JPH0690175A - Digital music signal compressing method - Google Patents

Abstract

PURPOSE:To reduce circuit scale and to improve musical sound compression by linearly predicting a sampling value yn of recorded information at the time (n) based on a specific expression. CONSTITUTION:The digital sampling value of the recorded information at the time (n) is linearly predicted based on an expression I. this expression I, yn-1 is the predictive value of the time n-1, yn-2 is the predictive value of the time n-2, and a1, a2 and a3 are arbitrary normal numbers. A differential circuit 11 calculates difference dn between musical signals Xn and Yn inputted from a signal repair circuit 2, and it is encoded by a shift register 12 and a judge circuit 13. At such a time, the initial value of quantizing width DELTAn can be any arbitrary number excepting '0' and the quantizing width DELTA can be updated by using a judge circuit 16, multiplier 17 and coefficient memory 18. Thus, the decibel of the music signal at the low level of a high frequency component is increased, the circuit scale is reduced, and the musical sound compression is improved corresponding to the width DELTAn.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for compressing a digital musical tone signal, which is used as a measure for skipping sound in a portable compact disc (hereinafter referred to as CD) player.

[0002]

2. Description of the Related Art Conventionally, as a measure for skipping sound in a portable CD player, there is a method in which a reproduction signal is once stored in a buffer memory to recover the skipping sound, and the signal is reproduced while maintaining the continuity of the signal. In this case, detection of sound skips, detection of sound skips, cueing of connection points, and repair of sound skips must be performed within the time for reproducing the tone signal stored in the buffer memory. The music signal must be compressed in order to make the delay. Since such a skip sound reproduction mode during compression is provided separately from the normal reproduction mode, a certain degree of sound quality deterioration is allowed.

The tone signal recorded on the CD is 16-bit pulse code modulation (PCM) data sampled at 44.1 kHz per channel (ch), and its transfer rate is 705.
It will be 6 kbps / ch. And for compressed signals of tone signals, Sony's ATRAC method (140 kbps)
/ Ch) or Philips PASC system (192 kb
ps / ch) has been proposed, and in these methods, a high-quality reproduced sound can be obtained by dividing a musical tone signal into several bands and allocating bits to each band using the auditory characteristics. Was there.

In addition to the above-mentioned systems, ATC (adaptive conversion coding) system, ADPCM (adaptive differential coding) system and the like are generally known as musical tone signal coding systems. In particular, the latter ADPCM method is the simplest encoding method, and it has been known that if a certain amount of information (5 bits) or more is assigned, reproduction of a normal music source is comparable to the original sound.

FIG. 3 shows a conventional ADPCM encoding circuit.
A tone signal compression apparatus for updating the quantization width in
It was done. In the figure, 11 is 16 bits to be input.
Digital tone signal x_n And the predicted signal y_n Difference d
_n A differential circuit for obtaining the
Signal L_n Shift register and decision circuit for making
This signal L_n To quantize the signal q_n Shift register to make
, 15 is this signal q _n Predicted signal y based on_{n + 1} Seeking
Adder 16 is the signal L_n About case classification
The decision circuit 17 is a function stored in the coefficient memory 18.
It is a multiplier that updates the quantization width by multiplication using a number.
It

In FIG. 3, the prediction method is the first-order difference prediction, which is expressed by the following equation 2.

[0007]

[Equation 2]

[0008]

However, the above equation 2
In the conventional first-order difference prediction method as shown in FIG. 1, even if a certain amount of information is assigned to the quantization, there is a drawback that the quantization noise becomes large due to a circuit limitation such as calculation accuracy. There has been a situation where quantization noise is conspicuous for a musical sound having a large level of low frequency components.

The present invention has been made in view of the above problems of the prior art, and it is an object of the present invention to reduce the circuit scale and realize good compression of a musical tone signal.

[0010]

SUMMARY OF THE INVENTION The present invention is suitable for applying recorded information.
Compress by encoding with the differential encoding method
A method for signal prediction in this adaptive differential encoding method.
During the measurement process, a digital sample of the recorded information at time n
Pulling value y_n The prediction value y at time n-1 _n-1 as well as
Predicted value y at time n-2_n-2 A straight line based on
It is a prediction.

[0011]

[Equation 3]

In Equation 3, a ₁ = 2 and a ₂ = −1, especially in the case of equal interval sampling.

[0013]

Function: The tone signal, especially the sampling frequency of CD is 4
It is as high as 4.1 kHz, and in such a case, the correlation between adjacent samples becomes very large. AD as shown in Equation 3 above
By using linear prediction as a prediction method in PCM (adaptive differential encoding method), a good decoding is possible even if the scale of hardware is reduced and the calculation accuracy of the circuit is reduced in the compression of a digital tone signal such as a CD. Better sound quality can be obtained.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A digital tone signal compression method of the present invention will be described below with reference to C.
An embodiment applied to the sound skip prevention device D will be described in detail with reference to the drawings.

FIG. 1 is a block diagram of a CD sound skip prevention device. In the figure, 1 is a CD, 2 is an optical pickup for reading a subcode having information such as a musical tone signal, an address, a track number, and an index recorded on the CD 1 at a double speed (sampling frequency 88.2 kHz / ch),
Reference numeral 3 is a signal restoration circuit that allows only the musical tone signal to pass through at a double speed as it is if there is no skip in the signal read by the optical pickup 2. Reference numeral 4 is a signal passed through the signal restoration circuit 3 (16-bit musical tone signal). ) Is encoded at double speed and 5
An encoder for producing a bit-encoded signal, 5 is a buffer memory composed of a 1M RAM for temporarily storing the signal encoded by the encoder 4 at double speed, and 6 is a buffer memory 5 Normal speed (44.1 kHz / c
a decoder for decoding the coded signal sent in h),
Reference numeral 7 is a D / A converter for converting the digital signal of the decoder 6 into an analog signal, and 8 is a low-pass filter for cutting a predetermined frequency from the analog output of the D / A converter 7 (cutoff frequency is 20 kHz). , 9 are speakers for outputting reproduced sound by the output of the low-pass filter 8. Reference numeral 10 denotes a system control circuit, which controls each component within the range surrounded by the broken line in the figure.

In the apparatus having the above construction, when the buffer memory 5 is in a memory full state, the control circuit 10 operates to stop the signal reading from the optical pickup 2. At this time, the control circuit 10 identifies the signal to be read next based on the subcode of the signal last fetched by the signal restoration circuit, and the buffer memory 5
As soon as the storage area is empty, signal reading is resumed, so that signal continuity is maintained.

Further, according to the above-mentioned structure, a stereo musical tone signal of 2.3 seconds at maximum can be temporarily stored in the buffer memory 5, so that even if skipping occurs in reading from the optical pickup 2, Continuous reproduction is possible if the coded signal remains in the buffer memory 5, and within the range of the time margin of 2.3 seconds, the control circuit 10 and the signal restoration circuit 3 detect skipping and skipping. It can detect the location, find the connection point, and repair the skipped sound.

The digital tone signal compression method of the present invention is applied to the encoder 4 and the decoder 6. The outline of the encoder 4 and the decoder 6 will be described below.

(1) Encoder FIG. 3 is a diagram showing a circuit configuration of the encoder. The same components as those of FIG. 3 described in the section of the prior art are designated by the same reference numerals and detailed description thereof will be omitted. To do. In the figure, the shift register 1
Y _n is obtained by linear prediction from the predicted values y _n-1 and y _n-2 stored in 9. In the case of the present embodiment, a ₁ = 2, a ₂ = −1, and a ₃ = 0 in the above mathematical expression ₃ . Therefore,
y _n is calculated by the shift register 19 and the adder 20 as in the following Expression 4. At this time, the double multiplication is replaced with a shift operation.

[0020]

[Equation 4]

In FIG. 2, the difference d _n between the 16-bit PCM digital tone signal x _n input from the signal restoration circuit 2 and the predicted signal y _n is calculated by the following equation 5 using the difference circuit 1.

[0022]

[Equation 5]

Next, the shift register 12 and the determination circuit 13
Then, it is encoded by the following equation 6 to obtain L _n .

[0024]

[Equation 6]

In the above formula 6, [] is a Gauss symbol and represents the maximum integer that does not exceed the number. In the case of 5-bit encoding as in this embodiment, the code L _n is -16 to 1
Since it is an integer up to 5, the division at the time of encoding is determined by the determination circuit 13 by determining whether the sign is positive or negative and dividing the case by a maximum of 15 times.
Will be done in. The initial value of the quantization width Δn may be any number other than 0, and is 8 in this embodiment.

Further, using the above L _n and Δn, the shift register 14 is used to quantize by the following equation 7, and q _n is obtained.

[0027]

[Equation 7]

Finally, the adder 15 is used to obtain the predicted quantized signal Y _{n according} to the equation (8).

[0029]

[Equation 8]

On the other hand, the quantization width Δn is updated by the following equation 9.

[0031]

[Equation 9]

In the equation (9), M (L _n ) is a function according to the following Table 1, and the decision circuit 16, the multiplier 17 and the coefficient memory 18 are used to update the quantization width.

[0033]

[Table 1]

(2) Decoder The decoder obtains the reproduced voice x _n 'from the ADPCM code L _n by the following equation 10.

[0035]

[Equation 10]

Also, the updating of the quantization width here is the same as in the above-mentioned equation 9.

[0037]

According to the present invention, as described above, the method of performing the prediction of the quantization for the compression of the musical tone signal by using the linear prediction makes it possible to reproduce the musical tone signal with a low level of high frequency components such as classical music. There was an improvement of 5-6 dB.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment configuration diagram in which the present invention is applied to a CD sound skipping device.

FIG. 2 is a block diagram showing an example of an implementation configuration of the encoder of FIG. 1.

FIG. 3 is a block diagram showing a conventional configuration example of an encoder corresponding to FIG.

[Explanation of symbols]

 1 Compact Disc 2 Optical Pickup 3 Signal Restoring Circuit 4 Encoder 5 Buffer Memory 6 Decoder 7 DA Converter 8 Low Pass Filter 9 Speaker 10 System Control Circuit 11 Difference Circuit 12, 14, 19 Shift Register 13, 16 Judgment Circuit 15 , 20 adder

Claims (1)

[Claims] 1. A method for compressing recording information by encoding it by an adaptive differential encoding method, which is a digital sampling value of the recording information at time n in a signal prediction process in this adaptive differential encoding method. predicted value prediction of time n-1 of y _n y _n-1 and time n-2 of the predicted value y _n-2
A method for compressing a digital musical tone signal, characterized by performing a straight line prediction based on Equation 1 using [Equation 1]