Stereo Tool

Issues 3 from the todo list is fixed. Only if Strictness is set at or above 6 (for lower values the old method seems to work better. New idea: Combine both )
Filters moved from Clipping to FM (note: You need to set them again, old preset values don't load anymore!)

Stand alone: http://www.stereotool.com/download/ster ... 20-011.exe
Winamp DSP: http://www.stereotool.com/download/dsp_ ... 20-011.exe
VST: http://www.stereotool.com/download/vst_ ... 20-011.dll

Note: "Performance: Not low freqs" and "Traditional bass/mids clipping" have a very similar effect, and I should probably remove one of them. For now, keep them both OFF to have composite clipping on ALL frequencies, turn the Traditional setting ON for standard clipping on bass/mids (but at a slightly lower level to compensate for expected pilot/RDS levels). You can also turn 'Performance' on to do even more things in non-composite mode; enabling both gives the best performance.

IMPORTANT: To get a good sound from the composite clipper:
1. Enable it It's an option under Advanced Clipper
2. Under Advanced Clipper -> CPU, set 'Strictness' at 11 or higher (higher is better; there's probably not much gain above 15 anymore though; 11 looks pretty good already so a bit lower values should still be ok).

Warning:
You will need a license update to use this! Otherwise the 'unregistered options in use' thingy will pop up, and you'll get messages through the music (every 4 hours).

DONE:
1. Move or copy (same as Stokkemask filter) composite clipping and SSB settings to FM section.
3. Perform same performance optimizations in traditional clipper; remove 'Stronger clipping' slider once that's finished.
6. Replace DIV_PS by RCP_PS + MUL_PS in new clipper
4. Check bass and mid clipping issues reported by Bojcha.
5. Check clipping bug when clipping a 20 Hz bass tone and set bass clipping (in code) from traditional to MPX clipping. Why is the output level higher??
Ad. 4: Clipping is set to 'traditional' clipping for both bass and midbass. But levels are not adjusted to make room for the pilot and RDS signals. Also, potentially stereo bass will not be clipped optimally. However, if I want to use composite clipping for the (mid)bass as well, I need to fix a bug (currently multi-threading gives different output levels, I don't know yet why), and bass clipping strictness must be greatly increased, leading to a much higher CPU load - so this is probably not a very good idea. The simple solution is to adjust the clipping level to make room for the MPX signal. (Stereo pilot 9% + RDS 4.5% means that there's 13.5% headroom needed, so for default settings that means the clipping level must be multiplied by 86.5%.)
7. Check holes in center of bass -> might be better with new clipper --> Checked, little difference (some are better, some are worse).

TODO:
2. There was a complaint in the beginning of this thread that there were 'shrill' sounds heard due to the changes in highs clipping. That's still not resolved.
9. Update settings (default strictness values etc.) Jeroen Platenkamp's presets strictness can be lowered from 13 to 8 and it's still stricter than before
11. Red output bars
12. SSB mode: Red output --> Not yet thoroughly tested, but should be fixed now. Added 0.01% headroom. Twice that for SSB mode. 1% for Hilbert mode (and that's still not enough - but you shouldn't use it anyway).
13. SSB mode: Check difference multicore/single core. --> Was pretty bad, now ok.
14. Fix behavior: When Composite Clipping enabled use normal clipping if FM or FM stereo coding is not enabled.
15. Maybe make separate Strictness slider for Composite???
16. GUI crashes. This is new... Needs to be fixed.


REMOVED:
8. Interesting observation: For Strictness settings below 7 (9 internally), the old clipping method works better than the new. So, it might be interesting to try to combine the two - that would again result in less steps. Not sure though what will happen to quality (I think the new clipping method is slightly better than the old one for that). --> Oddly, combining the 2 methods causes bigger spikes (overall the level seems slightly lower, as expected). Giving up on this one...
10. Since the old method works better for traditional clipping in low Strictness modes, is that also the case for composite clipping? --> Unlikely. And I implemented the new method because the old one didn't work well.

Stronger clipping slider is still there, even with strictness set to 15.

No change in CPU usage for non-isolated filter testing at strictness 7.

I know, I might even keep it there since using both methods will make it work better.
Good! I feared for a small increase actually. Thing is, you can use a lower strictness value than before for a similar output. Which does help with CPU load.

Then the recommended settings / general usage guidelines for both sliders need to be updated. Not everyone will use scopes to check the waveforms. FM has the other stuff you've been working on. Web streaming folks may or may not have the technical expertise to do waveform analysis.

I will change them, but I have some more changes (as I wrote before) that may help a bit more.

You don't need a scope; just turn off Hard Limit and look at the output waveform. Lower Strictness causes bigger spikes in the output. Hard Limit removes those.

Ok.
I was more focused on the "Stronger Clipping". From what you've said, higher strictness is better than stronger clipping, but still, if you think about it, that's vague. I had been using strictness 6, stronger clipping 10, but have changed to strictness 7, stronger clipping 8. I truly have no idea if that is optimal or even what I'm doing with those settings. If you're talking about the "output waveform" in your panel, then looking for spikes is quite difficult due to the smaller size, not to mention that a lot of current material has a fairly high base RMS level, thus having a lower dynamic range, which means spikes are somewhat less visible.

On another different, but possibly related topic, why exactly did the AGC band up/down controls get changed from Band 1 and Band 2+3 to Band 1+2 and Band 3? I'm still trying to work on the slap sound in Thriller, and I'm starting to wonder if it isn't the AGC squashing the 1K-3K range due to the linked 1+2 speeds and the spike protection.

Please make sure to read and/or comment on the AGC change in my previous reply, but I had further questions about "output waveform".

When I said "smaller size", bear in mind that I do not typically maximize the GUI. I typically don't even run with the GUI up, because it has an additional impact on total CPU load. Now that I maximized, and that I'm paying attention, there are light horizontal lines at the top and bottom. Are those intended to be spike indicators? If not, do they represent -1 dBFS or some other value?

0dB lines.

If that's the case, then that waveform must come from data BEFORE Hard Limit or Hard Limit is not entirely effective, because I routinely see small peaks that are above the line. My personal opinion is that it's based on data before hard limit, because when I run TTDR on a wav file output, there is a "hard limit" involved, which is that the peak level is limited to exactly the value set for postamp.

Oddly, this turns out to make things worse instead of better. So I'm removing this one from the todo list.