KR940006820B1 - Error processing method and circuit of decoder - Google Patents

Abstract

The circuit differently handles the error of the decoder according to the selected filter in order to eliminate the output signal saturation caused by the errors. The circuit comprises: data from CD-ROM; an error processing unit (10) handling the error position interpolation according to the selected filter number; an adding unit (20A) outputting the revision data plus the output data of the error processing unit (10); a saturation protection unit (30); a filter (50A) carrying out filtering for the output of the saturation protection unit (30); and an adder (20B) providing the input signal of the adder (20A).

Description

Decoder Error Handling Circuit

1 is a block diagram of a general decoder.

2 is a filter coefficient table.

3 is an error processing block diagram of the decoder of the present invention.

4 is a signal flow diagram for error cancellation of the present invention.

FIG. 5 is a diagram illustrating one embodiment of the saturation prevention unit in FIG.

6 is another embodiment of the saturation prevention part in FIG.

7 is another embodiment of the saturation prevention in FIG.

8 is attenuation characteristic diagram of FIG.

9 is an error processing explanatory diagram when the filter number is "0".

* Explanation of symbols for main parts of the drawings

10: error processing unit 20A, 20B: adder

30: Saturation prevention part 40A, 40B: Delay

50A, 50B: Filter

The present invention relates to error processing of a CD-I ADPCM decoder. In particular, when an error occurs irregularly in data, the error is concealed, and an error is accumulated to prevent the output of the decoder from being saturated in one direction. It relates to an error processing circuit of a CD-I ADPCM decoder suitably.

FIG. 1 is a block diagram of a typical CD-I ADPCM decoder, which is comprised of adders 1A and 1B, one sample delayers 2A and 2B, and filters 3A and 3B as shown therein. Referring to Figure 2 as follows.

The input data {X (n)} outputted through the adder 1A is passed through one sample delayers 2A and 2B, which are delayed by one sample and delayed by the delayer 2A. The data delayed by two samples through is selected through four types of filter coefficients according to the state of the input signal as shown in FIG. 2 through the filters 3A and 3B, respectively, and these are added to each other in the adder 1B.

The output signal of the adder 1B is supplied to the adder 1A again and added to the currently input data {X (n)}, and this value is output data {Y (n)}, which is the next D / A. Supplied to the transducer.

However, in the conventional decoder, when an error occurs in the input data, the unobtrusive sound is output as it is, and furthermore, the output signal is saturated due to the error, and a signal that is greatly distorted at the output is output.

The present invention has been made in order to solve the conventional problems, and to prevent the saturation of the error processing according to the selected filter will be described in detail with reference to the accompanying drawings.

3 is a block diagram of a CD-I ADPCM decoder to which the error processing means of the present invention is applied, and as shown therein, ADPCM receives data (D), an error flag (EF), and a filter coefficient (FN) from a CD-ROM. An error processor 10 for differently interpolating error positions according to the filter number FN selected in the encoding process, an adder 20A for adding correction data to the output data of the error processor 10, and outputting the correction data; After receiving the output data of the adder 20A and calculating an average value for a predetermined time, the sample is delayed by one sample through a saturation preventing unit 30 and a delay unit 40A to negatively feedback the output signal. The saturation prevention unit delayed by two samples through a filter 50A for performing filtering on the output signal of the saturation prevention unit 30 by a filter coefficient K ₀ , and delayers 40A and 40B. Filter by the filter coefficient (K ₁ ) The filter 50B to be performed and the adder 20B which adds the output signal of the said filter 50A, 50B, and supplies it as the input signal of the one side of the adder 20A, are attached to this invention. A detailed description with reference to FIGS. 4 to 6 is as follows.

The ADPCM used for CD-I has a different aspect from the general ADPCM, that is, to limit various parameters to implement a cheap decoder. One of them limits the parameters of the filter used in the predictor to four. Shown in FIG. 1 is a predictive filter used in the predictor, and shown in FIG. 2 is a filter value. Different filter numbers FN are selected and encoded according to input signals.

Referring to FIG. 4, the operation of the error processing unit 10 will be described. When data D, an error flag EF and a filter number FN are supplied from the CD-ROM to the error processing unit 10, an error occurs. By reducing the noise caused by the data unnecessarily, the interpolation of the error position is handled differently according to the filter coefficients K ₀ and K ₁ selected in the ADPCM encoding, which is illustrated in FIG.

If the selected filter number FN is 0, since the PCM data is encoded as it is, not the DIFFERENCE signal, that is, the input signal is encoded as it is in FIG. Since it is not accepted, the previous data is stored and the value is output as shown in FIG.

However, if the filter number FN is not 0, the data is compressed through the predictor filter, indicating the slope or the amount of change in the slope. If an error occurs in such data, it can be seen that the difference between the original signal and the original signal is reduced by correcting the change as the trend of the previous curve. This means that the input of the predictor is "0".

If the filter number (FN) is 1, the error-prone data is a form of elimination of redundancy, so the error-prone data value is set to 0 or replaced with a predetermined number of decoded samples. If (FN) is 2 or 3, since redundancy is already removed from the data D, the value of the data where an error has occurred is replaced with 0 and output.

Meanwhile, referring to FIG. 5, the operation of the saturation prevention unit 30 will be described. Even though the poor data contains noise processing in the error processing unit 10, the error is accumulated or a saturation of the audio output occurs due to a large error. In order to eliminate the saturation of this output, the audio signal has a long average value of zero, and if the negative value of the long output signal is negative feedback, the saturation caused by an error can be prevented.

In order to calculate the average value for the long time, the number of sampling is calculated. In order that the audio signal to be handled is not damaged in the low range, a sample of 1 / F _L time or more must be taken when the low frequency is F _L.

For example, if F _L = 100ZH, at a sampling frequency of 10MS or more, that is, at 37.8KHZ, more than 400 samples are taken and stored in the memory, and averaged and fed back. FIG. 5 shows a design example of this circuit.

The sixth turning chosen here to be seen that the implementation of the saturation prevention element (30) by using one memory switches are switched so that the multiplied ^2-1 is selected every N / 2 samples.

FIG. 7 is another exemplary embodiment of the saturation prevention part in FIG. 3, and FIG. 8 shows the damping characteristic of FIG. 7, which allows using a single memory to take an alternative average value, although not an accurate average value. Has a value between 0-1, and the response of the filter 30A is ego, This is shown in FIG.

As described in detail above, when an irregular error occurs in the CD-I ADPCM decoder, the error processing is changed according to the selected filter coefficient, and the mean value of the output is negatively feedback or the average value is implemented by using one register. This reduces the noise and has the advantage of preventing the output from saturating to one side.

Claims (5)

The error processor 10 receives the dada (D), the error flag (EF), and the filter coefficient (FN) from the CD-ROM and handles the interpolation of error positions differently according to the filter number (FN) to be selected in the ADPCM encoding process. After receiving the adder 20A for adding the correction data to the output data of the error processing unit 10 and outputting the output data of the adder 20A, calculating an average value for a predetermined time period, the feedback unit is negatively fed. A filter for filtering the output signal of the saturation prevention unit 30 to prevent the saturation of the output signal and the output signal of the saturation prevention unit 30 delayed by one sample through the delay unit 40A by a filter coefficient K ₀ . A filter 50B for performing filtering with a filter coefficient K ₁ on the output signal of the saturation prevention unit 30 delayed by two samples through 50A) and delayers 40A and 40B, and The output signals of the filters 50A and 50B are added to one input signal of the adder 20A. An error processing circuit of a decoder comprising: an adder 20B to be supplied. 2. The method of claim 1, wherein when the selected filter number (FN) is 0, the previous data for a predetermined time is stored and the value is output. If the filter number (FN) is 2 or 3, the error handling unit 10 is configured to replace the value of the data where the error occurs with 0 and output the same. Error processing circuit of the decoder. The saturation prevention unit (100) according to claim 1, further comprising taking a sample of 1 / F _L time or more when calculating the number of sampling to calculate an average value for a predetermined time period, and outputting the sample data as correction data by adding it with input data (D _in ). 30). The error processing circuit of the decoder characterized by the above-mentioned. 2. The error processing circuit of claim 1, wherein the saturation prevention section (30) is configured so that 2 ^-l is selected for every N / 2 samples for the input data (D _in ) and supplied as correction data. The method of claim 1, wherein the response characteristic of the input data D _in is An error processing circuit of a decoder comprising a saturation prevention section (30) configured to pass through an in-filter (30A) and multiply it by a constant (A) to be supplied as maintenance data.