1
$\begingroup$

If a signal's amplitude has a uniform PDF, which is better in the sense of producing a smaller quantization error: optimal scalar quantization or quantization with compressor (non-uniform quantization)?

$\endgroup$
  • 3
    $\begingroup$ It depends on your application and the dynamics of the signal you're trying to quantize. There is no general answer. $\endgroup$ – MBaz Oct 24 '17 at 13:23
  • $\begingroup$ What if the quantization with compressor is with µ-law algorithm ? $\endgroup$ – Spring Oct 24 '17 at 13:26
  • $\begingroup$ @MBaz what about this case ? $\endgroup$ – Spring Oct 24 '17 at 15:00
  • $\begingroup$ That quantizer is optimized for voice signals. $\endgroup$ – MBaz Oct 24 '17 at 15:25
  • 1
    $\begingroup$ normally, if the p.d.f. of the signal is uniform, then a uniform quantization error is the best you can hope for (ignoring time-dependent knowledge like what is assumed with Linear Predictive Coding). if the p.d.f. of the signal is not uniform, then, given the same number of quantized values ("words" or "symbols"), then it is better to trade larger quantization steps for the less probable values to get smaller quantization steps (and error) for the more probable values. Rabiner and Schaefer deal with this in Digital Processing of Speech Signals. $\endgroup$ – robert bristow-johnson Oct 24 '17 at 20:56

Your Answer

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

Browse other questions tagged or ask your own question.