[Imc-tech] Re: [Imc] U-C IMC on INN--14 Feb
Paul Riismandel
p-riism at uiuc.edu
Mon Feb 19 02:24:18 CST 2001
At 11:33 PM 2/18/2001 -0600, you wrote:
>Compounding data loss notwithstanding, it's strange to me that an MP3
>would degrade a long-distance telephone call, which is such a "data-poor"
>signal. For the curious, here's one way you might get at the bottom of
>this compression strangeness:
Well, it's not the mp3 by itself that I think causes degradation, but the
compounding of unlike compression algorithms, including digital phone
codecs, minidisc's ATRAC and then mp3 (which were the different codecs used
on our end). Nevermind the effects of poor phone interfaces and the like.
The degradation caused by intermixing codecs I have experienced time and
again x-ferring mp3s to minidisc and back. Loss in the high end can become
evident with just one codec->mp3->MD cycle, and sometimes more apparent
depending on the mp3 encoder and the complexity and quality of the original
sound recording.
With specific regard to INN, I believe we submitted our story in mp3 and
it's not unlikely that it was decompressed, edited in .wav or .aiff and
then recompressed with the final program in mp3 adding yet another codec
cycle. So the resulting cycle of codecs is: phone ->MD->mp3->mp3, each
resulting in further loss of data.
Again, I hear this degradation all the time even on the big radio news like
NPR. Part of that stems from the fact that nearly all national satellite
distributed programming goes through some sort of lossy codec cycle (ISDN,
MPEG or other proprietary format) after production and prior to or as a
part of sat distribution.
While I'm sure it might be sort of correctable for us on our end, the time
and expertise to do so I reckon would be prohibitive, and sometimes
rendered less effective by subsequent codec cycles out of our control.
--Paul
More information about the Imc-tech
mailing list