UV22 Creating Harmonics when Pushed

I made a 1khz stereo sine wave at 0dB with the signal generator and put it in a montage in order to raise the level above 0 to see how soon I could hear it clipping on my system. Not much it would seem, it’s pretty audible when it starts clipping. Something strange happened when I put the UV22 dither in though. I noticed that even if I was raising the level not far above zero by even +0.03 dB that the UV22 was creating obvious harmonics on the Spectrometer, and by the time I had raised the level +0.5 - +1.0 dB were well into the audible level range. I tried the same thing in Wavelab 7 with the same result so I guess it’s probably always been this way, but it doesn’t happen at all with the Internal or Izotope dither plugins with levels pushed, or with the UV22 with levels at or below 0 dB.

Wow, sounds like a wild Friday night :slight_smile:

Is it Friday already?

That is interesting though. I don’t think I’ve ever used the UV22 dither on a project. I’ve always used the iZotope MBit+ and more recently have been using Goodhertz Good Dither.

UV22HR Dither 1KHz +0.5 dB.png
Wavelab Internal Dither 1KHz +0.5 dB.png

Pretty bad, right? Even if I raise the level one-hundredth of a dB (+0.01) into the UV22, the harmonics start to show. Conversely I can raise the level into the Wavelab Internal dither by 10 dB and it still looks perfectly clean.

Wow… even though they output lower bit rates, I wonder if internally that particular plug-in is operating with 24-bit fixed processing so clipping occurs instantly above 0db, while the others may be 32-bit floating point processing with tons of headroom… interesting.

Uh oh… something even more bizarre. I think we’ve found a BUG.

  1. Remove all plugins from the master section
  2. Load the signal generator plugin in the 1st effect slot, hit space to enable with max level, master faders set at max = should overload to +10
  3. Load UV22 into the 1st dithering slot, notice lots of harmonics appear in the spectrum analyzer
  4. Drag UV22 into the 2nd dithering slot, NOTHING changes - as expected
  5. Drag UV22 back to the 1st slot - nothing changes
  6. Load MBit+ into the 2nd slot - nothing changes
  7. NOW - drag the uv22 into the 2nd slot (so the uv22 and Mbit+ change places) - THE HARMONICS DISAPPEAR
  8. Remove Mbit+
  9. From now on, UV22 no longer adds the harmonics seen prior in the spectrum analyzer
  10. Try dragging UV22 to different slots - it doesn’t matter, THE HARMONICS ARE GONE
  11. Unload UV22
  12. Reload UV22 into one of the dither slots - the harmonics re-appear

Wow, how’d you do that so fast??? You’re a genius!

You’re right, it does exactly as you say, and if you look closely after that, you can see the UV22 is doing some noise shaping above 20 KHz, like it looks when the Izotope or Internal are doing noise shaping.

And if you put the 1k file In a montage with some handles and render it, you can see the UV22 dither is there at about -80 dB

This is a strange one for sure! The honor is yours Bob for noticing the issue! I’m curious what PG will find is the cause… I don’t know if it’s a bug in the plugin, or a bug in the Master Section, but there’s a bug in there somewhere…

Thanks for finding the answer so quickly Toader. That was really amazing!

Thinking what might have been the effect of this on anybody’s master, I would tend to say it’s probably had no effect whatsoever. It doesn’t happen at all with signals at 0 dbfs, and I think most masters are probably not exceding that going into the dither, unless somebody’s intentionally overloading the output because “it sounds ok”. But in that case it sounds ok. In other cases where there isn’t a brick wall and peaks might occasionally be sneaking by, I can’t imagine such short infrequent things would have an audible effect. But that’s just my untested opinion, and might not cover all real world scenarios.

Thanks for the report. The bug is in fact: when you reorder the 2 dither plugins, the UV22 is moved just before the Master Section gain sliders. Visually you don’t see that from the Master Section panel, but this is what happens “internally”. Hence your observation. This will be fiex in 9.5.40.

When the plug-in is before the sliders, there is no harmonic distortion, but after, there is. Why is that? I’m just curious about the behavior of the plug-in… the Mbit+ does not show the same behavior…

I guess this plugin truncates to 0 dB, hence the distortion. When the level is not truncated, there is no distortion to display in the spectrum meter, because the spectrum meter does not truncate to 0 dB.

Thanks PG!

Yes, thanks PG and Toader.

PG, so the plugin reorder issue will be fixed in 9.5.40, is that correct? But the UV22 will still be the same, and will continue to truncate to 0 db? I think the UV22 in Cubase does the same thing, but somebody would have to check with the latest version.

…which illustrates why most of us left UV22 a long time ago. MBit+ is good, I use Toneboosters’ one, while TPDF Dither from Airwindows is a good free one.

Looking forward to the reorder fix (and whatever else gets updated) :slight_smile:

r,
j,

so the plugin reorder issue will be fixed in 9.5.40, is that correct?

Yes

But the UV22 will still be the same, and will continue to truncate to 0 db?

Yes, but this is somehow normal, as this is a dither plugin that change the resolution to eg. 16 or 24 bit.

Philippe

Would the condition then be better described as “UV22HR will continue to create harmonic distortion when input levels exceed 0 dBfs” ? Since the Wavelab Internal dither and Izotope dither plugins don’t.

Yes, but whatever the dither, there will be truncation anyway when rendering to a 16 or 24 bit PCM file.
I mean, even if a dither plugin does not truncate, the file writer will.