Stereo Tool

Whoa, that's why the betas sounded off >.<
I thought high steepness caused problems so I always had it at 30%

Hans, why did you change the steepness value from a decimal number is some betas to a percentage?

_________________
<-- I like this guy!

It's a percentage 'maximum difference between BINS' - note, bins, not bands! The difference between bands is at least one bin (bins are 86 Hz at latency 512, 43 at 1024, 21 at 2048, 11 at 4096).

High steepness gives more artifacts. So low steepness should sound better. If you use very extreme multiband behavior (very loud input), you might indeed get weird results with lower settings, especially at lower latencies.

Basically, if there were no artifacts, Steepness would not be there - or alternatively, the value would be set at +infinity.


Are you using such very high Multiband input levels, and/or very low latencies? Because I'd like to know why the sound was off in your case with level 30% - which is higher than the default level (25%)...

Pre-amp is at 18.00 and AGC is at 12.79 latency is 4096.

_________________
<-- I like this guy!

Then Multiband should be doing almost nothing, and Steepness should have no effect. Unless you're using very large Equalizer setting differences.

I posted the wrong settings I had AGC a lot higher(?).
My EQ settings aren't too high, the highest is about 4 DB on the higher freqs.

Would help if I posted the AGC levels, it was at something around 60.

_________________
<-- I like this guy!

just checked beta15 and 512 mode
much better!

But,
- band1 and 2 must be controled! i mean .. there is no control over it and there is bass missing at those freqs.
But again, Ludacris Ft. Nicki Minaj- My Chick Bad, that bass at 30Hz goes too much up with any preset, (checked with radio 0-84)
- In some tracks i have much better bass at 5% steepness. but in other much better at 250%, so best is at 125%
- there is strange mid-bass overshot and it very anoying.
- phaserotator, strange, strange i can't define it.
- how clipper works for band2 and not for band 1 and 3

Much better: Great!

1. There's no way to split band 1 and 2. 43 Hz at 44100 Hz is 44100/43=1024 samples... So with a buffer size of only 512 samples I cannot properly distinguish 43 from 86 Hz. An alternative solution is to increase the total level of band 2, where all the audio of band 1+2 ends up. Which doesn't solve the 30 Hz bass issue you mention..... Alternatively I could add a separate non-phase linear filter BEFORE any other processing that handles very loud freqs below - say - 50 Hz. Non-phase linear filtering can currently only be done BEFORE any other processing, but it could help in this case. (Actually, I could even add a non-phase linear bass filter for the lowest 3 bands......... . Maybe for 6.02 or so)

Edit: Hm. this might actually be a VERY good idea. I already have a 0-latency Multiband filter lying around which I created when I started with version 5.00. It doesn't support clipping, but I could use the current Multiband filter for that. It would effect the effectiveness of the noise gate, but it might be worth it. I would have to put it at the start of the processing (before AGC!), which might be a challenge...

2. Better at Steepness 5%? Steepness SHOULD only be used to get rid of artifacts, I was expecting higher Steepness levels to always sound better if you ignore those... Can you clarify what is better at Steepness 5%? And is that also better at higher latencies?

I still have a 'Loundess' issue on my TODO list that, if I manage to fix it, should allow using slightly bigger Steepness levels with the same level of artifacts.

3. Mid-bass overshoot? So it's too loud? Is that band 3, 4?

4. Phase rotator: Odd... Maybe I need to smooth the extremes and then slightly increase the default settings of the 2 new sliders. (You could for now increase both sliders to get closer to the 'perfect' filter).

5. Clipper. Band 1 is (or should be) empty, so that makes sense. Band 3 clipping sounded really bad, so I turned it off. But I was thinking that maybe I can modify the exact frequency range of band 3 vs. band 4 to get rid of that as well - if that works I can re-enable clipping. It's on my TODO list. Maybe this could also have an effect on issue #3.

i see...
Control over band1 and 2 must exist. Maybe is good to concentrate on 1024 mode then.
Or some tricks with upsampling-processing-downsampling and/or vice versa. Or to join band1 and band2, and band9 and 10.

Phase Rotator is actually good with dedault settings, bad thngs happen when i want to set start freq. at 200 or 150Hz.

btw
In BA Asio, Low Latency mode is non-phase-linear and sound really good with latency 17-20ms depend of soundcard, bit less then ST at 512 mode.

1. Already combining band 1+2, but maybe not correctly (why do you also want to join bands 9+10?)
2. With such values you may need to change the settings of the 2 new sliders. See the difference in dB value that is displayed; if it gets too big you'll get weird effects.
3. Non-phase linear: Probably a good idea, especially with my new idea (combining phase linear with non-phase linear - in this case a non-phase linear Multiband, with a phase linear Multiband for the remaining differences.)

Here is already was talk about how much bands is needed in MB, especially for highs.
My opponion is a litlle different...
In 512 and 1024 modes highs are different (does not mean bad), bit more more crispy. It much more sounds like big boxes. orban, omnia, vorsis... etc..
If i look at all this processors, at high freqs all of them have mostly two and sometimes one band for highs, where ST have 3 bands.
I guess there is a reason for that. I think it will be much better control over highs, mean, more natural for different sound format.