Which Dither?

[crumhorn]

Thanks for all your inputs.

Much reading to catch up on.

[davepermen]

I'm coming from a math background. I'm weird..

but i was always the one who had to teach the ones that didn't understood the teacher, so i always try to find ways to explain it strange, but understandable..

that's why i, almost always, refer to visual clues, like pictures..

so far no one could explain me music.. *crying*

the harmonic theories are hell for me. they are so illogical

[SubFunk]

Thanks for all your inputs.

Much reading to catch up on.

or leaving it out, honestly, unless you want to teach or going into programming and come up with a better dither algos then the POW-r

i am not being funny here, if it is out of interest, etc. cool, all i want to say is, that understanding dithering is not going in any way to enhance the ability to choose the better algo for you material, because it is always different how the dithering will affect the material, always!
tarekith's advise is OK to use the ears in the end of the day, it is just in most cases a very subtle and not easy to spot difference.

or simply go with that rule set, seriously: *unless you are able to hear it.

POW-R #1: optimized for simple program of average dynamic range, such as spoken word.
POW-R #2: optimized for low dynamic range program, such as rock music.
POW-R #3: optimized for complex, high dynamic range program, such as orchestral music."

ok, to not be flamed, in short... it's cool to know, but this kind of knowledge (talking about dithering) want affect the ability to choose better.

[Tone Deft]

I'm coming from a math background. I'm weird..

but i was always the one who had to teach the ones that didn't understood the teacher, so i always try to find ways to explain it strange, but understandable..

that's why i, almost always, refer to visual clues, like pictures..

so far no one could explain me music.. *crying*

the harmonic theories are hell for me. they are so illogical

I also come from a math background (EE) but I was one of the guys you'd be explaining it to after class. Linear Systems is where I could only LOL in terror at the abstract nature of math. Laplace transforms and the Z-plane. I knew it for the final but I probably can't even do a basic integral anymore. damn, I used to be able to do Maxwell's Equations - in several forms!! meh, once you run the gauntlet the concepts stick and you don't need the complex math, you just need to know the right DSP guy to ask.



I think subfunk's advice is good, don't go inventing problems that aren't there. do you need to dither? Ableton could implement a "Diamond Dither" that was the best sounding dither ever invented. they could rant about it and hype you up over it. in reality it would do absolutely nothing to the signal but you could use it with the warm fuzzy feeling that you used the best dither ever invented and you'd be convinced that your mix was better off with it. (no offense to you, just an example of how our heads can be fooled and we can overthink a problem.)

[davepermen]

yeah, like diamond shreddies

[Tarekith]

[davepermen]

they should build a diamond summing engine

[Tone Deft]

the 64 bit summing engine was pretty close. the performance gain was in the negative three digit range. but hey, it looks good on the box!

too true.

[jon_moore]

Dither is extremely scientific, and very reproducible. It's also a highly complex subject that makes most people's eyes (including mine) gloss over once you get into the details. Saying use your ears is a good thing usually, but with dither we're talking about something that's sooooo quiet compared to the music that like I said I doubt most people can even hear it, let alone the differences between the versions. Certainly worth a try though, don't get me wrong. This might help people with what to listen for:

The POW-r Dither plug-in provides three types
of noise shaping, each with its own characteristics.
Try each noise shaping type and choose the
one that adds the least amount of coloration to
the audio being processed.

Type 1 Has the flattest frequency spectrum in
the audible range of frequencies, modulating
and accumulating the dither noise just below
the Nyquist frequency. Recommended for less
stereophonically complex material such as solo
instrument recordings.

Type 2 Has a psychoacoustically optimized low
order noise shaping curve. Recommended for
material of greater stereophonic complexity.

Type 3 Has a psychoacoustically optimized high
order noise shaping curve. Recommended for
full-spectrum, wide-stereo field material.

OR:

POW-R #1: optimized for simple program of average dynamic range, such as spoken word.
POW-R #2: optimized for low dynamic range program, such as rock music.
POW-R #3: optimized for complex, high dynamic range program, such as orchestral music."


If you really want to know more, make yourself some popcorn and do a search on dither at Gearslutz. Take some aspirin in advance too for the inevitable headache.

Brilliant explanation. I learnt more in this post than I have in the previous ten years! Terekith, I salute you!

[crumhorn]

Dither is extremely scientific, and very reproducible. It's also a highly complex subject that makes most people's eyes (including mine) gloss over once you get into the details. Saying use your ears is a good thing usually, but with dither we're talking about something that's sooooo quiet compared to the music that like I said I doubt most people can even hear it, let alone the differences between the versions. Certainly worth a try though, don't get me wrong. This might help people with what to listen for:

The POW-r Dither plug-in provides three types
of noise shaping, each with its own characteristics.
Try each noise shaping type and choose the
one that adds the least amount of coloration to
the audio being processed.

Type 1 Has the flattest frequency spectrum in
the audible range of frequencies, modulating
and accumulating the dither noise just below
the Nyquist frequency. Recommended for less
stereophonically complex material such as solo
instrument recordings.

Type 2 Has a psychoacoustically optimized low
order noise shaping curve. Recommended for
material of greater stereophonic complexity.

Type 3 Has a psychoacoustically optimized high
order noise shaping curve. Recommended for
full-spectrum, wide-stereo field material.

OR:

POW-R #1: optimized for simple program of average dynamic range, such as spoken word.
POW-R #2: optimized for low dynamic range program, such as rock music.
POW-R #3: optimized for complex, high dynamic range program, such as orchestral music."


If you really want to know more, make yourself some popcorn and do a search on dither at Gearslutz. Take some aspirin in advance too for the inevitable headache.

Brilliant explanation. I learnt more in this post than I have in the previous ten years! Terekith, I salute you!

++1

[crumhorn]

@sub funk

No offence taken and I totally agree to be honest. In general I go along with the just use your ears argument, but in this case my ears/equipment just aren't up to the job, I can't hear any
difference at all, though I guess that it must make a difference to somebody.

But when I'm exporting audio I keep seeing all those options and, being a bit of a techy geek, I can't help but wonder what they all do and what the Abes were thinking when they decided which algorithms to include.

[Tarekith]

Those are the standard options any developer gets when they license the POW-r dither.

To be fair as well, some of what I posted about the different options was just copied from the dither page on Wikipedia cause I was too lazy to retype it.

[jon_moore]

Those are the standard options any developer gets when they license the POW-r dither.

To be fair as well, some of what I posted about the different options was just copied from the dither page on Wikipedia cause I was too lazy to retype it.

Dig your honesty mate! Still as you rightly said, whenever I've read detailed documentation on Dithering, I usually get to page 4 and my eyes glaze over, next I'm on page twenty thinking I haven't taken a single thing in for the last half hour...

[SubFunk]

Those are the standard options any developer gets when they license the POW-r dither.

To be fair as well, some of what I posted about the different options was just copied from the dither page on Wikipedia cause I was too lazy to retype it.

i thought so, it looked familiar.

@ crumhorn, no missunderstanding here, i am all into the importance of 'understanding' what is going on, what i mean in this particular case, dithering, any understanding will not allow you to make a better choice which one to use. if it is for your own interest, good, if you expect making better dithering algo decisions, forget it, use your ears, but it's in most cases extremely subtle, that's all.

[SubFunk]

And to add, as tarekith said in 95% of cases you are good to go with POW-r2
besides that is also a reason why you have mastering engineers for. you want to do it all by yourself, fine!
get some decent monitors and train your ears! learn how and what to listen for
do you think professionel sound & mastering engineers have all 'golden ears' ? BULLSHIT!
they simply trained their ears, just the way you learn swimming or cycling, by simply starting to rely (oh, did i spelled that right? I mean trust) on their ears and by paying attention to details and by investing into the most important item ever in any studio, MONITORS, instead of buying fancy synths first.

Period. end of also partly a rant.
Not addressed at anyone in particular, just a personal statement, summing up several threads involving 'listening' and 'use your ears'

maybe i should start a separate thread, this is bugging me since a very long time, everyone seems to be ok with learning, swimming or cycling or driving a car, but only a little amount wants to learn to listen.... golden ears... pffffft.... Muahhahaha!!!