JPH04113723A - Error correction system - Google Patents

Abstract

PURPOSE:To attain the correction close to a substantial correct voice signal data by comparing voice signal data outputted from an estimate value arithmetic circuit and a low-order word replacement circuit and selecting the voice signal data from the low-order word replacement circuit when the data are different from each other. CONSTITUTION:A voice signal data D1 from an error correction circuit 11 is inputted to an estimate value arithmetic circuit 13 and a low-order word replacement circuit 14 in the order of sample points t1, t2, t3. On the other hand, error flags EU, EL of each voice signal data are inputted to an error signal generating circuit 12. The estimate value arithmetic circuit 13 receives the error signal E1 and calculates the mean value of a voice signal data at sample points t1, t3 without an error flag and outputs the result as a voice signal data D2. The selective circuit 15 receives voice signal data D2, D3 respectively from the estimate value arithmetic circuit 13 and the low-order word replacement circuit 14 sequentially and compares them and selectively outputs any of the voice signal data D2 and D3 in response to the result of comparison.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an error correction method, and more particularly to an error correction method that divides quantized audio signal data into upper and lower words and performs correction for each word.

[Conventional technology]

FIG. 3 is a block diagram showing an example of a conventional error correction method. The error correction circuit 21 divides the quantized audio signal data D into upper words and lower words, performs error correction on each word, and outputs it as audio signal data D4. Word error 7 lag Eυ.

Output EL. The error signal generation circuit 22 receives error flags E[J, El of the upper and lower words, and outputs an error signal E when at least one word cannot be error corrected. When the estimated value calculation circuit 23 receives the error signal E, the estimated value calculation circuit 23 outputs the audio signal data D! Error correction is performed by calculating an estimated value based on audio signal data that does not include errors before and after , and outputting the estimated data. Estimation methods in this case include a method of obtaining an average value and a method of approximating an n-dimensional equation.

[Problem to be solved by the invention]

In the conventional error correction method described above, when an error cannot be corrected in at least one of the upper word and the lower word, an estimated value is calculated based on audio signal data that does not include previous and subsequent errors, and the estimated value is replaced with the estimated data. However, even when the effect of the error is relatively small, such as when there is an error only in the lower word, it is processed in the same way as an error in the upper word, where the effect of the error is large. There is a problem in that the correction is carried out.

An object of the present invention is to provide an error correction method that can perform correction close to the original audio signal data when there is an error only in the lower word.

[Means to solve the problem]

The error correction method of the present invention is an error correction method for quantized audio signal data. an error correction circuit that outputs an error flag for each word when the error flag of each word is disabled; and an error correction circuit that outputs a first error signal when at least one of the error flags of each word is received; an error signal generation circuit that outputs a second error signal when only an error flag is received; and an error signal generation circuit that outputs a second error signal when only an error flag is received; an estimated value calculation circuit that calculates an estimated value using audio signal data that does not include the audio signal data and outputs the estimated audio signal data; A lower word replacement circuit that replaces and outputs a lower word with a set word and audio signal data output by the estimated value calculation circuit and the lower word replacement circuit are compared to perform the estimated value calculation circuit and the lower word replacement circuit. and a selection circuit that selects and outputs one of the audio signal data output by the circuit. Further, the selection circuit compares the data value of the upper word of the audio signal data outputted by the estimated value calculation circuit and the data value of the upper word of the audio signal data outputted by the lower word replacement circuit, and If the difference is 2 units or more, the audio signal data output by the lower word replacement circuit is selected and output, and if the difference is less than 2 units, the audio signal data output by the estimated value calculation circuit is selected and output. It may be configured to select and output.

〔Example〕

Next, the present invention will be explained with reference to the drawings.

FIG. 1 is a block diagram showing one embodiment of the present invention. The error correction circuit 11 divides the quantized audio signal data D into upper words and lower words, performs error correction on each word, and outputs the audio signal data D. The error flags Eυ and EL of the upper word and lower word are output.

The error signal generation circuit 12 receives the error flags Eu and Et of the upper and lower words, and outputs an error signal E1 when at least one of the words cannot be corrected, and outputs an error signal when only the lower word cannot be corrected. Output E2. When the estimated value calculation circuit 13 receives the error signal E1, it calculates an average value based on the audio signal data that does not include the error before and after the audio signal data D1, and replaces the audio signal data D□ with the estimated data D2. and output it. When receiving the error signal E2, the lower word replacement circuit 14 replaces the lower word with a preset word and outputs it as audio signal data D3. In this case, the center value of the data value range of the lower word is set. The selection circuit 15 is
The audio signal data D2 and D3 are received and compared, and one of the audio signal data D2 and D3 is selected according to the comparison result and output as corrected audio signal data Do.

Next, the operation will be explained using an example of audio signal data shown in FIG. Here, the audio signal data D□ is 16 bits and is expressed in hexadecimal. Further, the original correct audio signal data is shown by a solid line, the audio signal data after the correction according to the present invention is shown by a broken line, and the case where the average value correction is performed is shown by a dashed line. Now, sample point t
There is an uncorrectable error in the lower word of the voice signal data No. 2, and the upper word without an uncorrectable error is selected (6B). It is also assumed that the audio signal data at sample points t1 and t3 have no uncorrectable errors, and (670
4) and (6BO4'). The audio signal data D1 from the error correction circuit 11 is sampled at sample points 11, 12, 1.
．． are input to the estimated value calculation circuit 13 and the lower word replacement circuit 14 in this order. On the other hand, the error flags Eu, Et, of each audio signal data are input to the error signal generation circuit 12.

Now, at the sample point t2, the error signal generation circuit 1
2 detects that only the lower word cannot be corrected, and sends error signals E1 and E2 to the estimated value calculation circuit 13 and the lower word replacement circuit 14, respectively. The estimated value calculation circuit 13 receives the error signal E1 and selects sample points 1. ．． 1. Sound audio signal data (670
4) Calculate the average value (6904) of (6B04) and use it as the audio signal D2 (6704), (6
904) and (6BO4) are output in this order. on the other hand,
The lower word replacement circuit 14 is set with the center value (80) of the range of data values of the lower word crimping <00) to (FF), and upon receiving the error signal E2, the lower word replacement circuit 14 replaces the lower word (
00) is replaced with (80) to obtain audio signal data D3 (68
80), (6704), (6880)
, (6BO4) are output in this order. The selection circuit 15 is
Audio signal data D2. D3 is sequentially received and compared, and depending on the comparison result, one of the audio signal data D2 and D3 is selected and output. In this case, the data value of the upper word of the audio signal data D2 from the estimated value calculation circuit 13 is compared with the data value of the upper word of the audio signal data D3 from the lower word replacement circuit 14, and the difference is less than 2 units. If the difference is 2 units or more, the audio signal data D3 is selected. First, compare (6704) and (6704),
Since the upper words match, the audio signal data D2
(6704) and output. Next, (6904) and (6880) are compared. The upper word (69) of (6900) and the data value and the upper word (690) of (6B80)
Since the difference from the data value of B) is 2 units, audio signal data D 3 (6880) is selected. Similarly (6B
O4) and (6BO4) are compared to obtain audio signal data D2.
Select (6BO4) and output. Therefore, the audio signal data D output by the selection circuit 16.

As shown by the dashed line, (6704), (6B80
), (6B04), and (6704), (69
04) and (6BO4), it is possible to perform correction closer to the original correct audio signal data shown by the solid line.

〔Effect of the invention〕

As explained above, according to the present invention, there is provided an estimated value calculation circuit that calculates an estimated value for audio signal data in which an error flag exists in at least one of the upper and lower words using audio signal data that does not include any preceding or following errors. For audio signal data in which the error flag is present only in the lower word, a lower word replacement circuit is provided to replace the lower word with a preset word, and the audio signal data output by the estimated value calculation circuit and the lower word replacement circuit is By comparing and selecting the audio signal data from the lower word replacement circuit if the data values are far apart, it is possible to distinguish between cases where there is an error in the upper word and cases where there is no error flag in the upper word and the effect of the error is small. Since the same processing is not performed, corrections that are far removed from the original audio signal data are not performed, and corrections that are close to the original correct audio signal data can be performed.

Circuit, 14...Lower word replacement circuit, 15.-
...Selection circuit, D, D 1. D2, D3
, Do --- audio signal data, Eu...
Upper word error flag, EL...-lower word error flag, E, E.

E2...-Error signal.

Claims (1)

[Claims] 1. In an error correction method for quantized audio signal data, the audio signal data is divided into upper and lower words, error correction is performed on each word, and the result is output.
an error correction circuit that outputs an error flag for each word when error correction is impossible; and an error correction circuit that outputs a first error signal when at least one of the error flags of each word is received; an error signal generation circuit that outputs a second error signal when only a word error flag is received; and an error signal generation circuit that outputs a second error signal when only a word error flag is received; an estimated value calculation circuit that calculates an estimated value using audio signal data that does not include an error and outputs the estimated audio signal data; and an estimated value calculation circuit that calculates an estimated value using audio signal data that does not include an error, and for audio signal data that the error correction circuit outputs when receiving the second error signal. A lower word replacement circuit that replaces and outputs a lower word with a word set in advance, and audio signal data output by the estimated value calculation circuit and the lower word replacement circuit, and compares the audio signal data output by the estimated value calculation circuit and the lower word replacement circuit. An error correction method comprising: a selection circuit that selects and outputs one of the audio signal data output by the word replacement circuit. 2. The selection circuit compares the data value of the upper word of the audio signal data outputted by the estimated value calculation circuit and the data value of the upper word of the audio signal data outputted by the lower word replacement circuit, and determines the value of these data values. If the difference is 2 units or more, the audio signal data output by the lower word permutation circuit is selected and output, and if the difference is less than 2 units, the audio signal data output by the estimated value calculation circuit is selected. 2. The error correction method according to claim 1, wherein the error correction method is outputted as follows.