Stereo Tool

Ok. I've heard it kinda work on a Mac station of one user. But something weird is going on, I don't know if it was something on that system or not but the CPU load was insanely high (Stereo Tool in bypass mode already took over 70% CPU load). As soon as we turned Bypass mode off it started missing chunks of audio constantly, even without processing.

I had to disable some features when I built the Mac version because Mac OS X doesn't support certain threading features, this might be related (I don't know). I can try building a Linux version 'pretending' to be a Mac so the same features get turned off, and see if I can reproduce it (and if so, I can test it myself which is a lot easier than looking at a remote system via TeamViewer).

@Edgar: Is this the Jack version that you tried?

http://www.jackosx.com/

BETA725-005:
Windows Winamp: http://www.stereotool.com/download/dsp_ ... 25-005.exe
Windows stand alone: http://www.stereotool.com/download/ster ... 25-005.exe
Windows VST: http://www.stereotool.com/download/vst_ ... 25-005.dll

Changes:
- Split Natural Dynamics into 4 sections, to make things more readable (no further changes, have been busy with the Mac version the last 3 days).

Meaning of sliders:
Initial detection
This section controls detection of which sounds need to be boosted
- Effect strength: How strong does the filter work. 0 = no action, 1 = standard etc.
- Dynamics detection: The effect is reduced if the signal is already very dynamic. This slider offsets the detection of how dynamic the sound already is. Result is displayed in the bars on the right (gray dotted part that comes in from the right: The bigger this is, the more dynamic ND thinks the sound is, and the effect of the filter is reduced as indicated by this gray dotted area).
- New method. This is a bit weird, there was an 'old' method that people liked, but I thought that I could make something better. This enables that "better" method; the following sliders only have an effect if New method is enabled!
- Multiply above: I calculate how much the current sound is compared to the 'average' level. 0 = equally loud, 1 = twice as loud etc. This calculated level is multiplied by Multiply above. Basically, bigger value means more effect.
- Subtract above: This is subtracted from the result of the multiply. Bigger subtract value means that the level where the filter becomes active shifts upwards (sounds must be louder for the filter to do anything).
- Clip above: Volume boost may never be more than this. (1 = no boost, 1.5 = 50% boost etc.)

I see that I can come up with much clearer names for these last 3 sliders, and I will in the next version.

Punch detection filter:
After testing a bit, I noticed that you only want to boost sounds that are 'punchy', in other words, if the volume goes up rapidly. This filter removes boosts found in the first step if the sounds are not punchy.
- Use punch: Turn this filter on. You should actually never turn this off, except when you're tweaking the settings of the initial detection.
- Punch rise time: Over how much time difference do we measure volume differences. Should be bigger for lower frequencies because a waveform takes more time to go up there.
- Punch spread time: The area in which we allow the volume to go up (see initial detection) is related to the punch rise time, but may be made a bit bigger.
- No speedup: O this should have been in the initial detection section There's probably no good reason to turn this off; in the initial versions it could take quite long sometimes upto 10 seconds) before the filter stabilized after a big volume change; the speedup fixes this. I've made it possible to turn the speedup off to test for bad effects, so far I haven't noticed any.
- Punch multiplier and subtract: Again, we calculate how much punch there is at each location; the calculated value is multiplied by multiply, then subtract is subtracted from it. Can be used to shift the response behavior.

Smoothing:
Controls how fast the volume may go up (should be fast!) and down (should not be fast, to avoid bad effects).


Note: I know that there are a lot of settings and that they are rather complex. I'll try to reduce things later, but first the filter must work properly.

Yes, this one!

There are (thin) lines on the top and bottom that indicate 0 dB, and on both sides the signal reaches exactly that point. So the output looks perfect.

Can you tell me exactly when this problem started? Because you first said that it was in one of the new beta's, but then you suggested that it might have been caused by the asymmetrical AM clipping, which I added in version 7.24 and is also available in the 7.24 release version. If possible, could you re-check with version 7.24 to make sure that it's not something else in your setup that changed around the time that you updated?

(Btw: About your test with asymmetric clipping: I never implemented it for composite clipping, so some pre-clipping steps are probably still doing it but the final clipping step is not).

something is wrong on Beta 5. Utilization goes very up and Gaps... anyone else???

_________________
Sexy Girls in your city

Tried this version on my new Macbook Air. It's playing audio through jack from iTunes but the audio is stuttering and playing too fast. After a few minutes the audio drops and needs to buffer again. Quality is way down and input and output buffer are on max. Even the evaluation "beep" sounds distorted.

Info from the console window:

MyApp::OnInit() ready!
Triggering sound card reset
Jack buffer granularity: 4096 samples
Activating Jack 8a00020
Jack connection opened
...............Setting priority to 98
.................................Restart Stereo Tool to restore default priority settings.
............Cannot read socket fd = 13 err = Socket is not connected
CheckRes error
JackSocketClientChannel read fail
JackMachSemaphore::TimedWait name = jack_mach_sem.501_default_stereo_tool usec = 2147483647 err = (os/kern) object terminated
SuspendRefNum error
JackClient::Execute error name = stereo_tool
Server is not running

So, your JackOSX is working? Which version, 0.89 or 0.90 beta 15? How did you get it to work, because I get the message from JackPilot that Jack server cannot start, with both versions. I work with an Apple MacBook Pro i7 2,2 gHz, so I guess enough power to let it work.

I found the cause of the stuttering; will post a fix tonight.

I'm using Jack 0.90 beta 15, just next next finish, start jack pilot, and it was working. I'm running Osx 10.8.4 on a i5 macbook air.