1
$\begingroup$

Standard DPCM

I began learning about DPCM and i've got one issue I am not sure about:

Is the incoming sample m[k] (see image for reference) already quantizied or not? I think it is but with an (assumed) very fine quanitization. So the Quantiser block is doing a requantization with a more coarse quantization, right?

$\endgroup$
0
$\begingroup$

Yes, in a practical setting, all the discrete-time samples in the DPCM block diagram that you've provided are quantized, however with different bit depths. The input sample $m[k]$ to the DPCM block is finely quantized (with high bit depth) while the output of the quantizer (which is the re-quantization of the difference between current sample and predicted current sample) is coarsley quantized, resulting in a bit rate reduction while maintainig the SNR quality of a finely quantized $m[k]$ sample. The bit rate reduction depends on the dynamical characteristics of the input signal and the quality of the predictor.

Note however that, from a dsp theory point of view, the input samples $m[k]$ can be assumed to take on a continuous range of amplitudes, hence unquantized yet.

$\endgroup$
0
$\begingroup$

No it is assumed that m[k] is not quantized, as the main purpose of DPCM is to quantize the difference between samples rather than quantize the whole samples.

ofcourse if m[k] is an image it must been quantized as the video sensor in any digital camera is a digital sensor by definition so it is doing some means of quantization.

$\endgroup$
0
$\begingroup$

Yes, the incoming sample is already digital (perhaps a more generic term than quantized), due to the meaning of PCM or Pulse-code modulation

Pulse-code modulation (PCM) is a method used to digitally represent sampled analog signals. It is the standard form of digital audio in computers, compact discs, digital telephony and other digital audio applications. In a PCM stream, the amplitude of the analog signal is sampled regularly at uniform intervals, and each sample is quantized to the nearest value within a range of digital steps.

Samples are most often coded with integers, possibly with some form of companding, but apparently floating point versions exist.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.