IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
Listen to dithering algorithms used out there, "The Great Dither Shootout", link
KikeG
post May 8 2003, 15:38
Post #1


WinABX developer


Group: Developer
Posts: 1578
Joined: 1-October 01
Member No.: 137



http://www.24-96.net/dither/

Here you can download examples of different dither types available at proffesional industry. The samples are amplified so that the dither noise is easily audible. You can also check using a FFT view the different noise shaping profiles used.

SSRC or FB2K dithering is not available, but If I had space enough to host the files, I could upload equivalent test samples to compare, when I had some time.

According to the available pool results, the best dither without doubt seems to be MegaBitMax Ultra. I fully agree with this, even when SSRC strong ATH noiseshaping would be among the best ones. The famous Apogee UV dithering would not wink.gif
Go to the top of the page
+Quote Post
Bonzi
post May 8 2003, 16:19
Post #2


A/V Moderator


Group: Members
Posts: 278
Joined: 22-February 03
Member No.: 5132



I found the most pleasing sample to be the MegaBitMax Ultra also, followed closely by the sample from airwindows which was the ten nines (IIR noise shaping). Nice test though, very informative. smile.gif
Go to the top of the page
+Quote Post
Garf
post May 8 2003, 16:58
Post #3


Server Admin


Group: Admin
Posts: 4883
Joined: 24-September 01
Member No.: 13



I uploaded samples from the FB2K ditherer:

http://sjeng.org/ftp/fb2k/fb2k_strongshape.flac
http://sjeng.org/ftp/fb2k/fb2k_softshape.flac
Go to the top of the page
+Quote Post
Garf
post May 8 2003, 17:20
Post #4


Server Admin


Group: Admin
Posts: 4883
Joined: 24-September 01
Member No.: 13



To me, MegaBitMax Ultra sounds best (easily even), next up is FB2K strong noiseshaper.

A comparison of the two:



Red is MegaBitMax, green is FB2k.
Go to the top of the page
+Quote Post
tigre
post May 8 2003, 17:31
Post #5


Moderator


Group: Members
Posts: 1434
Joined: 26-November 02
Member No.: 3890



[EDIT]No more sample. Garf was faster than me smile.gif [/EDIT]

From the ones I tested I tend to like MegaBitMax Ultra most, but I don't like the added high pitched tone (much more with fb2k). On the other hand I only notice it because I listen at 54dB higher volume than usual, so I think it's no problem in practice.

I think MegaBitMax' advantages are these:
- it's low volume flat dither until ~15.500 Hz, so it sounds somehow natural (like white noise)
- the additional high frequency noise has a steep highpass. It starts at ~15.5 kHz and reaches full volume at 18kHz, so it's almost inaudible and doesn't change the "natural white noise perception".

If you don't focus on annoyance but on volume caused by dither noise, there are other samples where noise seems to be a bit lower, but because it's high pitched and "unbalanced" it's perceived as more annoying.

These statements are my try to explain why certain samples sound better than others for *me* knowing their frequency view. No "objective" facts.

This post has been edited by tigre: May 8 2003, 17:34


--------------------
Let's suppose that rain washes out a picnic. Who is feeling negative? The rain? Or YOU? What's causing the negative feeling? The rain or your reaction? - Anthony De Mello
Go to the top of the page
+Quote Post
tigre
post May 9 2003, 16:38
Post #6


Moderator


Group: Members
Posts: 1434
Joined: 26-November 02
Member No.: 3890



I've just noticed that the samples with noiseshaped dither, e.g. fb2k with strong noiseshaping (see Garfs post) are killer-samples for lame 3.90.2 aps + ape (+ -Y, -Z, higher/lower lowpass). Awful ringing; api is ok. The alphas I tried (a12, 13, 14) are a little bit better, but still annoying. Not very fair/reallistic ... I know.

Anyway, maybe adding this type of noise (or maybe > 16kHz noise) to CDs on purpose will force people to use Ogg/Mpc (gt3b1 + mppenc 1.14 sound ok at q5). tongue.gif


--------------------
Let's suppose that rain washes out a picnic. Who is feeling negative? The rain? Or YOU? What's causing the negative feeling? The rain or your reaction? - Anthony De Mello
Go to the top of the page
+Quote Post
DigitalMan
post May 9 2003, 17:00
Post #7





Group: Members
Posts: 486
Joined: 27-March 02
From: California, USA
Member No.: 1631



QUOTE (Garf @ May 8 2003 - 08:20 AM)
To me, MegaBitMax Ultra sounds best (easily even), next up is FB2K strong noiseshaper.

Would it be reasonable to expect FB2K to switch to MegaBitMax Ultra (or something similar)? Seems in keeping with the high performance focus of FB2K. Not knowing about MegaBitMax Ultra there may be licensing issues, etc., so maybe FB2K strong noiseshaper can be improved to be better(?) than MBMU?


--------------------
Was that a 1 or a 0?
Go to the top of the page
+Quote Post
atici
post May 9 2003, 17:11
Post #8





Group: Members (Donating)
Posts: 1180
Joined: 21-February 02
From: Chicago
Member No.: 1367



What algorithm does the "Dither Output" option of Winamp MPC plug-in use?


--------------------
The object of mankind lies in its highest individuals.
One must have chaos in oneself to be able to give birth to a dancing star.
Go to the top of the page
+Quote Post
rjamorim
post May 9 2003, 17:30
Post #9


Rarewares admin


Group: Members
Posts: 7515
Joined: 30-September 01
From: Brazil
Member No.: 81



QUOTE (DigitalMan @ May 9 2003 - 01:00 PM)
Would it be reasonable to expect FB2K to switch to MegaBitMax Ultra (or something similar)?  Seems in keeping with the high performance focus of FB2K.  Not knowing about MegaBitMax Ultra there may be licensing issues, etc., so maybe FB2K strong noiseshaper can be improved to be better(?) than MBMU?

MegaBitMax is commercial and demands license fees.

http://www.megabitmax.com/
http://www.megabitmax.com/content/license


--------------------
Get up-to-date binaries of Lame, AAC, Vorbis and much more at RareWares:
http://www.rarewares.org
Go to the top of the page
+Quote Post
Xenion
post May 9 2003, 18:47
Post #10





Group: Members
Posts: 1041
Joined: 23-May 02
From: DE
Member No.: 2107



softshape.flac sounds better to me as stronshape has such a high-freq noise sound which sounds annoying to me. but i can be wrong also. just my first impression of those two.
Go to the top of the page
+Quote Post
Garf
post May 9 2003, 19:20
Post #11


Server Admin


Group: Admin
Posts: 4883
Joined: 24-September 01
Member No.: 13



QUOTE (DigitalMan @ May 9 2003 - 06:00 PM)
QUOTE (Garf @ May 8 2003 - 08:20 AM)
To me, MegaBitMax Ultra sounds best (easily even), next up is FB2K strong noiseshaper.

Would it be reasonable to expect FB2K to switch to MegaBitMax Ultra (or something similar)? Seems in keeping with the high performance focus of FB2K. Not knowing about MegaBitMax Ultra there may be licensing issues, etc., so maybe FB2K strong noiseshaper can be improved to be better(?) than MBMU?

MegaBitMax is propretary commercial stuff, but we can probably adjust the ditherer of FB2K to give a similar effect, and I will try this if I have time.
Go to the top of the page
+Quote Post
atici
post May 9 2003, 19:49
Post #12





Group: Members (Donating)
Posts: 1180
Joined: 21-February 02
From: Chicago
Member No.: 1367



I was actually reading about how dither works. I was wondering why we can't achieve a better effect in analog, i.e. after D/A conversion? Apparently the solutions we mention all dither digital files to make the the result more pleasurable but then don't we still have the discretization/quantization problem? Couldn't we achieve better results if dither is postponed after the digital to analog conversion?


--------------------
The object of mankind lies in its highest individuals.
One must have chaos in oneself to be able to give birth to a dancing star.
Go to the top of the page
+Quote Post
Pio2001
post May 9 2003, 20:20
Post #13


Moderator


Group: Super Moderator
Posts: 3936
Joined: 29-September 01
Member No.: 73



It would be too late. Think about a -100 db sine. If you postpone dithering after the 16 bits conversion you are just ditherig digital silence.
Go to the top of the page
+Quote Post
KikeG
post May 9 2003, 21:35
Post #14


WinABX developer


Group: Developer
Posts: 1578
Joined: 1-October 01
Member No.: 137



Just one thing:

In order to make a more meaningful comparison, you must compare the different dither types so that the noise is barely audible, to make a more realistic comparison. In this case, high-pitched noises tend to become inaudible.

If you do so, and compare MegaBitMax Ultra with SSRC/FB2K strong ATH, so that the background dither noise of the first is barely audible, you won't hear the high-pitched noise of the second, and the effect is that this second is less noisy than the first.

So, under these particular conditions, that could be considered to be closer to the real ones, FB2K/SSRC strong ATH wins for me. In practice both are so little audible that none of them would be audible at all in practice.

To verify this try one thing: at the bottom there are links to some generated 16-bit dither samples, without any amplification. Download them or just generate your ones, using FB2K and dithering to 16 bit.

Now, play them. Do you hear any noise? Turn up the volume, so that you start hearing noise. Compare with the silence sample to check dither noise audibility. In my case, even turning up amp volume at maximum, I can't distinguish the silence sample from the (strong) ATH dither sample, because my amp self-noise masks the audible dither noise in this case. This gives an indication of true power of the ATH noiseshaping dither, and gives an idea of audibility of it under real world listening conditions, and of possibilities of 16-bit audio.

Please REMEMBER TO LOWER DOWN THE VOLUME AGAIN BEFORE PLAYING ANY MUSIC, or you'll blow your speakers and your ears.

http://www.kikeg.arrakis.es/dither16bit/di...ither_flat.flac (48 KB)

http://www.kikeg.arrakis.es/dither16bit/di...triangular.flac (57 KB)

http://www.kikeg.arrakis.es/dither16bit/di...dither_ATH.flac (82 KB)

http://www.kikeg.arrakis.es/dither16bit/silence.flac (2 KB)

This post has been edited by KikeG: May 9 2003, 21:44
Go to the top of the page
+Quote Post
floyd
post May 29 2003, 05:49
Post #15





Group: Members
Posts: 630
Joined: 18-June 02
Member No.: 2332



so megabitmax is only available in some program for mac? no windows implementation (other than the time-limited demo)?
Go to the top of the page
+Quote Post
krabapple
post Mar 8 2007, 18:26
Post #16





Group: Members
Posts: 2181
Joined: 18-December 03
Member No.: 10538



The 24-bit original reference file seems not to be working any more. And when I download and open it with Audition, it's silence. The other two reference files look/sound OK.

This post has been edited by krabapple: Mar 8 2007, 18:30
Go to the top of the page
+Quote Post
Kees de Visser
post Mar 8 2007, 23:51
Post #17





Group: Members
Posts: 648
Joined: 22-May 05
From: France
Member No.: 22220



QUOTE (krabapple @ Mar 8 2007, 18:26) *
The 24-bit original reference file seems not to be working any more. And when I download and open it with Audition, it's silence. The other two reference files look/sound OK.
The music is at a very low level. Did you try to add gain (say 60 dB) to it ? Seems to work fine here.
The level is so low that the signal uses in fact only 7 bits (of 16). Increasing the monitor gain isn't quite fair, since dither is supposed to be inaudible, around the threshold of hearing. It can be interesting to find out where your "dither" threshold is. There is probably no need to make the noise perceptually quieter than that.

This post has been edited by Kees de Visser: Mar 8 2007, 23:52
Go to the top of the page
+Quote Post
airon
post Mar 9 2007, 03:37
Post #18





Group: Members
Posts: 31
Joined: 23-January 03
Member No.: 4708



The plugin Ozone has a dithering stage that uses MegabitMax. Find it at http://izotope.com .

You can do your own samples to demonstrate dither, easily.

Take any music, convert it to 24 bit, and fade out a part of at least 30 seconds. Lower the volume of those files by 60 dB. Now create truncated 16-bit and dithered 16-bit versions. Then amplify the files by 60dB again. Listen. Detail in the dithered version, pure shite in the tuncated one.
Go to the top of the page
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 28th July 2014 - 13:43