JPH0241935B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a delta modulation conversion method suitable for use when compressing digital information.

The delta modulation (hereinafter referred to as DM) conversion method is a method of digital information compression.
This is the simplest version of the PCM conversion method. The quantization in this DM conversion method is a two-level, 1-bit method for determining whether the comparison error is positive or negative, so it can be constructed using, for example, a D-type flip-flop. As described above, the DM conversion method can be realized with a simple circuit and can provide a high S/N ratio, so in recent years it has been reconsidered as an A-D conversion method.

In the DM conversion method, as mentioned above, only two pieces of information, increase or decrease, are assigned to the data at the sampling point, and since the magnitude of increase or decrease, that is, the Δ value, is constant, as shown in Figure 1A, When the input signal Si changes significantly, the DM output signal So
cannot fully follow, and sudden changes in the signal cannot be reproduced. In other words, if the rise or fall of an analog input signal becomes steep, slope overload noise occurs. In addition, where the gradient is gentle, signal values within the quantization step height are rounded off, resulting in a hunting phenomenon in which small oscillations occur around the correct signal value. In other words, as shown in Figure 1B, when the input signal Si hardly changes, for example when reproducing the DC component, vibrations at half the sampling frequency occur, and the DM output signal So
repeats step-like changes even though the input signal Si does not change. This corresponds to the quantization error in the PCM conversion method, and is called granular noise.

In order to eliminate these drawbacks, it is only necessary to adaptively change the magnitude of increase or decrease depending on the signal.

Conventionally, this method has been carried out by adjusting the magnitude of increase or decrease (Δ value) with respect to the tendency of the DM data sequence (sequence of increase or decrease information). In other words, if the same sign occurs two or more times, the Δ value is doubled the third time, and if a different sign occurs, the Δ value is reduced by half. , ADM) conversion method. In other words, ADM
In the conversion method, the Δ value is kept small in sections where the amplitude of the audio signal does not change significantly to suppress granular noise, and the Δ value is increased in areas where the amplitude changes significantly to reduce gradient overload noise. I have to.

However, in the case of such a conventional ADM conversion method, a delay of several data points inevitably occurs before it catches up with changes in the input signal, which has the disadvantage of distorting the reproduced signal. For example, in FIG. 2A, when the input signal Si suddenly rises, the predicted value cannot follow it and a delay occurs, and the reproduced output signal So becomes delayed as shown by the broken line. Furthermore, as shown in Fig. 2B, when the input signal Si suddenly reaches a constant level from a certain vibration state, the predicted value cannot follow the same as described above, and the reproduced signal So remains at a certain level as shown by the broken line. It gradually settles down to a certain level while repeating the vibration. In this way, in the case of the ADM conversion method, there is a delay of several data points before it catches up with the change in the input signal, distorting the reproduced signal.

In view of these points, the present invention provides a delta modulation conversion method that can eliminate gradient overload noise and granular noise as described above, and also eliminate distortion caused by a delay in a reproduced signal with respect to an input signal, which occurs in the ADM conversion method. It provides a method.

Hereinafter, one embodiment of the present invention will be described in detail based on FIG.

FIG. 3 shows the configuration of this embodiment. In the figure, 10 indicates the transmitting side and 20 indicates the receiving side. Now, on the transmitting side 10, input terminal 1
1 is supplied with an audio input signal, this signal is amplified in an amplifier 12 and then supplied to one input terminal of a comparator 13. The other output terminal of the comparator 13 stores the previous predicted value of the currently input signal from the integrator 14. In other words, the integrator 14 can be regarded as a predictor that has the function of storing past signal values and predicting the current signal value. Then, the comparator 13 compares the current value of the input signal and the predicted value, and then calculates the residual value, for example.
Quantize at a sampling frequency of about 12kHz,
A signal encoded as "1" when the residual value is positive and "0" when the residual value is negative is output to the output side.

Further, 15 is a Δ value setting circuit for adjusting the magnitude of increase or decrease, that is, the Δ value, in accordance with the magnitude of envelope information, which will be described later. When the output signal of the comparator 13 is "1", the Δ value setting circuit 15 transfers the Δ value adjusted according to the envelope information to the integrator 1.
On the other hand, when the output signal of the comparator 13 is "0", the Δ value adjusted according to the envelope information is subtracted from the value of the integrator 14.

Further, in order to obtain envelope information, the output signal of the amplifier 12 is supplied to a rectifying/smoothing circuit 16, where the envelope component is extracted and sent to the comparator 13 described above.
One input terminal of an equivalent comparator 17 is supplied.
Further, the output signal (predicted value) from an integrator 18 equivalent to the above-mentioned integrator 14 is supplied to the other input terminal of this comparator 17, and the comparison residual difference is, for example,
Quantized at a sampling frequency of approximately 3kHz.
Therefore, the encoded envelope information is taken out at the output side of the comparator 17.

19 is a Δ value setting circuit which generates information for increasing the Δ value when the envelope information obtained at the output side of the comparator 17 continues to be at the same level as, for example, [111] or [000]; On the other hand, when "1" and "0" alternate, information is generated in which the Δ value decreases. Therefore, the value of the integrator 18 is a Δ value added in the case of envelope information with continuous "1" or "0", while a Δ value is added in case of envelope information with alternating "1" or "0". is subtracted and supplied to one input terminal of the comparator 17 and also to the Δ value setting circuit 15 described above. Therefore, at the output side of the comparator 17, envelope information is extracted to be transmitted to the receiving side. Further, the Δ value setting circuit 15 adjusts the Δ value according to the level of the envelope information supplied from the integrator 18, and
4. Letting the envelope information supplied from the integrator 18 be V', the Δ value of the Δ value setting circuit 15 is expressed as Δ=kV'. Here, k is a constant. Therefore, when the value of envelope information V' is 1, the Δ value is k, and on the other hand,
When V' is 2, the Δ value is 2k. Then, the Δ value adjusted according to the level of the envelope information becomes "1" of the output signal of the comparator 13,
It is added to or subtracted from the value of the integrator 14 depending on "0".

In this way, the magnitude of the increase or decrease (Δ value) in the predicted value fed back to the comparator 13 is adjusted according to the level of the envelope information extracted from the input signal, so that the The input signal is information-compressed by DM conversion and transmitted to the receiving side 20.
Also, at the same time as this information compressed signal, the comparator 17
Envelope information synchronized with the data signal is also supplied to the receiving side 20 from the receiving side.

The receiving side 20 includes a Δ value setting circuit 21 equivalent to the Δ value setting circuit 15 used on the transmitting side to process the transmitted data signal, and an integrator 2 equivalent to the integrator 14.
2 is provided, and an output terminal 23 is led out from the output side of this integrator 22. Further, in order to detect the transmitted envelope information, a Δ value setting circuit 24 equivalent to the Δ value setting circuit 19 on the transmitting side and an integrator 25 equivalent to the integrator 18 are provided.

The Δ value of the data signal transmitted from the transmitting side 10 is determined by the Δ value setting circuit 21.
The envelope information is adjusted in proportion to the level of the envelope information obtained by the Δ value setting circuit 24 and the integrator 25 and is supplied to the integrator 22 . Then, the supplied Δ value is added to or subtracted from the value stored in the integrator 22 according to "1" or "0" of the transmitted data signal, and the original audio input is output from the output terminal 23. The signal is regenerated and extracted.

As described above, according to the present invention, the envelope information contained in the input signal is extracted, and the magnitude of increase or decrease in DM or ADM is adjusted using this envelope information before transmission. The delay distortion associated with ADM as well as gradient overload noise and granular noise such as those that occur can be eliminated. In other words, for slope overload noise, the level of envelope information is increased where the input signal changes suddenly, and the magnitude of the increase or decrease is adjusted in proportion to the level of the envelope information, so it increases. , the magnitude of the decrease is large enough to catch up with the change in the signal, without causing gradient overload noise. Regarding granular noise, as mentioned above, this noise becomes a problem mainly in silent parts, where there is no signal, and in these silent parts, the level of envelope information is also close to 0. Therefore, since the magnitude of increase and decrease is also small, the granular noise is substantially reduced, and this noise is not perceived audibly. Furthermore, the amount of increase or decrease is adjusted by referring to the envelope information of the input signal at the current point in time to avoid distortion caused by the delay in catching up with signal changes, which can occur with conventional ADM conversion methods. Therefore, no distortion occurs due to this delay.

Note that in the above-described embodiment, the DM conversion on the transmitting side and the receiving side may be replaced with conventional ADM conversion, and the amount of adjustment of the magnitude of increase or decrease may be controlled using envelope information.

In addition, in the above embodiment, the envelope information is
Although the case where the ADM conversion is applied to the Δ value setting circuits 15 and 21 has been described, the case is not limited thereto. The envelope information may be directly supplied to the Δ value setting circuits 15 and 21, or the above-mentioned envelope information may be temporarily recorded on a recording medium in synchronization with the data signal on the transmitting side, and then recorded on the recording medium when the data signal is reproduced on the receiving side. It is also possible to extract and use the envelope information provided.

[Brief explanation of the drawing]

Figure 1 is a diagram to explain gradient overload noise and granular noise accompanying DM conversion, and Figure 2 is a diagram to help explain gradient overload noise and granular noise associated with DM conversion.
A diagram for explaining the operation of the ADM conversion method,
FIG. 3 is a system diagram showing an embodiment of the present invention. 10 is a transmitting side, 12 is an amplifier, 13, 17 is a comparator, 14, 18, 22, 25 is an integrator, 15,
19, 21, 24 are Δ value setting circuits, 16 is a rectifier/
A smoothing circuit 20 is the receiving side.

Claims (1)

[Claims] 1. The first input signal is supplied to one input terminal.
a first Δ value setting circuit connected to the output side of the first comparator; and a first Δ value setting circuit connected to the output side of the first Δ value setting circuit, the output of which is connected to the output side of the first Δ value setting circuit. a first integrator supplied to the other input terminal of the rectifying and smoothing circuit for detecting envelope information of the input signal, and a second integrator supplied to one input terminal of the rectifying and smoothing circuit. a comparator; a second Δ value setting circuit connected to the output side of the second comparator; and a second Δ value setting circuit connected to the output side of the second Δ value setting circuit;
a second integrator whose output is supplied to the other input terminal of the second comparator, and changes the Δ value of the first Δ value setting circuit according to the output of the second integrator. A delta modulation conversion method characterized in that the output signals of the first and second comparators are transmitted respectively.