Apple Event: May 7th at 7 am PT

Looks like no one’s replied in a while. To start the conversation again, simply ask a new question.

Dither screwed up in WB 1.6

For all the posts about plugins not working and so on, what I've found is that you can't do anything in WB that would require a bounce or any kind of dither because that is broken. If you are putting files into WB and running plugins on them, then dithering (which you need to do for any kind of DSP), you are actually creating CDs of truncated 16 bits, I'm sorry to say.

I found this out because I wanted to hear what the different Powr algorithms sounded like, so I tried importing a -75dbFs 24 bit sine wave into WB and bouncing it with dither. Then reimporting it into Logic and putting enough gain on it to hear what's happening.

My experiments show that it turns into a nasty square wave or something similar when bounced.

Doesn't matter what the dither settings are.

I created a 16 bit dithered file of that same sine wave in Izotope RX using Mbit+ dither and burned it to a CD in WB directly without bouncing, then reimported it into Logic.
I heard a sine wave amidst a lot of noise. This is what it should sound like.

Did the same thing using Barbabatch. It's dither wasn't as good but it was basically the same thing.

Did the same thing with a bounce in WB, and it was corrupted. Night and Day. A buzzy wave with no noise at all. The dither doesn't work. It's just truncating the audio from what I can tell. I tried all the dither settings in the preferences I could find. They don't work.

In fact, even if you bounce a predithered file without doing anything else to it, it will get corrupted. Just the act of bouncing truncates the lowest bits.

You can't do mastering and you can't use plugins on files going to 16 bit without dithering.

The only way this software can be used is to import finished and dithered 16 bit files and not do ANYTHING to them. No bouncing of any kind. No plugins, no fades, no level adjustments. Nothing except ordering, spacing, and hitting the burn command.

Only under those conditions does it not corrupt the audio in the lowest bits.

Try my experiment if you don't believe it. You can create a low level sine wave with the test oscillator in Logic.

Hopefully this will get fixed.

MacPro 2.66 Quad, 11GB Ram, (3) WD Raptor SATA drives, Mac OS X (10.5.8), MOTU 896mk3, UAD-2 quad, (2) Unitor8s, MCcontrol/MCmix, KORE, Melodyne

Posted on Oct 16, 2009 11:34 PM

Reply
26 replies

Oct 20, 2009 5:55 AM in response to danseq

Fade tests not withstanding..

Here is a SIMPLE but highly accurate and informative test I performed by importing a track into WB, and bouncing it out both with and without dither.

Each of these files were then reimported into logic and aligned with the original (24 or 16 bit, I tried both, same results) source file, polarity inverted to null them and checked using "RMS Buddy" across the output bus. No level shifting required.

http://destroyfx.smartelectronix.com/docs/rms-buddy.html

This is a FREE plug-in meter. It will show numerical peak and RMS levels with great accuracy and even has an adjustable integration time for the RMS detector.

I only get 90.3 dB of null on the WB bounced files, both with and without dither.

The same test using Wave Editor resulted in more that 96dB of Null.. theoretically perfect.

I 'd like to see what each of you are getting when using this method. It's MUCH less picky than creating fades, etc and every bit as revealing (no pun intended).

-Chuck Zwicky

Oct 20, 2009 8:26 AM in response to danseq

danseq wrote:
Thanks for checking this. timrob.


Also very curious to see if it sounds the same in 10.5.8

Can you repeat with sine down 20 dB +/- from there? If it's clean then it's either something in my system or 10.6.


Yeah, it occurred to me that I should have gone with a lower level. I'm tied up mixing the rest of the week. Might have a few minutes here and there to try some more tests.

Sure wish someone from Apple would participate in these discussions.
I would think that would be the point of hosting them in the first place.

more later.

timrob

Oct 20, 2009 1:36 PM in response to zmix

zmix wrote:


Each of these files were then reimported into logic and aligned with the original (24 or 16 bit, I tried both, same results) source file, polarity inverted to null them and checked using "RMS Buddy" across the output bus. No level shifting required.


Thanks, Z. Will DL that.

Question...Do you just visually align them? One of the problems I've had is that in comparing the original files with WB bounces I find that WB seems to add somewhere in the neighborhood of 10-20 ms of space at the beginning of the new bounced regions (I haven't measured it exactly), making phase alignment trickier when trying to do null tests. Has this happened to any of you?

I'll try this metered null test when I get a minute and see what happens.

Oct 21, 2009 8:24 AM in response to danseq

danseq,
I've done some testing this morning using a -75db sine wave. Now, I'm getting results similar to yours.
Even with Ozone plugin on the mix using MBIT.
So, now I'm a believer. WB 1.6 does appear to be truncating, even when dither is applied via Plugin.

I suspect they are writing 16bit file before applying dither and possibly before applying plugins. Don't know that for sure, but it would make sense if that was the case.

It is a shame. WB has the potential to be a premiere mastering app, but continues to be mired with bugs. I don't understand it. I mean it would be one thing if there were only a couple of gotchas, but there are several.

Now this is all using Bounce Project. I haven't tried using Burn. But I wouldn't expect results to be any different.

best,

timrob

Oct 21, 2009 12:35 PM in response to timrob

Thanks Timrob.

My theory is that:

1. People complained about the bit bug

2. Apple went and tried to address it in ProKit 5.0 but missed some critical step.

3. People downloaded PK 5.0 and thought the situation was fixed, because they did some basic null tests but if your theory is correct and truncation is being applied first, then dither, then I can see why people would think there is a null. Because before there was no dither, and now there is--of some kind at least (this I can attest to). But if so, then it is a corrupted waveform being dithered, and the null is I guess better than before, but I bet it isn't to -96 as zmix observed.

Whatever the case, something still isn't right.

I suggest this be disseminated because a lot of people are under the impression that this situation has been fixed.

Oct 21, 2009 12:32 PM in response to danseq

danseq wrote:
zmix wrote:


Each of these files were then reimported into logic and aligned with the original (24 or 16 bit, I tried both, same results) source file, polarity inverted to null them and checked using "RMS Buddy" across the output bus. No level shifting required.


Thanks, Z. Will DL that.

Question...Do you just visually align them? One of the problems I've had is that in comparing the original files with WB bounces I find that WB seems to add somewhere in the neighborhood of 10-20 ms of space at the beginning of the new bounced regions (I haven't measured it exactly), making phase alignment trickier when trying to do null tests. Has this happened to any of you?

I'll try this metered null test when I get a minute and see what happens.


If the bounced file is offset from the original source, I suggest using the "sample delay" plugin to align the files. Which ever file is earlier gets the delay. Apple claimed to have instituted sample accurate waveform editing in the arrange page, but as always, best to trust your ears.

Oct 21, 2009 2:20 PM in response to danseq

danseq wrote:
Hmm... I haven't even tried Pounce Project recently. I've been doing Bounce Region for these tests.

I'll try some more tests to see if I can't match your findings.

If you can try bounce region and also see if it's adding the small offset, that would be great.


I'm not getting any offsets from any of the bounce functions.

timrob

Dither screwed up in WB 1.6

Welcome to Apple Support Community
A forum where Apple customers help each other with their products. Get started with your Apple ID.