JPH03216871A - Read data correction method - Google Patents

Abstract

PURPOSE:To reduce the reproduced read error by outputting N-th read data a specific component of the sampling frequency for at the time of reproducing ahead of the output time of operation results. CONSTITUTION:When the decimal part of the result of addition between the increment value of the read address of a waveform memory corresponding to 1/2fs (fs is the sampling period) and the preceding read address is smaller than 0.25 or larger than 0.75 in the sampling after the end of the processing of sampling NoS-2 for reproducing, the integer part of the read address of the addition result is used to read out waveform data. The read address of the waveform memory is not always designated by the fixed sampling period, but N-number of opportunities of sampling are made by periods obtained by equally dividing the sampling period into N and the point which is not distant from the integer part is designated as the read address of the waveform memory. Thus, sampling data at the time of reproducing is not remote from the analog waveform of original sounds, that is, the error is reduced.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a read data correction method, and particularly to a read data correction method suitable for reducing distortion of read data in a PCM sound source.

[Prior Art] Conventionally, when reading data from a waveform memory in a PCM sound source such as an electronic musical instrument, two waveform data read out using two integer part addresses are subjected to proportional calculation or trigonometric function using a decimal part address. A method of correcting by calculation using , etc. was used.

FIG. 3 shows an example of this, and is an explanatory diagram of a method that has means for controlling the integer part and the decimal part of an address and performs interpolation. The solid curve connecting the ● marks in the figure shows the analog waveform of the original sound, and the ▲ marks show sampling points. In this example, in one sampling period,
Read out two waveform data W(R) and W(R+1) corresponding to integer addresses R and R+1 in the figure, and from this and the decimal part S of the address, W(R)'=W(R)+s{ W(R+1)−W(R)}
However, W(R)': Corrected waveform data (address R+s
)W(R): Read waveform data (address R)W(R
+1): Successive waveform data (address R+1) R: Integer part of read address S: Decimal part of successive address Waveform data corresponding to address (R+s) is obtained by interpolation calculation. In addition, in FIG. 3, S indicates the sampled fish during reproduction, and fs indicates the sampling period.

In addition, Out(S-2) and Out(S-
2 >+ ... is sampling NIS-2, S-1
, . . . shows D/A conversion outputs corresponding to .

Incidentally, examples of the above-mentioned correction method include the methods disclosed in Japanese Patent Publication No. 54-19291 and Japanese Patent Publication No. 60-38716.

[Problem to be solved by the invention]

However, the above-mentioned conventional technology has a fundamental problem in that even if interpolation is performed, a readout error (E, E' in the figure) always occurs. In other words, in the case of Figure 3, the curve connecting the sampling points indicated by ▲ looks like a broken line, and compared to the solid curve connecting the ● marks (the analog waveform of the original sound), it is clearly distorted. Become. This results in subtle changes in the timbre of the reproduced musical sound.

The present invention has been made in view of the above circumstances, and its purpose is to provide a read data correction method capable of solving the above-mentioned problems in the conventional technology and reducing the read errors that occur. There is a particular thing.

(Means for Solving Problem II) The above object of the present invention is to reduce the increment data of 17N (N: natural number) of the waveform memory read address that determines the sampling frequency during playback to 1 in a read data correction method in a PCM sound source. The method includes means for adding N times in a sampling period, and means for determining whether the decimal part of the waveform memory read address is closer to O or 1 than an arbitrarily specified reference value every time the addition is performed N times. When the judgment result by the judgment means is true, the N-Mth (M
4N or less) is read from the waveform memory, and if it is not determined to be true in the middle, the waveform data is read from the waveform memory at the Nth time, and the result of calculating the waveform data read at the N-Mth time. This is achieved by a read data correction method characterized in that the read data correction method is characterized in that the read data is outputted earlier than the time at which the result of calculating the N-th read data is output by M/N of the sampling period during reproduction.

[Effect]

In the read data correction method according to the present invention, instead of always specifying the read address of the waveform memory at a constant sampling period, the sampling period is divided into N equal periods, and opportunities are created for sampling N times, and each sampling period is By selecting a point where the specified read address of the waveform memory is not far from the integer part, and by specifying such a point not far from the integer part as the read address of the waveform memory, sampling during playback can be improved. This ensures that the data does not deviate from the original analog waveform (that is, the error is small).

Thereby, it is possible to realize a read data correction method that can prevent subtle changes in the timbre of the intended reproduced musical tone.

[Example] Hereinafter, an example of the present invention will be described in detail based on the drawings. In addition, in the embodiment described below, the above-mentioned N=
Case 2 will be explained as an example.

1. [J] is an explanatory diagram of a read data correction method showing an embodiment of the present invention. In Figure 1, symbols R, S, f
s is used in the same meaning as shown in FIG.

In this embodiment, after the processing of sampling NaS-2 during playback is completed, in the next sampling, 1/
The increment value (F
D/2) to the previous read address, and if the decimal part is closer to O or l than the reference value (0.25 for N=2), that is, less than 0.25, or If it exceeds 0.75, the waveform data is read using the integer part of the read address resulting from the addition.

FIG. 2 shows a flowchart of the above-mentioned operation. The details of the operation will be explained below based on FIG. 2.

First, the cycle of the last playback, that is, not the address value actually read last time, but the final address value after adding the increment value (F D / 2) of the read address of the waveform memory twice, and the FD /2 (step 21), the decimal part of the addition result is O or l lower than the arbitrarily specified reference value (0.25 in this case).
It is determined in step 22 whether or not it is close to .

If the result of this judgment is true, the waveform data is read out at the current waveform memory read address and sent to the calculation section (step 24), and at the same time, the above-mentioned FD/2 is added to this read address again in step 25. , this is latched as the final waveform memory read address of this reproduction cycle.

If the result of the determination in step 22 is false, the waveform data is read out at the waveform memory read address after adding the above-mentioned FD/2 (steps 23 and 24), and is sent to the calculation section at the same time. , this is latched as the final waveform memory read address of this reproduction cycle.

In the example shown in FIG. 1, after the processing of sample S-2 during playback is completed, first, the increment value (FD/2) of the read address of the waveform memory mentioned above is added once,
In the process of sampling NaS-1.5 during playback, the decimal part of the read address at this point is set to the above-mentioned reference value 0.
．． It is compared with 25. In this case, since the fractional part of the read address at the point of sampling ffis-1.5 during reproduction is smaller than the above-mentioned reference value 0.25, the waveform memory data corresponding to the read address at this point is read.

Next, the increment value (FD/2) of the read address of the waveform memory described above is added once again to the value of the above address, and the same operation as above is repeated using this as a reference.

In this case as well, since the decimal part of the read address at the point where the increment value (FD/2) of the read address of the waveform memory is added once is smaller than the reference value 0.25, the read at this point is Read the waveform memory data corresponding to the address.

Similarly, the increment value (F.D/2) of the read address of the waveform memory is added once again to the value of the above address, and the same operation as above is repeated using this as a reference. This time, since the decimal part of the read address at the point where the increment value (FD/2) of the waveform memory read address is added once is larger than the reference value 0.25, the increment value (FD/2) of the waveform memory read address is FD/
Add 2) once more. This time, the decimal part of the read address at the point where the increment value (FD/2) of the read address of the waveform memory is added twice is greater than 0.75, so the waveform memory data corresponding to the read address at this point is read. put out.

Thereafter, by repeating the same operation, it is possible to obtain reproduction characteristic sampling data with relatively smaller errors than in the past. It should be noted that the output timing of the D/A conversion output needs to be adjusted according to the sampling timing during reproduction, as shown in the figure.

In the above embodiment, an example was shown in which N=2, but this is just an example, and the present invention should not be limited to this. The value of N is preferably about 2 to 5.

[Effect of the Invention 1] As explained in detail above, according to the present invention, in the correction method of read data in a PCM sound source, 1/N (N: (a natural number) N times in one sampling period; and means for determining whether the decimal part of the waveform memory read address is closer to 0 or l than an arbitrarily specified reference value each time the addition is performed N times. When the determination result by the determination means is true, the waveform data is read from the waveform memory at the N-Mth time (M: a natural number less than or equal to N), and when the determination result is not determined to be true on the way, the waveform data is read out at the Nth time. At the same time as reading the waveform data from the memory, the result of calculating the waveform data read out the N-Mth time is set to M/M of the sampling period during playback than the time when the result of calculating the data read out the Nth time is output.
Since the output is made earlier by N, it is possible to realize a read data correction method that can reduce the read error that occurs, which is a remarkable effect.

[Brief explanation of drawings]

FIG. 1 is an explanatory diagram of a read data correction method which is an embodiment of the present invention, FIG. 2 is an operation flowchart of the embodiment, and FIG.
The figure is an explanatory diagram of a conventional interpolation method. R-α, R, R10α: waveform memory address integer part, S-
α, S, S+α: Sampling seed during playback, OutD/A
Conversion output fish, 21-25: Processing steps. Figure 2 To the waveform data calculation section

Claims (1)

[Claims] (1) In the correction method of read data in a PCM sound source, 1/N (N:
a natural number) N times in one sampling period;
and means for determining whether the decimal part of the waveform memory read address is closer to 0 or 1 than an arbitrarily specified reference value every time the addition is performed N times, and the determination result by the determination means is true. Sometimes, the waveform data is read from the waveform memory at the NMth time (M: a natural number less than or equal to N), and if it is not determined to be true on the way, the waveform data is read from the waveform memory at the Nth time, and at the same time, the waveform data is read out from the waveform memory at the N-Mth time. A read data correction method characterized in that the result of calculating the waveform data read out is outputted earlier by M/N of the sampling period during reproduction than the time at which the result of calculating the data read out the Nth time is outputted. .