JP3272438B2 - Signal processing system and processing method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a signal processing technique for audio and video signals, and more particularly to coding and decoding of signals using spectral control of quantization noise in consideration of human hearing and the like. Etc. and its applications.

[0002]

2. Description of the Related Art In digital signal processing, how to control and reduce quantization noise is an essential issue for performing high-fidelity reproduction. In general, when the number of quantization steps is sufficiently large or when accurate dither is superimposed / subtracted, the quantization noise becomes white noise uncorrelated with the input signal. In this case, the power of the quantization noise in the band of 1/2 of the sampling frequency is given by
A Δ ^2/12. Therefore, if the sampling frequency is increased, the quantization noise in the signal band can be relatively reduced, and a dynamic range of 100 dB or more can be theoretically secured by 1-bit quantization. However, since the sampling frequency is actually several GHz, it is actually difficult to implement hardware. Therefore, attempts have been made to secure a sufficient dynamic range at a realistic sampling frequency by concentrating the distribution of quantization noise in a high frequency range.

FIG. 26 shows various A / D / D / A conversion methods, and FIG. 27 shows a frequency spectrum thereof. FIG. 26 (a)
Is the most basic PCM transmission and processing method. FIG.
(b) is a method called oversampling method.
Sampling is performed at a frequency higher than usual, and thinning is performed by a digital low-pass filter that also serves as a band limit, thereby achieving matching with ordinary digital signals.

FIG. 22 (c) shows a system called ΣΔ modulation in which a quantizer is provided in a feedback loop to give a higher-range differential characteristic to quantization noise. FIG. 28 shows a configuration of first to third order ΣΔ modulation. In Figure 28, when the input signal by the quantizer assuming that quantization noise N _q uncorrelated occurs, the output of the n-order ΣΔ modulation, Y = X + (1- z -1) n · N _q , and the quantization noise appearing in the output signal is obtained by differentiating _Nq to the nth order. FIG. 29 shows the input / output relationship of such ΣΔ modulation, the spectrum of quantization noise, and the quantization noise power in the bands of 0 to fs / 2 and 0 to fs / 32 when the sampling frequency is fs. FIG. 30 shows the spectral distribution of quantization noise of ０〜Δ modulation of the 0th to 4th orders.
30A is an enlarged view of the band of 0 to fs, and FIG. 30B is an enlarged view of the band of 0 to fs / 4. As can be seen from FIGS.
Increasing the order of Δ modulation increases the total power of the quantization noise,
At a frequency 1/6 of the sampling frequency, the quantization noise power in the lower frequency band decreases rapidly. FIG.
Shows the power ratio SN _q of a sine wave and the quantization noise of the maximum amplitude of the band of the 0~f in 1-4 order 1 bit ΣΔ modulation. From this figure, for example, in the first-order ΣΔ modulation, a sampling frequency of 112 MHz is required in order to obtain a dynamic range of 100 dB in a band of 20 kHz.
Next, it can be seen that a sampling frequency of 1.1 MHz is sufficient.

[0005]

The above-mentioned A / D, D / A
In the conversion method, as can be seen from FIG. 26, since the band is limited by the low-pass filter, the signal band to be recorded and transmitted has a certain limitation. Particularly, in the ΣΔ modulation method, complicated signal processing is required on both the A / D side and the D / A side.

An object of the present invention is to provide a signal processing system capable of performing highly accurate processing with a simple configuration by controlling the spectral distribution of quantization noise.

[0007]

According to a first aspect of the present invention, in a signal processing apparatus for recording a signal on a recording medium by 1-bit high-speed sampling, a first analog input signal is modulated by modulating a quantum signal. 1 that focuses on the high frequency noise
A first 1-bit signal processing unit that generates a first 1-bit digital signal sampled at a high speed, and a 1-bit high-speed sampler that concentrates quantization noise on a high frequency band by modulating a second analog input signal A second 1-bit signal processing unit for generating a converted second 1-bit digital signal, and a converter for converting the first 1-bit digital signal into a first multi-bit digital signal at a normal sampling frequency An analog-to-digital converter for converting the second analog input signal into a second multi-bit digital signal; a cross-fade period before the cross-fade period; In which the first multi-bit digital signal and the second multi-bit digital signal are cross-fade. Ri to generate a recording signal, after the cross-fade period and a recording signal generating unit to recording signals the second 1-bit digital signal.

According to a second aspect of the present invention, the signal is set to 1
In the method of recording on a recording medium by bit high-speed sampling, a 1-bit high-speed sampled 1 bit in which quantization noise is concentrated in a high band by modulating a first analog input signal before a cross-fade period. A bit digital signal is generated and recorded on a recording medium. During the cross-fade period, the first digital signal is converted into a first multi-bit digital signal at a normal sampling frequency, and a second analog input signal is generated. A signal is converted into a second multi-bit digital signal, and a cross-fade process is performed between the first multi-bit digital signal and the second multi-bit digital signal to generate a recording signal, which is recorded on the recording medium. After the end of the cross-fade period, the quantization noise is modulated by modulating the second analog input signal. To generate a second digital signal 1-bit high-speed sampling concentrated on high frequency band is recorded on the recording medium. According to the signal recording device or method configured as described above, first, the first analog signal is sampled at high speed, converted into a first 1-bit digital signal, and recorded on a recording medium. Next, when entering the crossfade period,
The first 1-bit digital signal is converted to a first multi-bit digital signal, and the second analog signal is converted to a second multi-bit digital signal at a normal sampling frequency. And recorded on a recording medium. When the crossfade period ends, the second analog signal is sampled at high speed, converted into a second 1-bit digital signal, and recorded on a recording medium.

[0009] According to a third aspect of the invention, in the signal processing system, modulates the analog signal, and encoding means for outputting the 1-bit high-speed sampling signals centered quantization noise to a higher frequency, the 1-bit high-speed Positive or negative of the same frequency as the sampled signal
Signal generation means for generating a gate signal which is a repetition signal of the above, an EX-OR circuit for performing an exclusive OR operation of the 1-bit high-speed sampling signal and the gate signal,
And-a signal processing means for performing predetermined signal processing including recording on a recording medium or transmission via a transmission line with respect to the output signal of the OR circuit. According to the system configured as described above, the 1-bit high-speed sampled signal is converted to the EX-
The result of the exclusive OR operation in the OR circuit is signal-processed.

According to a fourth aspect of the present invention, in the signal processing system according to the third aspect, another EX-OR circuit that performs an exclusive OR operation of an output signal of the signal processing unit and the gate signal. And a decoding means for inputting the output signal of the another EX-OR circuit and restoring the original analog signal. This allows
The signal calculated by the EX-OR circuit is decoded.

According to a fifth aspect of the present invention, in the signal processing method, a first step of modulating an analog signal and outputting a 1-bit high-speed sampling signal in which quantization noise is concentrated in a high frequency band;
Positive / negative repetition of the same frequency as the 1-bit high-speed sampling signal
A second step of generating a gate signal as a return signal, a third step of performing an exclusive OR operation of the 1-bit high-speed sampling signal and the gate signal, and a step of performing an exclusive-OR operation on the output signal of the EX-OR circuit. And performing a certain signal processing including recording on a recording medium or transmission via a transmission path. According to the system configured as described above, a signal obtained by performing an exclusive OR operation on a signal obtained by sampling one bit at a high speed by an EX-OR circuit is processed. According to a sixth aspect of the present invention, in the signal processing system according to the fifth aspect, a fifth step of performing an exclusive OR operation of an output signal of the signal processing unit and the gate signal, and using a low-pass filter. And a sixth step of inputting the signal generated in the fifth step and restoring the original analog signal. Thus, the signal calculated by the EX-OR circuit is decoded.

[0012]

According to the present invention, a signal is sampled at a higher frequency than usual, so that a faster signal processing than usual is possible.

Further, by controlling the spectral distribution of the quantization noise, highly accurate signal processing can be performed with a simple configuration. Further, since the output of the ΣΔ modulation or the like is processed as it is as a 1-bit digital signal, complicated processing such as a digital filter after the ΣΔ modulation or an inverse quantizer (D / A converter) at the latter stage of the processing system becomes unnecessary. . Further, in this method, since band limitation by a low-pass filter or the like is not performed on either the encoding side or the decoding side, processing of a high-frequency signal can be performed as compared with a general ΣΔ method or the like.

[0014]

[Embodiment] Basic principle of the present invention As described above, generally, as shown in FIG.
While performing band limiting using a digital filter after Δ modulation, matching with a desired sampling frequency is achieved by decimation processing (decimation), but the above-described processing is not performed on the signal after ΣΔ modulation, A simple processing system that treats the digital signal as it is can be considered.出力 Δ modulation output,
In particular, the output of 1-bit ΣΔ modulation is a digital signal, but the spectrum of the original signal exists as it is in the signal band. Therefore, at the time of decoding, a D / A converter (inverse quantizer) is not required, and an analog signal can be obtained only by providing a filter if necessary. Here, Σ
Although only the case of Δ modulation will be described, the same processing can be performed by a method other than this, as long as the encoded output is only one bit, only the spectrum of the quantization noise changes.

FIG. 1 is a block diagram of a high-speed sampling 1-bit system according to the present invention. Since the ΣΔ modulation output is processed as it is as a 1-bit digital signal, there is no need for a digital filter after ΣΔ modulation or an inverse quantizer (D / A converter) at the subsequent stage of the processing system as compared with FIG. Become. FIG. 2 shows a frequency spectrum according to this method. Since the quantization noise is concentrated in the high band, the quantization noise power in the signal band is relatively reduced. In this method, the encoding side,
Since no band limitation is performed by a low-pass filter or the like on any of the decoding sides, a general ΣΔ method (FIG. 26)
(c)) Higher frequency signal processing can be performed. Also,
In this method, since the sampling frequency is high, there is also an advantage that high-speed processing can be performed. FIG. 3 shows the basic configuration of the 1-bit high-speed sampling method according to the present invention. This system is roughly divided into an emphasis unit 1, a ΣΔ modulation unit 2, a signal processing unit 3, a decoding unit 4, and a de-emphasis unit 5. The emphasis unit 1 and the de-emphasis unit 5 are generally provided to reduce the amount of high frequency components in an audio signal and to suppress quantization noise. Depending on the characteristics of the signal to be processed, it can be omitted. The ΣΔ modulation section 2 performs 部 Δ modulation of about the second to seventh order, and has a specific configuration as shown in FIG. The signal processing unit 3 is a unit that performs recording and transmission of an original signal, and corresponds to DAT, CD, satellite communication, and the like. The decoding unit 4 is a part corresponding to a normal D / A converter. For example, as shown in FIG. 5B, a flip-flop can be used as a 1-bit inverse quantizer.

The signal waveforms (S1 to S1) of each part in FIG.
4) and the frequency characteristics are shown in FIG. As described above, this system can be realized with a simple configuration including several operational amplifiers and flip-flops, and does not require a complicated multi-bit quantizer or digital filter. FIG. 7 shows the frequency characteristics in the case of 1.536 MHz, 1 bit by the fourth-order ΣΔ modulation, and FIG. 8 shows the waveforms when a 10 kHz square wave is recorded / reproduced to / from the DAT by the present method and the normal method. From FIG. 7, it can be seen that this system can process high-frequency signals up to about 100 kHz. Also, in FIG. 8, the waveform of the present method reproduces the original waveform almost as it is, whereas the output of the normal method is limited to the fundamental wave because the band is limited.

Next, processing of a plurality of channel signals according to the present method will be described. FIG. 9 shows a signal processing system for four channels. The input analog signal of each channel is 1 bit in 1 bit A / D sections 2a to 2d.
Are converted to digital signals Sa to Sd. These signals are time-division multiplexed by the multiplexer 6 and sent to the processing system 3 as a 4-channel signal Sm. Signal Sa ~
FIG. 10 shows Sd and Sm. To multiplex a four-channel signal with a frequency f, the frequency of the control signal St is 4 · f
is necessary. In the processing system 3, processing such as recording / reproduction of CDs and DATs and transmission are performed. The signal obtained from the processing system 3 is divided into signals of respective channels by a demultiplexer 7, and each of the signals is divided into 1-bit A / D units 4a to 4d.
Is decoded into an analog signal.

The basic configuration of the 1-bit A / D unit 2 is shown in FIG.
(a), which is composed of an integrating circuit 27, a quantizer 28 and a delay circuit 29. Although FIG. 11 (a) is based on the first-order ΣΔ modulation, it is possible to adopt a configuration of a plurality of orders as shown in FIG. 11 (b). It is determined by controlling.

FIGS. 12 and 14 show A for performing high-order ΣΔ modulation.
4 shows a circuit configuration example of a / D section. FIGS. 13 and 15 show frequency characteristics of quantization noise by the circuits of FIGS. As can be seen from these, the spectrum distribution of the quantization noise can be controlled by arbitrarily setting the configuration of the A / D section.

In the above description, the case where the spectrum distribution of the quantization noise is controlled by ΣΔ modulation has been described, but another modulation method may be used. Next, 1bitA
FIG. 16 shows the configuration of the / D section 4. The 1-bit digital signal distributed to each channel is inverted by the gate 41, and output as a signal having positive and negative amplitudes around 0V by the differential amplifier at the subsequent stage. Also,
This circuit also serves as a de-emphasis circuit, and a desired de-emphasis characteristic (see FIG. 4) can be obtained by changing a resistance value or a capacitor value.

In FIG. 9, the signals from the processing system 3 are distributed to the signals of the respective channels by the demultiplexer 7 and then decoded. A signal can be obtained, and it also operates as a mixing circuit. When recording a signal on a crossfade, a fade-in / fade-out tape, or the like, cross-fade processing is performed when a signal is switched from one signal to another signal. FIG. 17 shows a cross-fade processing method in this method.
This will be described based on FIGS. In this method, as shown in FIGS. 17 and 18, switching is performed by switching the input of the signal of CH1 and the signal of CH2. That is, from the state of inputting only the signal of CH1 (Step 1), the input of the signal of CH2 is gradually started (Step 2), and the frequency is further increased (Steps 3 to 6). Only the user inputs (step 7).

In performing the above-described cross-fade processing, it is necessary to control the quantization noise distribution. In general, when the original signal is sampled at the sampling frequency fs, a spectrum of its harmonic appears, so that it is 信号 of the sampling frequency.
Noise is generated at the boundary of (fs / 2). Now, when a 1-bit signal having a quantization noise distribution as shown in FIG. 19A is sampled at a sampling frequency of fs, the frequency fs /
Although aliasing noise occurs around 2, the spectrum overlaps with the spectrum of the original signal, so that the aliasing component of the quantization noise does not overlap the signal band (low frequency band). The same result is obtained by sampling at a frequency that is a positive multiple of the sampling frequency. On the other hand, when the same signal is sampled at a frequency other than a positive multiple of the sampling frequency, such as 1/4 or 1/3 of the sampling frequency, as shown in FIGS. 19 (b) and (c), Since the aliasing noise component overlaps with the signal band, a large noise is included.

In the above-mentioned cross-fade processing, processing for switching signals at regular time intervals is performed as shown in FIG. 18. Such processing is equivalent to sampling the original signal at a different sampling frequency. Become. For example, focusing on the signal A (CH1) in FIG. 18, there is no problem because the signal A is sampled at twice the frequency (2 fs) in step 4, but in steps 5 and 6, the signal A is 1/3 and 1/4 frequency (fs / 3, fs /
Since this is equivalent to the sampling performed in 4), FIG.
As shown in (c), the aliasing noise component overlaps the signal band. In order to prevent such a problem, it is necessary to control the distribution of the quantization noise so that the aliasing noise component does not overlap the signal band during the period in which the cross-fade processing is performed. FIG. 20 (a) controls the quantization noise distribution so that the aliasing noise does not overlap the signal band when the signal is switched at a frequency of 1/4 of the original signal and FIG. 20 (b). The spectrum obtained is shown. Such control of the quantization noise distribution can be realized by changing the configuration of the 1-bit A / D unit shown in FIG. Therefore, if A / D conversion is performed so as to have a quantization noise distribution corresponding to the switching frequency at each cross-fade stage (each step in FIG. 18), the cross-fade is not affected by aliasing noise. Processing becomes possible.

The crossfade processing can also be performed by the following method. FIG. 21 shows the configuration. DAT
The digital recorder is configured to perform a cross-fade process using a signal of 16 bits and 48 kHz. Therefore, in the cross-fade section, a signal of 16 bits, 48 k
Cross-fade is performed with a signal of Hz. In FIG.
Consider a case where the signal of A is switched to the signal of B. A /
The D converter 20 has a 1-bit, 768 k
Hz signals A and B are input. SW1 is normally connected to the terminal a side (when only the signal A is used), but is switched to the terminal b side during the cross-fade period.
At the time of cross-fading, the signals A and B are converted from 1-bit and 768 kHz signals to 16-bit and 48 kHz signals, respectively, by the bit converter 21 and cross-fade processing is performed. The signal after the cross-fade is reconverted into a 1-bit, 768 kHz signal by the bit converter 23, and is recorded on the tape via the modulator 24. The features of the 1-bit signal processing according to this method are that the quantization noise is low in the low frequency range and that signal processing up to the high frequency range is possible. However, during the cross-fade period, the two signals are mixed. Since the state is not a state in which the degree of noise or the like should be considered, the above method can be used as a simpler method.

The case of the cross-fade processing has been described above. Other than this, for example, start recording on tape
At the end, etc., processing such as fade-in / fade-out is performed in order to prevent instantaneous noise. In this case, one of the signals to be mixed is set to a zero signal, and a method similar to cross-fade is performed. Can be processed. Noise suppression at the time of burst code error In a general DAT or the like, digital zero data is output when a burst-like code error occurs during reproduction or when a portion where no signal is recorded is reproduced. Muting processing is performed. In DAT and the like, since a 2's complement system is used for encoding, digital zero data outputs "000-000" or "111-111". Therefore, when recording / reproducing a 1-bit signal according to this method, since each bit is data, it becomes a positive or negative full-scale signal, and is reproduced as noise. Therefore, a circuit for preventing such noise is shown in FIG. Fig. 22 (a)
The signals from the terminals L and R are both 1-bit signals according to the present method, are mixed into two-channel signals by the multiplexer 51, and are input to the EX-OR circuit 52. The signal S3 from the gate signal generator 56 is input to another input terminal of the EX-OR circuit 52. Signal S
Reference numeral 3 denotes a positive / negative repetitive signal having a constant frequency (for example, the same frequency as the signal S1), and the output from the EX-OR circuit 52 is an inverted signal of the signal S2. This signal passes through the processing system and is input to the EX-OR circuit 54 on the reproduction side. EX-
The other input terminal of the OR circuit 54 includes a gate signal generator 5
6, the signal S3 from the same signal generator is input, and the signal S4 is inverted again.
Is the same signal as the signal S2. The signal S5 is distributed to two channels by the demultiplexer 55. Here, when muting occurs due to a code error or the like in the reproduction system, the signal S4 becomes a full-scale signal as described above.
Since the output of No. 4 becomes a positive / negative repetition signal by the signal S3, the output of large noise can be prevented. FIG. 22 (b)
An EX-OR circuit is inserted for each channel, and has the same function as that of FIG.

The above description is mainly for recording a 1-bit signal.
・ The aim was to prevent noise in playback, but the same idea
To be actively applied to signal transmission systems such as wireless communication
Prevents wiretapping of communication contents (a kind of scramble processing)
Can also be performed. That is, from the gate signal generator
Pseudo-random sequence such as M-sequence
If the output signal of the EX-OR circuit is a kind of encrypted signal
It can be. Therefore, the same signal is used on the receiving side.
And decrypt the encryption to get the original signal.
You. However, the signal here must be a known signal that has been determined.
Is necessary. Application to systems requiring multiplication or convolution In digital signal processing, digital signals, especially complex
Multiplication of several bit signals requires complicated operations,
Calculation time is also required. High-speed sampling 1-bit method
Can be processed at high sampling frequency for high-speed processing
In that the calculation can be performed between 1-bit signals.
However, multiplication with analog signals (convolution operation)
In other words, using the features of this method,
High-precision signal processing is possible. An example of the application is shown below.
You.

FIG. 23 shows a case where the 1-bit signal according to this method is used in an active noise control system. In the figure, 34 is set as a point of interest, and noise at this point is controlled using an additional sound source. The signal from the noise source 30 is
And is input to the filter 32. The signal from the filter 32 is output from the additional sound source 33.
Here, the transfer characteristic from the noise source 30 to the microphone 31 is H ₁
(f), the characteristic of the filter 32 is H ₂ (f), the transfer characteristic from the additional sound source 33 to the point of interest 34 is H ₃ (f), and the noise source 30
The transfer characteristic of the target point 34 from the H When (f), H (f) = -H 1 (f) · H 2 (f) · H 3 (f) As is true filter characteristic H ₂ (f) Is controlled, the noise at the point of interest can be canceled. Filter 3
The output Y (n) of No. ₂ is represented by Y (n) = (X (n) · H ₂ (nk), where N is the coefficient of the filter. Here, X (n) is an analog signal, which requires a filter characteristic H ₂ and a convolution operation for N filter coefficients. Therefore, if the convolution operation is performed using the filter characteristic H ₂ as a 1-bit digital signal of the present method, the quantization noise, which is a feature of the present method, is concentrated in a high frequency range.
A signal with less noise in the signal band can be obtained. As a method of updating the coefficient of the filter 32, there is a method such as LMS.

FIG. 24 shows a case where the 1-bit signal according to the present method is applied to cancel howling of a hands-free telephone or the like. The sound input from the microphone 41 is transmitted to the amplifier 4
3 and is output from the speaker 44. Howling can be canceled by performing a convolution operation on the characteristics (transfer characteristics of the room, etc.) from the speaker 44 to the microphone 41 by the calculator 45 and subtracting this from the input. is there. Here, the signals x and y are both analog signals, and the same high-precision processing as described above can be performed by using the 1-bit signal of the present system as the convolution signal.

FIG. 25 shows a method of adding reverberation to music information and the like using a 1-bit signal of the present system. Conventionally, as shown in FIG. 25 (a), an audio signal is AD-converted, a DSP processor performs convolution calculation of a specific reverberation characteristic (such as an impulse response in a concert hall or the like), and further DA-converted to add reverberation. Was getting a signal. According to this method,
As shown in FIG. 25 (b), for an analog sound signal,
If a 1-bit reverberation characteristic signal is calculated by switching a switch or the like, processing such as AD conversion or convolution calculation of a plurality of bits becomes unnecessary.

In the above description, the audio signal processing has been described. However, the present invention can be applied to an image signal by controlling the quantization noise distribution while paying attention to human visual characteristics. is there.

[0031]

As described above, according to the present invention, by controlling the spectral distribution of quantization noise, highly accurate signal processing can be performed with a simple configuration. In addition, since the output of ΣΔ modulation or the like is processed as it is as a 1-bit digital signal, high-speed processing becomes possible, and complex filters such as a digital filter after ΣΔ modulation and an inverse quantizer (D / A converter) at the subsequent stage of the processing system are used. Unnecessary processing becomes unnecessary. Further, since the signal is a 1-bit signal, multiplication with an analog signal can be performed with a simple configuration. Further, in this method, since band limitation by a low-pass filter or the like is not performed on either the encoding side or the decoding side, processing of a high band signal can be performed as compared with a general ΣΔ method or the like.

[Brief description of the drawings]

FIG. 1 is a block diagram of a high-speed sampling 1-bit system according to the present invention.

FIG. 2 is a diagram showing a frequency spectrum of a high-speed sampling 1-bit system according to the present invention.

FIG. 3 is a basic configuration diagram of a high-speed sampling 1-bit system according to the present invention.

FIG. 4 is a diagram showing an emphasis / de-emphasis characteristic of a high-speed sampling 1-bit system according to the present invention.

FIG. 5 is a configuration diagram of a ΣΔ modulation section / decoding section.

6 is a diagram showing signal waveforms and frequency characteristics in FIG.

FIG. 7 is a diagram showing frequency characteristics of a high-speed sampling 1-bit system according to the present invention.

FIG. 8 is a signal waveform diagram of a high-speed sampling 1-bit system according to the present invention.

FIG. 9 is a configuration diagram of a high-speed sampling 1-bit multi-channel signal processing system according to the present invention.

FIG. 10 is an explanatory diagram of a high-speed sampling 1-bit multi-channel signal processing system according to the present invention.

FIG. 11 is a configuration diagram of a 1-bit A / D unit.

FIG. 12 is another configuration diagram of the 1-bit A / D unit.

FIG. 13 is a diagram showing a quantization noise spectrum based on the 1-bit A / D shown in FIG.

FIG. 14 is another configuration diagram of the 1-bit A / D unit.

FIG. 15 is a diagram showing a quantization noise spectrum based on the 1-bit A / D shown in FIG. 14;

FIG. 16 is a configuration diagram of a 1-bit D / A unit.

FIG. 17 is an explanatory diagram of the cross-fade processing by one method.

FIG. 18 is an explanatory diagram of the operation of the crossfade processing system according to the present method.

FIG. 19 is a diagram illustrating a frequency spectrum of a crossfade processing system according to the present scheme.

FIG. 20 is another diagram showing a frequency spectrum of the crossfade processing system according to the present scheme.

FIG. 21 is another configuration diagram of the crossfade processing system according to the present method.

FIG. 22 is a configuration diagram of a noise suppression circuit for a burst code error.

FIG. 23 is a diagram illustrating an application example of the present method.

FIG. 24 is a diagram showing another application example of the present method.

FIG. 25 is a diagram showing another application example of the present method.

FIG. 26 is a diagram showing various A / D / D / A conversion methods.

FIG. 27 is a diagram showing a frequency spectrum of each A / D / D / A conversion method.

FIG. 28 is a configuration diagram of a ΣΔ modulation method.

FIG. 29 is a diagram illustrating characteristics of 次 Δ modulation of each order.

FIG. 30 is a diagram illustrating a quantization noise spectrum of each order ΣΔ modulation scheme.

FIG. 31 is a diagram illustrating noise power characteristics of ΣΔ modulation of each order.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Emphasis circuit 2 ... Sigma-delta modulator 3 ... Processing system 4 ... Flip-flop 5 ... D / A conversion part 6, 51 ... Multiplexer 7, 55 ... Demultiplexer 52, 54 ... EX-OR circuit 56, 57 ... Gate signal generation circuit

Claims (6)

(57) [Claims] 1. A signal processing apparatus for recording a signal on a recording medium by 1-bit high-speed sampling, wherein 1-bit high-speed sampling in which quantization noise is concentrated in a high band by modulating a first analog input signal. The first one done
A first 1-bit signal processing unit that generates a bit digital signal; and a 1-bit high-speed sampled second signal that modulates a second analog input signal to concentrate quantization noise in a high frequency range.
A second 1-bit signal processing unit that generates a bit digital signal; a conversion unit that converts the first 1-bit digital signal into a first multi-bit digital signal at a normal sampling frequency; An analog-to-digital converter for converting an analog input signal into a second multi-bit digital signal; a first 1-bit digital signal as a recording signal before a cross-fade period; A recording signal generating unit that generates a recording signal by performing a cross-fade process on a multi-bit digital signal and the second multi-bit digital signal, and that uses the second 1-bit digital signal as a recording signal after a cross-fade period And a signal processing device. 2. A method of recording a signal on a recording medium by 1-bit high-speed sampling, wherein a quantization noise is concentrated on a high frequency band by modulating a first analog input signal before a cross-fade period.
A first 1-bit digital signal sampled at a high speed is generated and recorded on a recording medium. During the cross-fade period, the first digital signal is converted into a first multi-bit digital signal at a normal sampling frequency. And a second analog input signal is converted into a second multi-bit digital signal, and the first multi-bit digital signal and the second multi-bit digital signal are subjected to cross-fade processing. Is generated and recorded on the recording medium. After the end of the cross-fade period, the second analog input signal is modulated to concentrate the quantization noise in a high frequency region and the 1-bit high-speed sampled second signal is sampled. A signal recording method, wherein a digital signal is generated and recorded on the recording medium. 3. An encoding means for modulating an analog signal and outputting a 1-bit high-speed sampled signal in which quantization noise is concentrated in a high band, and a positive / negative repetition having the same frequency as the 1-bit high-speed sampled signal.
Signal generation means for generating a gate signal as a return signal ; an EX-OR circuit for performing an exclusive OR operation of the 1-bit high-speed sampling signal and the gate signal; and an output signal of the EX-OR circuit Signal processing means for performing a predetermined signal processing including recording on a recording medium or transmission via a transmission path. 4. An EX-OR circuit for performing an exclusive OR operation on an output signal of the signal processing means and the gate signal, and a low-pass filter, and an output signal of the other EX-OR circuit is provided. 4. The signal processing system according to claim 3, further comprising: decoding means for inputting and restoring an original analog signal. 5. A first step of modulating an analog signal and outputting a 1-bit high-speed sampled signal in which quantization noise is concentrated in a high frequency band, and performing a positive / negative cycle having the same frequency as the 1-bit high-speed sampled signal.
A second step of generating a gate signal as a return signal; a third step of performing an exclusive OR operation of the 1-bit high-speed sampling signal and the gate signal; And performing a predetermined signal processing including recording on a recording medium or transmission via a transmission path. 6. A fifth step of performing an exclusive OR operation on the output signal of the signal processing means and the gate signal, and using a low-pass filter to input the signal generated in the fifth step to obtain an original signal. The signal processing system according to claim 5, further comprising: a sixth step of restoring an analog signal.