Stereo Tool

Beta091 - reduction in load displayed in :gasp: Task Manager (unreliable, yadda yadda, I know, I know, but comparitive measurements on the same equipment are statistically relevant), from 75-80% CPU to 68-75% (both shortcuts checked, Advanced Highs disabled), so relatively little change, and certainly not enough to make a difference on usability for anyone below a Core 2, which includes P4, the entire AMD K8 line (including the X2 and Opteron), and the first-gen Core. The oversampling is a severe penalty for older processors. Not sure I can even hear any fidelity difference.

For reference, load with oversampling off and neither shortcut box checked is 35-45%...

Oh, and back to Task Manager and usability, YES, it does cause significant sluggishness with other applications. Perhaps not basic email or web browsing, but doing other things that are more demanding, most definitely.

beta 79 ahead
AGC is very strong to loud peaks .... which causes a drop down to all the sound ... in my opinion these loud peaks must be absorbed in multiband ..... and this is the meaning of sound balance.
Drop downs are gone !! only in band where needed 1 st , 2 nd , 3rd ........
Thatswhy .... more bands more balance .
If we have loud peak in mids ( 1, 1.5 KHz ) only band 6 is responding.
if we have loud peak in 9 Khz olnly band 9 responds .......
so on so on .......
This is the right way because in FM or only sound processing the schematic is : INPUT->FIlters ->Pre-AMp->AGC->Multiband processing->limiting/clipping/hard limit->(here is the FM checkbox) Stereo encoding-> pre-emphasis->final limit/clipping/hard limit. Maybe some where in the first components there is another filter added .. don`t know , this is optional.
But .............. my opinions are ignored ...... so this is the last one .
AGC is only to monitor the RMS level of all the song ... respectively RMS levels of the two or three bands ( if enabled ) and must be very soft --> more constant level ( or have a slider to change the speed respond like i propose one time ) .

I think the strong effect of waveform display on that mobile CPU depends on the CPU architecture itself, it has a build-in GPU, maybe that is measured with task manager too. Maybe a computer with extra graphiccard does have no problems with the waveform display on/off. I´m running beta69 atm, but i dont need the natural dynamics on my netbook because its too week, will test beta79 soon.

It doesn't give any - actually it causes a deterioration.

However, on *some* sound cards, if the sample rate is changed before playback that might cause distortion. And for FM processing it is mandatory, because without it you might exceed the +/- 75 kHz bandwidth.

Anyway, thinking about this, it might make sense to turn oversampling off by default (for FM it should always be forced on). Then someone who wants it can still turn it on.

Auch. For the last 2 hours or so I've been looking through the protection filters BEFORE the clipper (voice vibrations, volume drops due to loud S sounds, distorted S sounds caused by loud bass). Any my conclusion so far is that there's REALLY a lot of code, but that it's far too complex and doesn't do much. In other words: This needs a redesign.

It already has controls for that, the up speeds & down speeds. Your up/down speeds are too fast.

Likely a cause of additional overhead.

I know I'll get some hatin' on, but this is why I keep preaching optimizing code. I understand getting something going, but if you don't revisit it and optimize it fairly quickly, you end up with a mess like what you're saying you've found. In my prior job, we had patched code out the wazoo... The code would wind back and forth all over the place. Upwards of 50 people worked on the code base at one time or another, and naturally people would code hard returns here and there, odd variable names, etc, etc, etc... On the one hand, being a single developer gives you an advantage of not having to worry as much about code branching and merging, but on the other hand, you can get overwhelmed, then find a "shiny object" and focus on that for a while, and put off the code review / profiling / optimization.

Not that speed man ....
I`m talkin about another speed that is inside in the program . Time of respond ( if this time is small only the colour indicator moves , but if it`s large ... clipping indicator appears and then the main colour indicator moves) . i don`t know if You understand me ...
I`m using all the versions with one fixed values of that You talkin about + fixed gating . A lot of difference. There is something that is edited in AGC. 6.10 is one , to beta 79 is another .... after 79 is very different .
this means that AGC must be soft responding not to cause drop down of all the song level.
Peaks must not be a work that AGC is take care of .

@DJ-DOGGY: Try increasing the 'Force level below' slider. That slider is new since some point after BETA079. Its default value is 200% - that _seemed_ to be enough. But if it's not, set it higher. (And let me know ).

If this fixes the issue, please let me know which up/downspeed settings you are using.

Hans , just test it
I think this fixes the issue
looks like values above 140% fixing the problem - > 150 - 170% is good enough.
So no doubt 200% that You choose is not noticeable at all . I confirm .

Values of up speed that i use are 5.001 -> 5
Values of down speed are 20.001 -> 20 maybe fast but good for me .
I use two bands .

Between Hans , what happens to the filter that causes about 400 Hz popping sound ( oscillating of the filter ) maybe overloaded ? or code issue ? What do you think
What is the type of filter ? like Bessel , Cauer or Chebyshev ?