Stereo Tool

With Hard Limit on, you should never see spikes above or below these 0 dB lines. They can touch it (ie. overwrite the darker line) but should never surpass it. If you disable Hard Limit however, you will see that it clearly does. (Unless you set Strictness very high).

Find a combination that doesn't let too much peaks through. Increasing 'Strictness' increases CPU load, increasing 'Stronger clipping' does not but MAY *very* slightly hurt audio quality.

It has always been this way! The reason for allowing different speeds in band 3 is to remove brief excessive highs (which were mainly an issue in older versions of the FM clipper and multiband; it could probably even be removed now). Band 1+2 are normally locked together - they should not change the way the audio sounds, just keep the volume constant.

AGC squashing? That should be clearly visible in the AGC meters!

By the way, for streaming this isn't too relevant since lossy compression will add large peaks anyway. So, you can easily get away with using a low Strictness value and turning Hard Limit off. (If you keep it on it will hurt the sound quality if there are big peaks, Hard Limit was made to remove small spikes only).

Chris, do you have some tracks where this happens (plus the settings that you used)? It's one of the last things I need to fix, but I need to hear it first! (If possible, could you re-check that this is really an issue?)

"Should" is a different thing than "will".

Attached is the latest refinement of Cobalt. Yes, it is still using the deprecated features, so I also am attaching the same preset with the deprecated features (classic multiband and classic singleband) turned off. Also, just in case, the non-deprecated version does not use the MP3 preemphasis. In both instances, I can clearly see the blue peaks going over with Hard Limit turned ON.

I again believe that the data you're using to build that image is being obtained BEFORE at least Hard Limit, and possibly something else. The reason being is that TTDR Offline always shows a peak of -0.45 dB (post amp), and the output bars do not turn red.

Edit: Almost positive that what's going over are mids and/or highs. I can see some things that are squared off below the line that appear to be bass, possibly advanced clipper output.

A great track to check for surpassing the lines is DJ Orkidea - Metaverse (Gareth Emery remix)

It is visible. If you use the presets I posted, you can see band 2 getting hit pretty hard with the spike protection. I think that down/up speeds and clipping in multiband would give more refined control over band 2, but I could be wrong.

As for the linkage, I went back and installed the older versions, and to you it may have always been that way, but for me it wasn't clear. There were no labels on the sliders before, and so my brain had filled in that the right-hand slider was band 2, and that band 3 was linked to band 2, and the degree of linkage of band 1 to band 2 was controlled by the other sliders. This interpretation was further reinforced by the "remove remaining peaks" checkboxes. Combine that with the lack of labels...

At any rate, I think it may be good to see what the sound would be like if the spike protection was decoupled and you could choose a different spike protection level for band 1 vs. band 2, and possibly even for band 3, if you still keep a band 3. This would allow a linked or non-linked spike protection level, including perhaps no spike protection for one band, but spike protection for the other. It would allow experimentation to determine if a blend of spike protection that was combined with nuances in multiband gave a better sound.

Yesterday I had some expirience with Orban 5500 FM and settings is easy
But I only compared some default presets, I noticed something common for most MB coupling:
1->2 0%
2->3 0%
3->4 30-50%
4->5 80-100%

PEQ (not sure):

257Hz Q3.3 2-3dB
4100Hz Q3.3 1-2dB

_________________
Urban Junglist

What you see (i think) is highs peeks. At output scope they are visible if pre-emph is used (10-30us), and not visible if pre-emoh is not used. By my logic it's vice versa, and Hans can explain you why is that.

I'm sorry, I forgot something in this discussion. The output DISPLAY shows upsampled output - if you don't enable upsampled clipping (in FM modes it's automatically enabled, in non-FM modes you can choose), you'll indeed still see spikes.