naturerecordists
[Top] [All Lists]

Re: ATRAC don't get no respect

Subject: Re: ATRAC don't get no respect
From: Walter Knapp <>
Date: Thu, 24 Apr 2003 00:57:08 -0400
Dan Dugan wrote:
> Walt, you wrote,
> 
> 
>>As I've noted it's a error to think ATRAC even keeps 20% of the original
>>data. The more I dug into it, the more it became clear that it should
>>rather be thought of as synthesizing a whole new set of data, and
>>storing instruction on how to do this. Yes, a approximation, but so is
>>anything that comes out of a A/D - D/A cycle.
> 
> 
> Since ATRAC has higher fidelity than, say, pre-Dolby analog master 
> tape recording (compare to 15 ips 2-track: better s/n, flatter 
> frequency response, lower distortion, better group delay), one could 
> ask "what percentage of the data does tape recording throw away?"

Different world. One more or less assumes that analog recording methods 
don't throw away data, a infinite amount of data points in and a 
infinite amount out. But it may store it imperfectly. I suppose you 
might could come up with something, since analog is a infinite amount of 
data it would be some percentage of infinity.

And you are shooting at one of my original favorite recording methods. 
Really hated it when cassette pushed reel to reel aside. I'm not 
surprised by your statement, however.

In digital one has data points that one could count, in theory. So how 
many are saved can also be counted. Because ATRAC occupies 20% of the 
disk space that the uncompressed data it stored occupies there is this 
leap that somehow it saves just 20% of the original data, each point it 
saves unchanged. I can't see how 20% of the raw data saved like that 
would come out all that good. And reading the descriptions of what's 
saved and how leads to the conclusion it does not save the original data 
at all. It saves a sort of description of what the original data 
contained, a blueprint for making it. You certainly would not get 
anything usable out of a D/A if you fed it what's recorded. The decoder 
understands the description and using that makes a new set of data 
points to feed the D/A. Not a mix of old and new points. The distinction 
is a important one. As I'm sure we all know a description of a thing can 
be many times smaller than a thing and yet describe it in a very 
detailed manner. ATRAC manages to describe all of a sound's samples in 
just 20% of the space the samples occupy.

Walt




________________________________________________________________________
________________________________________________________________________

<Prev in Thread] Current Thread [Next in Thread>
Admin

The University of NSW School of Computer and Engineering takes no responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the naturerecordists mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU