Stereo Tool

Well, you tell me if this is a valid test or not:

I've gone back to BETA022. I have Pre Amp at maximum. Post Amp is at x 1.00. Only AGC and multiband are on. AGC is set to -6 Target Output. Remove Remaining Peaks at 150% for Band 1, 133% for band 2. I'm using my posted HypersonicV7 preset bands in multiband, and have clipping set to 0 (full clip) for everything but the first 4 bands (35 Hz, 75 Hz, 110 Hz, 220 Hz). Multiband is set to Compress, which it needs to be for bass (read up on it if you're not convinced). I have a 25% steepness (I think that's the default), no band linking, and no vocal protection.

With those settings, when the bass guitar hits especially hard, there is distortion and some artifacts.

If I switch off multiband, bring the preamp down to x 14.00, and use the Bandpass set for 0 - 400, there's all sorts of crackling. If I enable Hard Limit, the crackling goes away.

I did the same test with BETA034, and got the same results.

So, the questions are:

1) Is this a valid test?
2) If yes to 1, just how well does multiband actually handle high input levels?

Hm, interesting.

The test with only bandpass does more or less what you would expect - without Hard Limit and with Pre Amp at 1.00x you'll get major overshoots which sound especially horrible if there are only low frequencies present. So that behaves as it should.

But, in the multiband test I hear lots of very weird artifacts that I didn't expect. Although I think I can explain those as well: By setting the clipping level of bands 5+ to 0, you create a VERY steep filtering slope. The Prevents Artifacts/Steepness slider was NOT designed to cope with this type of extreme situations (it is if you use the equalizer or soft limit values, NOT if you use insane amounts of clipping). If you just slightly increase the clipping level for band 5 (even if you just set it to 0.05 or so), the sound immediately improves a lot. So I think this is not really a problem - unless these artifacts are also there under normal circumstances (and since they already almost disappear as soon as I increase the band 5 clipping only very very slightly, I don't *think* that that's the case). --- EDIT: Ow. If I turn that slope thing on, I still hear it. Oops... Something is not working as it should here!

I have not yet noticed distortion in this case - or are you talking about overshoots in the ouput (which can be solved by turning on Hard Limit)?

I have no idea if you're agreeing or disagreeing. Seems like agreeing at this point. I'll wait a while and see. Will do more tests later today based on what you said before the edit.

Although... bear in mind that it is possible that these "volume controls to 11" folks are pushing the signal to levels that you may not have tested with, and thus the distortion that might otherwise be "expected" and "masked" ends up being louder and not able to be masked.

Hi Brian,

1. The artifacts that I *do* hear when I do what you said are coming from an artifact cleanup algorithm that appears to be too aggressive in this case. But since I've *never* heard these artifacts when using normal settings (well, not since about a year ago when they were actually reported to me by Bojcha, and I changed the processing to fix them) I'm going to assume for now that there's not really a problem.

Last night I have been chatting with several people about these voice distortions, and what it seems to come down to is that there are 3 types of people:
- Some people hear them very clearly
- Some people hear them very clearly, but if you invert the signal they don't hear them AT ALL anymore (*)
- Some people just don't hear them at all

(*) Naturally, when inverting the signal there must be other places where they start to occur where they didn't hear it before. Not yet confirmed.

This was all tested on different equipment (Bojcha even exchanged both his sound card and his headphones to make sure).

So basically, some people hear it, some people hear it half of the time and some people (including me) just don't hear it at all. People who do hear it hear it clearly and find it really annoying. And they don't hear it in other processors.


My current theory is that whether you hear it or not depends on how your brain processes audio. And apparantly there are different methods to cope with distortion that is created by the ears (that also explains why you can hear some forms of distortion much better at a very low volume, but that also doesn't seem to be true for everyone).

Anyway, the good news is that I think I know now what is causing this: Asymmetrical bass. (I've seen it in the audio that I've received so far. And it has to be, otherwise inverting the signal would not have any effect.) And that's often improved by using the non-phase linear highpass filter, which matches the reports so far. Solving it should be pretty simple. Unfortunately I cannot test the solution myself because I just don't hear the distortion (well I can check if the waveform looks better, but that's it).


Short cut from last nights chat:
[2:20:00] Hans van Zutphen: If you listen to sounds with distortion and you increase the volume, you don't hear the distortion anymore. My guess is that - since the ear only contains one thing that receives all the frequencies, if there's a loud low frequency the mid and highs get reduced also, it would make sense that the brain masks this effect.
[2:20:48] Hans van Zutphen: Actually, it would even make sense that different brains use different masking strategies

Hi Hans,

And of course the comparison was made ​​with the same preset.

the cursor is at the maximum quality

I do not use the "bass agc"

I tried to lower the frequency of the bass clipper 40-60-80-100 Nothing doing, the sound is always too much saturated and low frequencies vibrate (as you say it is definitely a problem of asymmetry)

I also tried to lower the level of the limiter 1 to 0.80 but the result is the same, low saturation and vibrant (of course this happens only and exclusively on recent recordings, with strong processes of limitation)

and what strange decrease if the "preamp" saturation increases.

In fact the problem is not the distortion, which as you say, but I think there is, but almost certainly is a problem of asymmetry.

I hope you manage to understand my "google english"

best Regards

Ok... Since I don't hear this issue for some reason, can you check if this probem is also there in BETA032? That's just before I changed the bass clipping change - which now allows MORE of the higher bass freqs to come through. My guess is that the change happened between 032 and 035 (33 and 34 are bad).

Hi Hans,

Tested the beta 32, I tried to lower the frequency of the bass clipper 40-60-80 Hz, the result is always the same, low vibration and overall quality deteriorated.

beta 23 on the air again.

Best Regards

Agreed that it's bass...
... however, NPL introduces phase changes, which you previously said were part of this problem. The other presets posted have Phase Rotation ENABLED.

Make sure you're taking their whole processing chain into account.

In my opinion, it is simply a matter of too high input going into multiband.

However, if you are determined to do something, the question I'd ask is how far away do you think that a redesigned multiband would be? If it is 8-12 months or more, then yeah, if this change is quick and easy, then might as well. If, however, a redesigned multiband could be had in 2-4 months, then it may not be worth spending the additional time on...

The new multiband won't completely solve it (Bojcha already checked by using another multiband compressor). But... The good news is that I've made a few changes and according to 2 people who hear the effect it's improved now. So I know what to do, and I expect to have a fix tomorrow or Monday.

Actually it's really simple: If the bass is asymmetrical, I need to clip it a bit harder.