Groups | Search | Server Info | Login | Register
Groups > comp.speech.research > #86
| Newsgroups | comp.speech.research |
|---|---|
| Date | 2015-07-24 22:17 -0700 |
| References | <370ED5CF.DC0838E6@kt.co.kr> <370FA02A.77483118@earthlink.net>#1/1> |
| Message-ID | <894a9149-aa83-4fec-8d3e-b9c51ed6d99e@googlegroups.com> (permalink) |
| Subject | Re: [Q]Are the ADPCM and the PCM lossy or lossless??? |
| From | amitvec1014@gmail.com |
On Saturday, April 10, 1999 at 12:30:00 PM UTC+5:30, Joseph S. Wisniewski wrote: > "Baik, Seong-bok" wrote: > > > > Usually, the PCM seems to be thought as lossless, but most people > > say the ADPCM is lossy. Could anyone please tell me why? > > > > Thanks in advance... > > PCM (Pulse Code Modulation) is loseless becuase it's not a data > compression method. It's just simply the storage or transmission of the > raw values (the Code for the Pulse) that come from an A/D converter. > > ADPCM (Adaptive Differential PCM) takes the differential of a PCM signal > (the difference between successive samples). At this point, you haven't > actually lost any information yet, since the difference between two 16 > bit samples is (technically) a 17 bit number. The "lossy" part comed > from the "adaptive" slgorithm. An adaptive scaling factor is applied to > these differences, to get them down to 4 bit numbers (well, some ADPCM > schemes use from 2 to 6 bit numbers, but 4 is most common). When the > scaling factor is small, you capture low level changes in the signal > well (and most any low frequency data) but high level, high speed > transients (large values of high frequency components) will distort > because you don't have enough dynamic range to track them, so you run > into a slew rate limit (clipping of the differential). When the adaptive > scaling factor is large, you capture large dynamice well, but small > variations are distorted, since the quantitization step size is too > large, and you get quantitization noise. So the signal you get back > almost always looses some information. > > Hope this helps. > > Joe if any system using quantizer then it will be definitely lossy
Back to comp.speech.research | Previous | Next | Find similar
Re: [Q]Are the ADPCM and the PCM lossy or lossless??? amitvec1014@gmail.com - 2015-07-24 22:17 -0700
csiph-web