|
What the hell is dithering ?
|
View this Thread in Original format
| harriz |
I am planing on buning some reedits on to cd using a progran called bias peak le.
Which of the dithering options I sould sellect ?
It gives me three options
powr 1
powr 2
powr3
Also Sould I normalize the recording or will that ruin the quality?
Help me out. |
|
|
| aquila |
| I've also wondered what Dithering is. I know what it is in graphics, but I dunno exactly what it's for in audio. |
|
|
| harriz |
| quote: | Originally posted by aquila
I've also wondered what Dithering is. I know what it is in graphics, but I dunno exactly what it's for in audio. |
Yeah I am not sure either.
I dont know if I should burn a cd with this program.
The stereo is prerecorded and mastered, all I did was did an edit.
I dont know weather I should normalize or not
and which of the dithering options to use..
hopefully somebody who knows will post. |
|
|
| FrancoR |
| Waiting for DigiNut or Derivative answer :p |
|
|
| Invertika |
Dithering is the process of converting a 24 bit audio file down to a 16 bit, which is the CD standard. Due to 24 bit having a higher dynamic range, to convert the audio file down to 16 involves some loss of the original signal. I'm sure Diginut or someone can provide the proper terms, but basically its a more complex method of shaping the sound to get the 16 bit file to sound the same as the 24.
The opposite of dithering is called 'truncating', and i read somewhere that if truncating is like hacking off the excess 8 bits with a chainsaw, then dithering is like doing it with a fine scalpel. There's still loss, but its more refined and less noticeable.
I'm pretty sure pow-r 3 is the best to use, its the most sophisticated algorithm |
|
|
| harriz |
| quote: | Originally posted by Invertika
Dithering is the process of converting a 24 bit audio file down to a 16 bit, which is the CD standard. Due to 24 bit having a higher dynamic range, to convert the audio file down to 16 involves some loss of the original signal. I'm sure Diginut or someone can provide the proper terms, but basically its a more complex method of shaping the sound to get the 16 bit file to sound the same as the 24.
The opposite of dithering is called 'truncating', and i read somewhere that if truncating is like hacking off the excess 8 bits with a chainsaw, then dithering is like doing it with a fine scalpel. There's still loss, but its more refined and less noticeable.
I'm pretty sure pow-r 3 is the best to use, its the most sophisticated algorithm |
thanks:) |
|
|
| DigiNut |
Yes, that's pretty much the correct answer. Understand that dithering is not necessarily from 24-bit down to 16-bit; it can be from any resolution down to any lower resolution. Most Cubase users probably work with their track in 32-bit and dither it down to 24-bit for offline processing (mastering), then dither it again down to 16 bit at the end.
If you want to know why dithering is important then try using a bitcrusher. ;) That's truncation, and essentially it cuts off or screws up a lot of the high frequency and makes for a generally noisy sound.
UV22-HR is the best dithering algorithm... Waves IDR is pretty good too in my experience. |
|
|
| harriz |
| quote: | Originally posted by DigiNut
Yes, that's pretty much the correct answer. Understand that dithering is not necessarily from 24-bit down to 16-bit; it can be from any resolution down to any lower resolution. Most Cubase users probably work with their track in 32-bit and dither it down to 24-bit for offline processing (mastering), then dither it again down to 16 bit at the end.
If you want to know why dithering is important then try using a bitcrusher. ;) That's truncation, and essentially it cuts off or screws up a lot of the high frequency and makes for a generally noisy sound.
UV22-HR is the best dithering algorithm... Waves IDR is pretty good too in my experience. |
Thanks for explaining.
Hey by the way :
''aigis-breaking up the wrong tree''
Yay !! FUN!!! :D :D :)
I was plesently suprised!
This is very cool and funky!!
I will save this !!
I am listenng to it right now...
Kinda has a ''zabiella-tyrant'' type of grove going to it ...
How did you make this? |
|
|
| Derivative |
sorry guys but you are all wrong. digi is close though :)
dithering is the process of adding low level noise to a recording.
why would you want to do that you say?
when you convert a 24 bit render to a 16 bit render.
when you do this you are truncating bit depth and the result is called quantisation noise. depending on how many bits you are losing, this noise is more or less evident. at 24 bit down to 16 bit though, in most cases its noticeable only by complete audiophiles on really good soundsystems and studio techs with really accurate monitors.
for all intents and purposes, all you need to know at this point is that a 24 bit soundwave dithered correctly down to 16 bits will always sound more accurate than a 24 bit soundwave rendered straight to 16 bit. but never dither a track more than once or you the low level noise will become more audible. you can do this accidentally if you are reimporting dithered wavs into a project. it will sound when they are dithered again and it is noticeable.
dithering is best explained in terms of pictures because it will make more sense. if you have photoshop or paint shop pro or any program with a dither filter then try it out.
ill explain if you dont have any of those programs. lets say you have a picture of an airplane on your pc. lets say its a 32 bit bmp which allows for a possible reproduction of 16.7 million colours. the bit depth allows for this number of possible colours. the result is a very lifelike image that captures most, if not all of the perceived detail of the real thing.
when you save this file to an 8 bit 16 colour bmp then you will find that the image loses most of its detail. you only have 16 colours now so there is no smooth transition between colours and shades - now there are gradients. moreover the image only consists of flat colour and looks more two dimensional. you cant make out light or dark or shadow or colour intensity anymore because the image is now only described in terms of 16 shades.
if you apply a dither filter before you render the 32 bit bmp down to 8 bits what you do is add loads of speckled 'noise' to the image. when you render it down to 8 bits you will notice that the image retains more detail than without dithering. you will notice that you can make out transitions between colours better because the speckled noise adds a stipled effect (lots of tiny dots) making it easier to make out light and shadow. overall, more detail of the original is retained even though its still only described using 16 colours. if you can try this out in photoshop you will instantly see what im describing.
take this idea and transpose it to audio since it works the same way. you apply dither to a 24 bit render before you render down to 16 bit. the low level noise takes the emphasis away from the quantisation noise introduced into the recording by losing 8 bits.
you have to do this right though or it can sound worse than without dithering. i dont even bother dithering any of my tracks. i dont even master them since i think this stuff is best left to a mastering engineer. besides, my production and mixing isnt up to scratch yet so i feel i can get a more noticeable increase in the quality of my tracks by working on them better prior to the mixdown. for most people i would imagine this is the case. i wouldnt bother with dithering until your mixes are good enough where there isnt really much you can do to improve this. i can think of very few artists where this would apply. photek being one of them. but even he leaves that stuff to the engineers at metropolis. |
|
|
| Dj Thy |
Derivative is right.
You might ask how it's possible to cure noise by adding noise. That's because quantisation noise is correlated to the signal, which means it's pretty much cyclic. In other words, it's noise with a recognisable pattern it. And humans are pretty good at detecting such patterns as something unnatural.
The noise you add will improve the matter a bit. Even if the new noise is audible, it's pretty much white noise, which is random, so not as annoying.
Wether it's audible or not is not only a matter of the system it's played on, but also how you listen. You can still detect "decimated" 16 bit audio (meaning the last 8 bits have been truncated, when you start from a 24 bit source to a 16 bit destination). But then again, most youngsters are so used to hearing mp3 (which is pretty much their "standard" audio nowadays, sadly), so the last thing you'll notice between all that's been lost due to the compression is the detail you'd retain by dithering.
The advanced algo's you're referring to (UV-22, POW-r, etc...) are more than dithering alone. They use noise shaping too. In simple terms noise shaping is just applying a filter on the dithering noise. Why? The big idea is to retain the advantages of "pure" dither noise, but filter it (according to psychoaccoustic principles) so the resulting noise is even less audible (POW-r = Psychoaccoustically Optimized Wordlength reduction by the way).
On http://www.24-96.net/dither/ you can find a blind dither algorhythm shootout (it's pretty old, so isn't up to date anymore).
Actually, you can't say one algorhythm is better than the other, because they react differently on different source material. A good mastering engineer usually has different algo's ready, and will test which one will give the best results for that given source.
But derivative gave very good advice :
If you want to risk doing it yourself, you should only dither once in the whole signal chain, and if possible as the last step before putting it on the final media.
If you know your stuff will end up in a decent mastering studio, keep everything at the highest quality possible, without dithering at all. The mastering engineer is a trained person that will know what is best for your tracks, and 9 chances out of 10, he'll have gear that's much better than yours for the job. |
|
|
| Derivative |
| thy are you an engineer? why do you always have to 1up me? :( on the flipside, i just learned something new about dither. cheears for the extra info |
|
|
| Dj Thy |
Yup, I am actually.
Knowledge is power my friend. You can never know too much.
And frankly, I see no reason why I should keep that knowledge to myself. |
|
|
|
|
