JPH10135829A - Analog/digital converting error reducing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce a conversion error by estimating the correct quantizing level of a code generating a conversion error and outputting the quantizing level as the quantizing level of a code generating the conversion error so as to detect a conversion error at the time of analog/digital conversion from a digital signal. SOLUTION: A digital audio signal prepares a histogram by a level distribution detecting means 1. An error code detecting means 2 detects codes whose appearing frequency are extremely many or less at each code. A level estimating means 3 estimates a quantizing level which is estimated to be correct in an error code from an inputted digital audio signal. A conversion error judging means 4 compares the estimated quantizing level and inputted digital audio signal with each other. A converting error correcting means 5, at the time of judging a conversion error to be large, outputs the digital audio signal by lowering the quantizing level of the digital audio signal by 1LBS (a lowest order bit) and when the conversion error is small, outputs as it is.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an analog / digital conversion error reducing apparatus for reducing a conversion error during digital / analog conversion of a digital signal.

[0002]

2. Description of the Related Art At the beginning of recording of a music signal by PCM (Pulse Code Modulation) modulation, an analog / digital converter (A / D converter: Analog / Digital Converter) for converting an analog signal into a digital signal is used. , 13-bit or 14-bit successive approximation type A / D converter has been used. The successive approximation type A / D converter can reduce a conversion error by adjusting a resistance value inside the A / D converter or adjusting a bit weight by an external circuit. At the time of the / D conversion, the A / D conversion error could be reduced.

[0003]

However, in the past, many digital audio signals digitally recorded by using a low-bit A / D converter are low-bit A / D converters.
It has many conversion errors due to poor A / D conversion accuracy by the D converter. Since the characteristics of the A / D converter change with time, even if the A / D converter was used at the time of recording, it was not possible to detect a conversion error from the A / D converter and correct the digital audio signal. . That is, it is difficult to detect and reduce the conversion error from the digital audio signal recorded with the conversion error at the time of A / D conversion.

The present invention provides an analog / digital conversion error reducing device capable of detecting a conversion error at the time of analog / digital conversion from a digital signal having a conversion error by an analog / digital converter and reducing the conversion error. It is intended to be.

[0005]

According to the present invention, a plurality of samples of a digital signal converted to a code by an analog / digital converter are provided for each code representing a quantized amplitude level. Level distribution detecting means for detecting the frequency of occurrence, and error code detection for detecting a code having a large difference in the frequency of occurrence of an adjacent code as a code having a conversion error from the frequency of occurrence for each code detected by the level distribution detecting means Means, a level estimating means for estimating a correct quantization level of a code having a conversion error detected by an error code detecting means, and quantizing a code having a conversion error by converting the quantization level estimated by the level estimating means. And a conversion error correction means for outputting a level.

The present invention according to claim 2 is based on claim 1.
In the analog / digital conversion error reducing device described above,
The level estimating means estimates the quantization level of the code in which the conversion error has occurred, from the average of the quantization levels of the code in which the conversion error has occurred and the code adjacent in time.

The present invention described in claim 3 is based on claim 1.
In the analog / digital conversion error reducing device described above,
The level estimating means obtains an average estimated value from a quantization level of a code immediately before and after the code in which the conversion error has occurred, and obtains the average estimated value and a plurality of temporally estimated codes in which the conversion error has occurred. A first extrapolated estimated value is obtained from a quantization level with the previous code, and a second extrapolated value is calculated from the average estimated value and the quantization level with a code that is temporally plural codes later than the code in which the conversion error has occurred. An extrapolated estimate is obtained, an average extrapolated value is obtained from the first extrapolated value and the second extrapolated value, and a conversion error occurs from the average of the average estimated value and the extrapolated average estimated value. It is characterized by estimating the quantization level of the code.

According to the present invention, a code having a conversion error caused by an analog / digital converter is detected from the frequency of occurrence of each code representing a quantized amplitude level of an input digital signal, and the correct code in the code is detected. A quantization level is estimated, and the estimated quantization level is output as a quantization level of a code having a conversion error. As a result, an analog / digital conversion error of a digital signal can be reduced.

[0009]

FIG. 1 is a schematic diagram showing a schematic configuration of an embodiment of an analog / digital conversion error reducing apparatus according to the present invention. As shown in FIG. 1, the analog / digital (A / D) conversion error reducing device includes a level distribution detecting unit 1, an error code detecting unit 2, a level estimating unit 3, a conversion error determining unit 4, and a conversion error correcting unit 5. I have it. Digital audio signals are analog audio signals
/ D conversion and digital audio tape (DAT: Digi
tal Audio Tape) and compact discs (CD: Compac)
t Disk) or the like, and are read from the recording medium.

The level distribution detecting means 1 detects the frequency of appearance of each code representing the quantized amplitude level of the digital audio signal. The digital audio signal is a signal converted from an analog signal to a digital signal by an A / D converter, and the amplitude level is quantized and represented as a code. A histogram is obtained by detecting the frequency of occurrence of the code in the digital audio signal.

Here, the histogram of the digital audio signal will be described. FIG. 2 is a schematic diagram illustrating an example of a histogram of a digital audio signal. (A)
Shows the case where the A / D conversion error is small, and FIG.
This is the case where the / D conversion error is large. As shown in the histogram of FIG. 2A, the frequency of appearance of the digital audio signal having a small A / D conversion error has a substantially symmetric distribution in the positive and negative directions with the code 0 as the center. When the code is 0, the appearance frequency is high, and as the code increases in the positive direction, and as the code decreases in the negative direction,
The appearance frequency gradually decreases. Such a feature of the histogram is unique to each input audio signal.

However, as shown in FIG.
In the case of a histogram of a digital audio signal in which a conversion error has occurred, a code whose appearance frequency sharply increases or a code sharply decreases with respect to an adjacent code is generated.

The level distribution detecting means 1 uses the code itself representing the quantized amplitude level as an address, and an address memory (not shown) storing a value corresponding to the address, and a histogram for creating a histogram. Histogram memory (not shown) that accumulates the appearance frequency of each code
Is provided. The level distribution detecting means 1 accesses a histogram memory from a code of the digital audio signal based on the code, and reads a value corresponding to an address stored in the address memory. One is added to this data, and the added value is written to the same address. This process is repeated for a plurality of samples.
As a result, the cumulative number of times for each code, that is, the frequency of appearance is obtained and stored in the histogram memory.

The number of samples for which the process of creating a histogram is performed is not all of the digital audio signals for which an A / D conversion error is to be detected, but is a signal converted into a digital audio signal by the same A / D converter. , A portion of the digital audio signal, eg, samples for the first few seconds. The greater the number of samples, the better the accuracy of the histogram. For example, if a sample of a period of 2,500,000 samples in which a relatively high level portion is continuous is used, a favorable histogram distribution can be obtained. This histogram can be realized by a digital counter circuit or software.

The error code detecting means 2 detects a code having an A / D conversion error from the histogram detected by the level distribution detecting means 1. FIG.
FIG. 3 is a schematic diagram showing a code having a large D conversion error. A / D
In a digital audio signal in which a conversion error has occurred, a code in which the frequency of occurrence of an adjacent code sharply increases, or
A low code occurs. The error code detection means 2
The code whose appearance frequency changes abruptly is detected from the histogram. As shown in FIG. 3, the appearance frequency is -35 LS
There is a difference of about 20,000 samples between B and -36 LSB. Originally, the sample to be converted to -36 LSB is
About 10,000 samples, -35L due to A / D conversion error
It has been converted to SB. In such a case, the error code detecting means 2 detects a code having a level of -36 LSB as a code having a conversion error.

The level estimating means 3 calculates a quantization level which is estimated to be correct from a code temporally adjacent to the code having a large difference in appearance frequency detected by the error code detecting means 2. It is an estimate. There are two methods for estimating the quantization level. The first estimating method is a method of estimating the quantization level from the values of codes before and after the code temporally adjacent to the code for estimating the quantization level. In addition, the second estimation method includes a method of estimating a value estimated from a quantization level of a code temporally preceding a code for estimating a quantization level and a value of a code temporally subsequent to the code estimating a quantization level. This is a method of obtaining a value estimated from the quantization level and estimating the value from an average of the two values.

The first estimating method will be specifically described. FIG. 4 is a schematic diagram showing a first level estimation method in the analog / digital conversion error reduction device of the present invention. In FIG. 4, for example, it is assumed that there is a difference of about 20,000 samples between the appearance frequencies of −35 LSB and −36 LSB. That is, it is assumed that a sample to be converted to −36 LSB is converted to −35 LSB due to an A / D conversion error.

The quantization level Sn of the detected time T _n is-
It is a 35LSB, the quantization level S _n-1 of the time T _n-1 -38LSB, the quantization level S _{n + 1} of the time T _{n + 1} -34
The LSB, and the average estimated value S ′ _n of the quantization level at the time T _n estimated from the quantization levels of the preceding and succeeding samples can be obtained by the following equation. S ′ _n = (S _n−1 + S _{n + 1} ) / 2 (1)

The second level estimation method will be specifically described. FIG. 5 is a schematic diagram showing a second level estimation method in the analog / digital conversion error reduction device of the present invention. In FIG. 5, for example, the frequency of occurrence is −35 LS.
Assume that there is a difference of about 20,000 samples between B and -36 LSB. That is, it is assumed that a sample to be converted to −36 LSB is converted to −35 LSB due to an A / D conversion error.

The quantization level S _n-1 of the time T _n-1 is -38L
SB, the quantization level S _{n + 1} of the time T _{n + 1} is -34LSB, quantization levels S _n of the detected time T _n is -35L
SB. First, the average estimated value S _'n of quantization levels of the sample at the time was estimated from the quantization level T _n temporally before and after the sample is calculated by the following equation. S ′ _n = (S _n−1 + S _{n + 1} ) / 2 (2)

Next, from S _n−2 , S _n−1 and S ′ _n , the time T
The first extrapolated estimated value Z _n1 of the quantization level at _n−1 is obtained by the following equation. _{Z n1 = S n1 - {(} S 'n -S n-2) / 2} (3)

Further, from S ' _n , S _{n + 2} , S _{n + 1} , the time T
_A second extrapolated estimated value Z _n2 of the quantization level at _{n + 1} is obtained by the following equation. _{Z n2 = S n + 1 -} {(S n2 -S 'n) / 2} (4)

Furthermore, the first outer挿推value Z _n1 from the second outer挿推value Z _n2, the average outer挿推value Z _n of estimate at time T _n, calculated by the following equation. _{Z n = (Z n1 + Z} n2) / 2 (5)

[0024] Then, 'the average outer挿推value Z _n and _n, the estimated value S' average estimate S a _'n, calculated by the following equation. S ″ _n = (S ′ _n + Z _n ) / 2 (6)

The conversion error judging means 4 comprises the level estimating means 3
Is compared with the value of the code having a large difference in appearance frequency, and whether the difference between the two values is greater than a predetermined value,
It is determined whether or not to correct the value of a code having a large difference in appearance frequency depending on whether the code is small.

The estimated value S according to the first level estimation method described above _'n or the estimated value S according to the second-level estimation method' from _'n, the following equation, whether sample A / D conversion error is large judge. S ′ _n <S _n −１ ／ LSB (7) S ″ _n <S _n −１ ／ LSB S ′ _n ≧ S _n −１ ／ LSB (8) S ″ _n ≧ S _n −１ ／ LSB

The conversion error correction means 5 corrects the input digital audio signal and outputs it when the conversion error determination means 4 determines that the code has a large A / D conversion error. That is, the value of the estimated value S ′ (S ″) obtained from the equations (7) and (8) is equal to S _n −1/2.
If the value is smaller than the LSB value, it is determined that the A / D conversion error is large, and the value of the quantization level Sn of the sample is set to 1
Output with LSB lowered. Also, the value of the estimated value S ′ (S ″) obtained from the equations (7) and (8) is Sn− ／ LSB.
If the value is equal to or greater than the value, it is determined that the A / D conversion error is small, and the input digital audio signal is output as it is.

The processing of the A / D conversion error reducing device having the above configuration will be described. The digital audio signal is input to the level distribution detecting means 1, and the frequency of appearance is detected for each code to create a histogram.

The error code detecting means 2 refers to the created histogram to detect a code having an extremely high appearance frequency or a code having an extremely low appearance frequency for each code.

Based on the error code detected by the error code detecting means 2, the level estimating means 3 estimates a quantization level estimated to be correct in the error code from the input digital audio signal.

The conversion error judging means 4 comprises the level estimating means 3
Is compared with the input digital audio signal to determine whether the conversion error is large.

The conversion error correction means 5 converts the code determined to have a large conversion error by the conversion error
Set the quantization level of the input digital audio signal to 1
The digital audio signal that is output with the LSB lowered and that is determined to have a small conversion error is output as it is.

By the above processing, a conversion error generated at the time of A / D conversion of a digital audio signal can be detected and reduced. For example, even if the characteristics of the used A / D converter are not known, A / D conversion can be performed. The distortion of music software having a D conversion error can be reduced, and the sound quality can be improved.

[0034]

According to the present invention, in an analog / digital conversion error reducing device, a conversion error at the time of analog / digital conversion is detected from a digital signal having a conversion error by an analog / digital converter, and the conversion error is reduced. be able to.

[Brief description of the drawings]

FIG. 1 is a schematic diagram showing a schematic configuration of an embodiment of an analog / digital conversion error reduction device of the present invention.

FIG. 2 is a schematic diagram illustrating an example of a histogram of a digital audio signal. (A) shows the case where the A / D conversion error is small, and (b) shows the case where the A / D conversion error is large.

FIG. 3 is a schematic diagram showing a code having a large A / D conversion error.

FIG. 4 is a schematic diagram showing a first level estimation method in the analog / digital conversion error reduction device of the present invention.

FIG. 5 is a schematic diagram showing a second level estimation method in the analog / digital conversion error reduction device of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Level distribution detection means 2 ... Error code detection means 3 ... Level estimation means 4 ... Conversion error determination means 5 ... Conversion error correction means

Claims (3)

[Claims] 1. A level distribution detecting means for detecting an appearance frequency of each code representing a quantized amplitude level of a plurality of samples of a digital signal converted into a code by an analog / digital converter, and detecting the level distribution. Error code detecting means for detecting a code having a large difference in the frequency of occurrence of an adjacent code from the frequency of occurrence of each code detected by the means as a code having a conversion error, and the conversion detected by the error code detecting means. Level estimating means for estimating a correct quantization level of the code in which the error has occurred, and conversion error correcting means for outputting the quantization level estimated by the level estimating means as the quantization level of the code in which the conversion error has occurred. An analog / digital conversion error reduction device, comprising: 2. The analog / digital conversion error reducing device according to claim 1, wherein said level estimating means calculates the average of quantization levels of a code in which the conversion error has occurred and a code adjacent immediately before and after the code. An analog / digital converter for estimating a quantization level of a code having a conversion error.
Digital conversion error reduction device. 3. An analog / digital conversion error reducing apparatus according to claim 1, wherein said level estimating means calculates an average estimated value from a quantization level of a code in which the conversion error has occurred and a code immediately preceding and succeeding in time. , And a first extrapolated estimated value is obtained from the quantized level of the average estimated value and the code having a plurality of codes before the conversion error.
A second extrapolated estimated value is obtained from the quantization level between the average estimated value and the code having a plurality of codes temporally subsequent to the code in which the conversion error has occurred, and the first extrapolated estimated value and the second extrapolated value are calculated. An analog extrapolated value is obtained from the extrapolated estimated value, and the quantization level of the code having the conversion error is estimated from the average of the average estimated value and the extrapolated average estimated value. / Digital conversion error reduction device.