Stereo Tool

Hans, this is specifically intended to you. How do you (or did in the past..) calibrate multiband settings? Specially soft limit and individual clipping for each band.
Is there any methodical approach? If so could you please let me know unless it's not too trivial...
Thanks in advance.

_________________
visit website

Sorry I am still not sure if everything works the way it should (especially realtime listening versus file rendering):
viewtopic.php?f=15&p=8146#p8146

Is it a problem of Stereo Tool ?
Or is it only a problem of my limited understanding ?
Phoenix, what do you think ?

See the new help page, http://help.stereotool.com/6.00/multiba ... iter.shtml , section about measure & show median. That's what I use.

BETA601-012:
- Most of the performance loss for HIGHER sample rates is restored.
- Upsampling and downsampling quality has been improved again! So output is again cleaner.

Winamp DSP plugin: http://www.stereotool.com/download/dsp_ ... 01-012.exe
Stand alone version: http://www.stereotool.com/download/ster ... 01-012.exe
VST version: http://www.stereotool.com/download/vst_ ... 01-012.dll
VST version (No SSE2): http://www.stereotool.com/download/vst_ ... 01-012.dll
Command line version: http://www.stereotool.com/download/ster ... 01-012.exe
Linux command line version: http://www.stereotool.com/download/ster ... ETA601-012 [not available]
Linux GUI version: http://www.stereotool.com/download/ster ... ETA601-012 [not available]

What remains for 6.01:
- Performance: Multiband steepness: Moving UP can be done inside current loop, no separate loop needed
- Performance: Multiband: Remove sqrt(sqrt(cos())), pow(x, .75) etc. - too expensive, replace by lookup table.
- Reduce downsampling frequency because currently very high frequencies (21-22 kHz @ 176.4 kHz input sampling rate) in the input can cause spikes even if Hard Limit is used.
- And I need to check how much the performance is impacted by the latency improvements, in the case where no upsampling and downsampling is needed. This seems to be impacted way more than I expected (could also be Multiband steepness).
- MAYBE: Make Steepness smarter. That would much better preserve the audio, especially at very low latencies! How: Instead of setting all the levels at AT MOST the level of neighboring bands + a bit, combine adjacent bands, determine total output level, and then fix it such that this combined output level is approached more. (So one very low, one very high --> one a bit less low, one a bit less high instead of both low).
- Fix NOISE GATE behavior in VST plugin ???

Questions:
* Multiband: Question: Is Steepness behavior ok?
* Loudness: QUESTION: The changed Punch behavior, is that good or bad?

Check:
Check difference in behavior between 44.1 and 48 kHz input for multiband! This could potentially result in really big differences.

Hi Chris,

last, is that the current or the previous (which I called 'bad').

At higher latencies, you should see very little effect on higher bands! Basically, high Steepness values make multiband more effective, but can lead to artifacts, especially at lower latencies and lower frequencies. That's why the filter has the most effect in those situations.

The last one Hans... it had more affect on mids and hights and made the sound more clear.. my opinion.

_________________
secret profiles of sexy girls - https://privateladyescorts.com - erotic connections online

Ideally 'steepness' should have NO effect at all on the sound... It's just there to prevent artifacts


Some good news about low latency processing: I've found a better method of protecting low latencies against distortion. This is, with far less (expected - still need to test it!) effect on the audio.

In case anyone is interested in the maths:


Say there are 3 adjacent bands, with maximum amplification level difference between bands (Steepness) of 25% - at bands with the maximum amount of artifacts (so the level difference really should not exceed 25%).

Say the suggested amplification levels for those bands are: .9 .1 .9

CURRENTLY I use the lowest value to limit everything, so if the 2nd band is 0.1, the adjacent bands cannot get higher than 0.125 (25% extra). This means a HUGE loss of adjacent frequencies (which is usually bass, because for bass frequencies the artifact level is the highest): 0.125 0.100 0.125

I could also have used the maximum level instead, to avoid bass loss - but then protection against loud tones does not work at all: .9 / 1.25 = 0.72 --> 0.900 0.720 0.900

Combining maximum and minimum also doesn't work too well, (.100 + .720) / 2 would still be far too high for the center band.

NEW calculation: 2 / ( 1 / min_i + 1 / max_i ).

Gives:
min_i: .125 .1 .125 (reciprocal = 8.00 10.000 8.00)
max_i: .875 .7 .875 (reciprocal = 1.142 1.429 1.142)
Sum of reciprocals = 9.142 11.429 9.142
2* reciprocal = .219 .175 .219

So the center frequency is in this case 75% louder (which is not too extreme), and the frequencies around it are also 75% higher -> resulting in 75% more bass!

For a more likely situation:
.7 .3 .7
min_i: .375 .300 .375 rcp = 2.667 3.333 2.667
max_i: .700 .560 .700 rcp = 1.429 1.786 1.429
Sum of rcps: 4.096 5.119 4.096
2* rcp = .488 .391 .488

Which is much better than the currently used min_i values.


Let's say the input levels are also the reciprocals of the amp values (1/.7, 1/.3, 1/.7).

Then that results in (intended behavior): .7/.7 + .3/.3 + .7/.7 = 1+1+1 = 3

With MIN values that's: .375/.7 + .3/.3 + .375/.7 = .54+1+.54 = 2.08 - less.

With new calculation: .488/.7 + .391/.3 + .488/.7 = .69 + 1.30 + .69 = 2.68 - still less, but much closer to the intended 3.00.

Thanks for indirectly answering...

Be careful that you're not getting too wrapped up into your own coding. Many times, as a single developer works on something, they become convinced of the behavior of the application being a certain thing. A different set of eyes on the situation can sometimes reveal things that the original designer may not have thought of... In this case, if we know what you're aiming for, those of us with decent hearing can help, even if we don't have fancy meters...

So a lot of time needs to be devoted to go by this approach. I would rather stick with the default settings that you provided. I like this setting. Nothing inherently wrong with it. It's just that I had been trying to replicate a station's signature. It's a new one (92.7 Big FM) in the city and using Omnia 6 EX FM. It's also the loudest. Stereo Tool sounds very clean and fuller as compared to the former. It's difficult to describe the station's sound, but I can say the highs are overtly bright(to the point of getting irritated) and very dense. But then even at that loudness it substantially preserved the bass kick. I had a tough time trying to get that in Stereo Tool but eventually did (ofcourse, barring the dense highs).

I will upload the preset in a short while along with an improved version of the Pristine preset.
On an intuitive note Hans, I might like to suggest to extend the range of Bass Boost Strength. You can achieve wonderful results at competitive loudness. Just a thought, entirely your call. Please don't get baffled!

Uploaded Now! Please refer to : viewtopic.php?f=10&t=2804&p=8172#p8172

_________________
visit website

This explanation helped a lot!
Using the new preset now and everything sounds good!
No distortion or artifacts and 4096 works perfectly.

Left the steepness at 25% in the beginning but turned the down speed up and I saw what you added to the multiband.
In the lowest 3 bands I saw a lighter version of the meters, exactly the same kinda I saw on the output band when I enabled pre-emphasis.
Is it ok to see this in 4096 mode though? I turned up steepness to 75% so its not there any more.

_________________
<-- I like this guy!