September 25, 2007

Dither this...

by Chris Randall

Has anyone here tried the Crane Song analog dither CD, and if so, does it work as advertised? I realize that statements like "warmer, fuller, nicer" are subjective, but for the sort of people that would order and use something like this in the first place, I would expect the ability to make a statement to the effect of "it's better than UV22HR because..." or something like that.

I suppose at thirty bucks, it's kind of a no-brainer, and I guess I'll just order it and try it out for myself, but I'm curious as to whether it adds Magic Smoke or not.



Page 1 of 2

Sep.25.2007 @ 1:27 PM
Now, I'm a thick eared moron who has never tried this CD.
But, when I think that noise shaped dithering all happens at the 16 bit marker, which is at -96db and most of my songs have an average dynamic range of about 23db. It makes me wonder.

I'm trying to imagine a case where I have my amplifier up to +96db (slightly quieter than a jackhammer) and I can still discern all those near silent 1db details which were salvaged from the 32bit range.

Logically I just can't figure out exactly when this would happen. Dither is only working on the -96db mark. Turn up stereo to +96 with a 1db signal playing sure you might notice differences. But, my pants make 1db of sound when I am just sitting in them


Sep.25.2007 @ 1:39 PM
Chris Randall
Yeah, I'm with you there. I can _sort_ of tell, but not enough to make a qualified statement one way or the other.



Sep.25.2007 @ 3:27 PM
Strange - I might be getting in over my head here, but my understanding is that most implementations of bit downconversion also do noiseshaping of the quantization itself as well as dither shaping. So I'd think using dither shaping only would likely be worse than processes involving both (assuming all else is equal).

But maybe they intend this to be mixed in above -96db, dwarfing quantization noise entirely?


Sep.25.2007 @ 4:41 PM
Chris Randall
That was my thinking. From what I've read, you just drop the audio file in with the fader at 0 (which zero, though? That's the question) and you're good to go. It would necessarily have to be significantly louder than the quantization noise to do anything whatsoever, so it would stand to reason...

In any event, I've ordered it, and when it gets here we'll all know. My Moogerfooger FreqBox shows up tomorrow, though, and I'm way more excited about that.



Sep.25.2007 @ 5:33 PM
You gotta post sound samples of the FreqBox. Preferably dithered all analog-like.

Sep.25.2007 @ 6:03 PM
the freqbox is fun as hell i love mine

Sep.25.2007 @ 6:28 PM
I tried it. It was okay, but sounded heaps better after I put it through my telefunken valve pre-amp then time-stretched it in Abelton Live...

Seriously though, how can a CD claim to be analogue?


Sep.25.2007 @ 7:26 PM
Aenemone Carbuncler
what about the Eventide Timefactor?

or ..wait the new c-labs SSL comp clone :
link [chameleonlabs.c5.htm...]">link []
still their 7602 thrills me for the price!


Sep.25.2007 @ 7:28 PM
Now, I'm no expert, and I don't have experience with this product, but given the theory of dithering, I would say the following.

1) This has less importance if you aren't doing word length or sample rate conversion.

2) This is more or less pointless if the output will be data compressed with a lossy algorithm (this is specifically the type of stuff that gets "lost" ).

3) This will only work for 16bit files (not a killer unless you're releasing more DVDs than I realize)

4) The "any simple digital audio editor" that you are using should actually be something that implements a good quality floating point mix bus that you are fairly sure is losing the minimum amount of bit depth when mixing in the dither signal.

Number 4 kind of depends on whether the signal is on the CD at something like full scale, or if it is already recorded at the 2-3 lowest order bit level that you would want it mixed at. In the case that it is the ultra-low level signal, you would probably be best off trying to find some utility that you know will only be doing a simple addition between the 2 streams. If multiplications are involved, then you are probably undoing some of the "betterness" that could be had (even if this is just being multiplied by the 1.0 coefficient that would represent 0dB).


Sep.25.2007 @ 10:45 PM
Dither is low level so you dont hear it. If you create a 24bit at +23db then downconvert to 16bit without dither, you will hear new sounds. The randomness of the dither noise helps 'smear' the distortion so it's warmer to ear.

It's analogous to a previous thread here about adding street noise to recordings to make them sound less sterile.


Page 1 of 2



Sorry, commenting is closed for this blog entry.