home *** CD-ROM | disk | FTP | other *** search
- Path: news.ottawa.istar.net!usenet
- From: hellmanc@spots.ab.ca (Clay Hellman)
- Newsgroups: comp.sys.amiga.audio
- Subject: Re: Paula chip and Amiga audio
- Date: Thu, 01 Feb 1996 05:59:48 GMT
- Organization: fONOROLA i*internet inc.
- Message-ID: <4epkvi$f80@news.fonorola.net>
- References: <wfblanDL5rJB.IK8@netcom.com> <4doo4s$r5g@news.jhu.edu> <4dsren$omc@news.fonorola.net> <DLLE8z.7rs@GTS.NET>
- NNTP-Posting-Host: pm111.spots.ab.ca
- X-Newsreader: Forte Free Agent 1.0.82
-
- news@GTS.NET (Operator) wrote:
-
- >Clay Hellman <hellmanc@spots.ab.ca> wrote:
-
- >|Example: Take 2 samples, one at 8 bit 44khz and one at 16 bit 22khz.
- >|They both take exactly the same amount of memory. However, if you
- >|listen to them both, there is no comparison to which one will sound
- >|better. The 8 bit 44khz one will obliterate the other one. Bit
- >|resolution ONLY makes a difference with very low level signals (QUIET
- >|SAMPLES!!!).
-
- > Ummm, you're not taking all quantising errors into account -
- > the number of bits/sample also determines how closely the
- > analog signal is replicated in the amplitude domain, whereas
- > the sampling rate determines the fidelity in the time domain.
-
- Yep, I know. But, with 8 bit resolution, you have 256 possible levels
- for the signal (-128 to 127). That is alot. But like I said above,
- when you have a low level signal (say, 1/16th of the full possible
- amplitude), you in essence are only hearing a 4 bit sample (amplitude
- range = -8 to 7). With most ANY sound that is using the full
- amplitude of an 8 bit sample, there is hardly ANY audible distortions
- due to the 8 bit sampling resolution. My point stands, that the
- sampling rate makes MUCH more of a difference than sample resolution,
- in almost all applications.
-
- > Take a 20 KHz sine wave for example: sampled at 44 Khz, there
- > are only 2.2 samples/cycle, so it's not going to produce a
- > very accurate representation of the original analog input.
-
- Yes, you are right. This is a property of the RATE.
-
- > At 1 Khz, with 44 samples/cycle, it should be easy to visualise
- > that time quantisation errors are diminished, but the size of
- > amplitude steps still produce the equivalent of 1/256 peak
- > error for 8-bit encoding, or approximately 0.4% distortion.
-
- Yep.. And this is a property of the resolution. I do see where you
- are coming from.
-
- Anyway, Ive never been able to hear these resolution quantization
- errors when listening to 8 bit low-frequency sine waves. And yes, Ive
- listened for them. (I made a realtime FM synth program for the
- Amiga). However, the lack of frequency response due to low sampling
- rates is always audible - to anyne.
-
- L8r
-
- Clay.
-
-
-