[Top] [All Lists]

Re: Re: again about compression

Subject: Re: Re: again about compression
From: Dan Dugan <>
Date: Wed, 3 Apr 2002 17:26:45 -0800
>  > What I've done to evaluate the effect of bit rate reduction is to
>>  take a fine example of nature sound (a bit of a Gordon Hempton CD),
>>  transfer it digitally to an MD, then transfer that back, again
>>  digitally, to PCM.
>>  I can then do blind A/Bs between original and MD copy with my ABX
>>  setup. No one has been able to distinguish the recordings yet. And I
>>  can subtract the original and the encoded-decoded, resulting in a
>>  difference file that is what the encoding took out. An interesting
>>  experiment.

>I take it you managed to find a way around clock syc to get the
>difference file. When I tried this using the Portadisc via USB to my G4,
>I got a beat frequency in the difference file that basically made it
>useless. Did find out the two clocks differed by about 1 cycle every 2.7
>seconds or so.

My Pro Tools can be set to synch to the incoming digital signal.

>I should try it again using the Roland and see if the results are the same.
>Your experience is typical of what's found doing these tests carefully.
>The blind testing has worked out this way since ATRAC 4 came in.
>What I found when I did the version for the group using a analog
>transfer was that the difference file simply contained what appeared by
>sonogram to be a attenuated version of the original (or the MD copy).
>Which may explain the inability of people to distinguish. It removes the
>same thing it keeps. Which is what a good compression method is supposed
>to do.

What I hear in the difference file is not what you describe. It's 
low-level, swishy and staticy sound.

-Dan Dugan


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

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