[Top] [All Lists]

Re: Minidisc Mic Input

Subject: Re: Minidisc Mic Input
From: Walter Knapp <>
Date: Mon, 23 Aug 2004 11:22:42 -0400
From: Dan Dugan <>

>>>This will only work if the signal level is not just reduced, but
>>>eliminated. The quiet stuff gets compressed along with everything else.
>>>So you would want to drop the signal by 90+ dB relative to your loudest
>>>signal I'd assume.
> I thought perceptual coders like MP3 and ATRAC work with a "bit
> budget." When the signal has a lower level it needs a smaller
> bandwidth and thus less data compression. Maybe I'm wrong about this.

I see no reason why this should be so, any more than it would be with
high level signals.

Yes, ATRAC intelligently allocates bits where needed. But it's much more
focused on transitions. Sharp transients will invoke a more detailed
compression, for instance.

>>>I'm also not so sure about eliminating all the low end. That, too, is
>>>part of the natural ambiance. Even if it's a problem to record and play
>>>back properly.
> I feel the same way, I'd rather preserve all the bandwidth my mics
> can pick up--unless I'm having a problem with something that isn't
> part of what I want, like freeway rumble or wind noise in the mic. I
> often high-pass stuff later, but I want the option.

I agree, often we have no choice but to mangle the low end to get rid of
noise. But I want it as a choice so if I can get that low end I will.

>>>It would actually make more sense to eliminate all the
>>>high end, as most don't really hear that.
> ATRAC -does- eliminate the high end, which eliminates a lot of
> bits--I see what appears to be a signal-dependent variable low-pass
> around 16K.

You must be working with ATRAC 4. Newer ATRAC it's quite a bit above
that. Even ATRAC 4 got the high stuff, just dropped the very quiet parts
of the high stuff. I used to estimate it's dropoff at about the -50dB
point below the highest levels at that frequency. And if you work that
over against the normal adult hearing it's not a problem.

ATRAC 4.5, which is what the portadisc has is pretty good up to
18-20khz, then some slight effect like ATRAC 4 above that. Much less
than ATRAC 4. I've not examined the even newer ATRAC.

When you think about it, at the high frequency end there are few samples
relative to the waveform. Any compression will loose some of the detail
at the highest end where every bit counts. But, it will still beat
analog tape by a long ways, and that's still being touted as good for
nature recording.

As I've stated before, if the highest third of the frequency range is
what's important for you then you should be using a lot higher sampling
rate than 44khz. For most nature recording it's not that important.



<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