naturerecordists
[Top] [All Lists]

Re: dither

Subject: Re: dither
From: Walter Knapp <>
Date: Tue, 10 Dec 2002 02:37:50 -0500
Rich Peet  wrote:
> Well, I figured it didn't matter before when I was running standard 
> mics.
> 
> But, I got a few low noise mics and need to understand dither more.
> In recording "faint field" sound with a low noise mic, should I be 
> running "dither" on for edits?
> 
> There appears to be a trade off on using this with little guidance as 
> to when to use it and not for "faint field".
> 
> Anybody know about this stuff here?
> I am just confused.

Dither will somewhat smooth out odd jumps in sound due to the 
processing. It applies some randomization to the calculated results. If 
applied right it's only doing that on a very small scale to cover up 
round off errors and such like. You probably don't want it to be applied 
to the point where you can hear obvious differences you can clearly 
identify.

One trade off is that dither can smooth out what you want. It's 
generally not able to tell processing errors from actual sharp 
transitions in sound you want. So, it's effect can be to dull or muddy 
the sound down. Especially if it's applied multiple times.

Ideal is if your software does it's internal calculations at a higher 
bit depth. That's where something like 24 bit makes a big difference, 
not in storing the sound. With the higher bit depth there is not much 
need for dither. The comparison with graphics would be like where 256 
color gifs need dither, but 16 million color tiff files don't. 16 bit 
sound is good enough for storage, but for processing will cause some 
problems.

I found that the higher the quality of the sound the more careful you 
have to be in processing it. Before starting in to make up stuff for the 
Frog CD, I spent a couple months just trying various processing on the 
sound as my previous scientific work did not allow much filtering. And I 
still feel like I'm lost half the time.

My own guess is you should try it first without dither. I'm guessing 
that will be better unless your software is doing calculations at very 
low bit depths. The dither is likely to muddy the faint sounds more than 
the loud ones. I use different software since I use a mac, so don't run 
into the problem quite the same way. All my software is processing at 
higher bit depth to start with so dither is not used much. Since some of 
this is in the realm of the style of sound you like, you will have to 
experiment to find what you like the best.

I find that faint field handles no different than the high end in many 
ways. Digital generally has a greater dynamic range than the mics or 
what they are recording, so you can fairly easily amplify faint stuff 
and bring it up with little processing. But note this only became 
possible fully after I got the Portadisc. Even feeding a walkman with a 
pre won't get the full ability. And it's easier to mess up the faint end 
with filters. With low noise mics there is a more or less hidden reserve 
of high quality sound and filters are really designed to just deal with 
the main, higher level sound.

It is one of the problems when you bring in some higher quality piece of 
equipment into a processing chain. It usually will highlight some other 
part of the chain that's now the limiting factor. You will nearly always 
get improvement, however.

My own experience where I went to quality mic first, replaced the 
walkman later was that I did not get the full quality out of the mic 
until I replaced the walkman. The higher quality mic did improve what I 
got with the walkman, but then the Portadisc made another jump. And with 
the quality the mic and portadisc could do it became a little more 
crucial to use digital sound input. Using a preamp with the Portadisc 
may give me a slight bit more, but it's not much of a jump. The MKH mics 
I've gotten will open up possibilities as well. Though I'm more noticing 
the ability to choose my recording field better than the low noise aspects.

It all points to keeping a outfit fairly balanced. I tend to try to have 
the highest quality front end first, then add down the chain later.

Walt




________________________________________________________________________
________________________________________________________________________



Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/ 


<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