input terminal 61.
The frequency conversion unit 66 performs the following processing for each frame of a predetermined time.

The frequency conversion section 66 converts the digital audio signal into a QMF (Qu
The image is divided into a plurality (i) of frequency bands (blocks) by an adrature mirror filter. Next, the digital audio signal is divided into MDCs in units of divided blocks.
The frequency is converted (converted into a spectrum signal) by T (Modified Discrete Trance Form) processing. Further, the sum of squares of the converted MDCT coefficients is obtained, and is converted into power data of each of the i blocks.

The peak calculating section 70 sends a signal to the system control microcomputer 21 provided in the MD recording / reproducing apparatus.
A flag is transmitted to notify that the peak frequency band has been determined. Furthermore, coefficient selection information according to the determined peak frequency band is transmitted to the filter coefficient selection unit 71.

The above system control microcomputer 21
When the information is received from the peak calculation unit 70 by the flag, the switch 63 is connected to the switch 64b and the switch 67 is connected to the switch 68b.

Upon receiving the coefficient selection information, the filter coefficient selection section 71 selects a filter coefficient suitable for a peak signal of one frame from a filter coefficient storage memory 72 in which several sets of filter coefficients are stored in advance. I do. Thus, the variable digital low-pass filter 65 is determined. As described above, by changing the filter coefficient, the characteristics (cutoff frequency and the like) of the variable digital low-pass filter 65 are determined.

FIG. 3 is a block diagram showing the configuration of the variable digital low-pass filter 65, and FIG. 4 is a graph showing the frequency characteristics of the variable digital low-pass filter 65.

As shown in FIG. 3, the variable digital low-pass filter 65 has an n-order FIR (Finite Impulse Filter).
Response) type digital filter.

In the figure, Z^-1Is one of the sampling data
Shows Ray. Also, h₀~ H _nIs the filter coefficient
Is shown. Filter output to y_nAnd the filter input
x_nThe above digital filter can be expressed by the following equation.
become that way.

[0059]

(Equation 2)

[0060] h ₀ the to h _n, advance the filter coefficient storing several sets that are prepared in the memory 72 (one set of h ₀ ~
from the coefficient h _n), 1 set selected by the filter coefficient selection unit 71 is used. Filter output y _n
Is calculated by successively accumulating as in Equation 2.

In the frequency characteristic of the variable digital low-pass filter 65 shown in FIG. 4, the reference filter characteristic shows a case where the cutoff frequency is 11 kHz. The filter coefficient is selected according to the input digital signal so that the peak frequency of the input signal <the cutoff frequency. As described above, the cutoff frequency can be changed by changing the filter coefficient.

The processing performed for each predetermined time frame subsequent to the above-described processing will be described again with reference to FIG.

The signal input to the variable digital low-pass filter 65 is
5 is a digital audio signal used in selecting the filter coefficient of No. 5. Therefore, the variable digital low-pass filter 65 used here is optimal for the input digital audio signal. This allows the digital audio signal to pass through the variable digital low-pass filter 65, thereby effectively reducing quantization noise in a high frequency band.

As described above, the digital audio signal from which the quantization noise in the high frequency band has been removed is converted into i number of spectrum data in each frequency band by the frequency conversion unit 66 in the same manner as in the above sequence. .

Further, to the converted spectral data, a bit allocation unit 69 performs bit allocation applying human auditory characteristics, and further calculates a spectrum data length (word length). Is output.

As described above, the audio compression circuit 6 is configured to select the coefficient of the digital low-pass filter according to the input digital audio signal and select the characteristic. Therefore, since the quantization noise in the high frequency band can be effectively removed, digital dubbing can be performed without reducing the gain even for a digital signal having a large quantization noise component in the high frequency band. It becomes possible.

In this embodiment, the characteristics of the n-order FIR digital filter are determined by referring to the peak frequency of the digital audio signal subjected to the first-order noise shaping. It is also applicable to digital audio signals on which high-order noise shaping is performed.

In this case, the order (n) of the FIR digital filter is increased to correspond to a digital audio signal that has been subjected to higher-order noise shaping processing. This is because by increasing the order (n), a filter characteristic having a steeper integration characteristic can be obtained. However, if the order (n) of the FIR type digital filter is dynamically increased during the music, the delay amount differs and becomes noise. Therefore, the user switches the order (n) in a silent portion between music. Will be.

It is also possible to determine the characteristics of the n-th order FIR digital filter by referring to frequencies other than the peak frequency, for example, the local peak (the second and third largest peak) frequencies.

In this case, it is set so that the filter coefficient is selected from the filter coefficient storage memory 72 as shown in (a) and (b) below.

(A) In the case of peak frequency <local peak frequency The filter coefficient is selected so that local peak frequency <cutoff frequency.

(B) When peak frequency> local peak frequency: Filter coefficients are selected so that peak frequency <cutoff frequency.

Further, in the present embodiment, it is possible to adopt a configuration in which the user selects whether or not to pass the digital audio signal through the variable digital low-pass filter 65.

This will be specifically described as follows. When digital dubbing a digital audio signal that has not been subjected to noise shaping, the user himself chooses not to pass through the variable digital low-pass filter 65. This information is sent to the system control microcomputer 21. The system control microcomputer 21 connects the switch 63 to the side 64a and connects the switch 67 to the side 68b. Therefore, the digital audio signal is output without passing through the variable digital low-pass filter 65. Thus, the digital signal processing device according to the present embodiment can be applied to a digital audio signal that has not been subjected to noise shaping.

[0075]

As described above, the digital signal processing apparatus according to the present invention is a digital signal processing apparatus used for digital dubbing of a digital signal subjected to noise shaping processing. And a digital low-pass filter that passes low-frequency band components in which quantization noise has been reduced by noise shaping processing.

Therefore, a digital signal in a high frequency band in which quantization noise is increased is removed. That is, the dubbing-side device can remove the quantization noise in the high frequency band of the digital signal to be dubbed. Furthermore, since the high frequency band component of the digital signal can be removed by the dubbing-side device in this way, when performing frequency conversion, encoding, or the like, even the quantization noise in the high frequency band, which is not the original signal component, is removed. Will not be targeted. That is, bits are not allocated to the quantization noise in the high frequency band increased by the noise shaping processing.

Thus, when digitally dubbing a digital signal subjected to noise shaping processing, quantization noise in a high frequency band can be suitably reduced, and unnecessary bit allocation at the time of encoding can be reduced. This has the effect.

Further, the digital signal processing device according to the present invention comprises: a frequency conversion means for converting the digital signal into a spectrum of a plurality of frequency bands; a filter characteristic storage means for storing a plurality of digital low-pass filter characteristics; It is preferable that the apparatus further comprises a filter characteristic selecting means for selecting a digital low-pass filter characteristic adapted to the digital signal from the filter characteristic storing means according to the spectrum.

Therefore, an optimal digital low-pass filter can be selected according to the quantization noise characteristics of the input digital signal. As a result, the quantization noise of the input digital signal can be advantageously reduced.

Further, the digital signal processing device according to the present invention can select whether or not to pass the digital signal through the digital low-pass filter according to whether or not the input digital signal has been subjected to noise shaping processing. It is preferable to have a configuration.

Therefore, for example, when the input digital signal is not subjected to the noise shaping processing, it can be prevented from passing through the digital low-pass filter. As a result, there is an effect that processing according to the input digital signal can be performed.

A digital signal processing method according to the present invention is a digital signal processing method used when digitally dubbing a digital signal subjected to noise shaping processing. A digital low-pass filter that passes a low-frequency band component in which quantization noise is reduced, including the step of passing the digital signal.

Therefore, a digital signal in a high frequency band in which quantization noise has been increased is removed. That is, the dubbing-side device can remove the quantization noise in the high frequency band of the digital signal to be dubbed. Further, since the high frequency band component of the digital signal can be removed by the dubbing-side device in this way, when performing frequency conversion, encoding, or the like on the digital input signal, a high frequency component that is not an original signal component is obtained. The quantization noise in the frequency band is no longer targeted. That is, bits are not allocated to the quantization noise in the high frequency band increased by the noise shaping processing.

Thus, when digitally dubbing a digital signal that has been subjected to noise shaping processing, quantization noise in a high frequency band can be suitably reduced, and unnecessary bit allocation at the time of encoding can be reduced. This has the effect.

Further, in the digital signal processing method according to the present invention, before the step of passing the digital signal through a digital low-pass filter, the digital signal is
It is preferable that the method further includes a step of converting to a spectrum of a plurality of frequency bands and a step of using the spectrum to select a digital low-pass filter characteristic adapted to the digital signal.

Therefore, an optimal digital low-pass filter can be selected according to the quantization noise characteristics of the input digital signal. As a result, the quantization noise of the input digital signal can be advantageously reduced.

Further, in the digital signal processing method according to the present invention, it is possible to select whether or not to pass the digital signal through the digital low-pass filter according to whether or not the input digital signal has been subjected to noise shaping processing. It is preferable to use a method.

Therefore, for example, when the input digital signal is not subjected to the noise shaping processing, it can be set so as not to pass through the digital low-pass filter. As a result, there is an effect that processing according to the input digital signal can be performed.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an audio compression circuit according to an embodiment of the present invention.

FIG. 2 is a block diagram illustrating a configuration of an MD recording / reproducing apparatus including the audio compression circuit.

FIG. 3 is a block diagram showing a configuration of a variable digital low-pass filter provided in the audio compression circuit.

FIG. 4 is a graph showing an example of a frequency characteristic of the variable digital low-pass filter.

FIG. 5 is a block diagram illustrating a conventional primary noise shaping circuit.

FIG. 6 is a graph showing a conventional primary noise shaping characteristic.

[Explanation of symbols]

6 Voice compression circuit (digital signal processing device) 65 Variable digital low-pass filter (digital low-pass filter) 66 Frequency conversion unit (frequency conversion unit) 70 Peak calculation unit (filter characteristic selection unit) 71 Filter coefficient selection unit (filter characteristic selection unit) 72 Filter coefficient storage memory (filter characteristic storage means)

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) H03H 17/02 681 G10L 7/04 G H03M 7/30 9/18 M

Claims (6)

[Claims] 1. A digital signal processing device used for digitally dubbing a digital signal that has been subjected to noise shaping processing, wherein the digital signal has a low frequency band in which quantization noise has been reduced by noise shaping processing. A digital signal processing device comprising a digital low-pass filter that passes components. 2. A frequency conversion means for converting the digital signal into a spectrum of a plurality of frequency bands, a filter characteristic storage means for storing a plurality of digital low-pass filter characteristics, and the filter characteristic storage means according to the spectrum. 2. The digital signal processing apparatus according to claim 1, further comprising: filter characteristic selecting means for selecting a digital low-pass filter characteristic adapted to the digital signal. 3. The method according to claim 1, wherein whether the input digital signal has been subjected to noise shaping processing can select whether or not to pass the digital signal through the digital low-pass filter. 2. The digital signal processing device according to claim 1. 4. A digital signal processing method used for digitally dubbing a digital signal that has been subjected to noise shaping processing, wherein the digital signal has a low frequency band in which quantization noise has been reduced by noise shaping processing. A digital signal processing method, comprising: passing a digital signal through a digital low-pass filter that passes components. 5. A step of converting the digital signal into a spectrum of a plurality of frequency bands prior to the step of passing the digital signal through a digital low-pass filter; and using the spectrum to adapt the digital signal to the digital signal. 5. The digital signal processing method according to claim 4, further comprising selecting a low-pass filter characteristic. 6. The apparatus according to claim 4, wherein whether or not the input digital signal is passed through said digital low-pass filter can be selected according to whether or not the digital signal has been subjected to noise shaping processing. 2. The digital signal processing method according to 1.