Stereo Tool

I've just been clicking around in RadioCaster, loading mutiple plugins at the same time, and it doesn't seem to like that. Looking into it further now...

Aaaaahhh!! Memory fragmentation! So, if you have a directory with a few dozen DLL's, it loads all of those, then unloads them, but somewhere it's leaking memory. That doesn't have to be much, but it's enough to not have big enough memory blocks left anymore. So then when you load and unload a DLL a few times, at some point memory allocations will start to fail.

If i remove some old dll's everything is fine and stable.

Stereo Tool 8.54 BETA009

Windows 32 bit:
Windows stand alone: http://www.stereotool.com/download/ster ... 54-009.exe
Winamp DSP: http://www.stereotool.com/download/dsp_ ... 54-009.exe
VST: http://www.stereotool.com/download/vst_ ... 54-009.dll

Windows 64 bit:
Windows stand alone: http://www.stereotool.com/download/ster ... 54-009.exe
VST: http://www.stereotool.com/download/vst_ ... 54-009.dll

CHANGES:
Verified the code with Valgrind, found a few somewhat suspicious things (calculations that were using uninitialized values on the first run) that are fixed now. This probably has no noticeable effect, but it could potentially solve some issues. Also found one bug with Valgrind when loading presets.

OLD CHANGES:
- Check if goniometers take latency into account
- RMS compressor mode fails in 8.53??? Just dies, no more compression, but audio output and limiters still there. (Possibly fixed, waiting for feedback)
- Omnia.9: Give more output values for smoother meters
- SST installer
- SST web gui fixes Koek
- Check report of VST crash in ProppFrexx (does not reproduce here, more info needed) ==> Solved, was not a Stereo Tool bug.
- Dropout when playing MPX2.flac in Winamp plugin
- Add MP3 detection sensitivity for variance to avoid detecting lowpassed audio as lossy
- Check MB1 Digital mode, reported to not work anymore
- Add Goniometer level compensation (in fact, just moved Pre Amp further in the chain)
- *Probably* fixed crash reported by Wunjo (CPU related, so I can't easily test it).
- Combined and reorganized Compressor calls for better cache behavior and fewer separate threading calls.
CPU load with identical settings: 6.7% before combining threads, 6.5% after, 6.4% after improving cache.
- Omnia.9: Clean up L-R area when AutoPilot is active -> L-R should not even be demodulated in that mode.
- Optimize Peak mode as well? -> No, but Average is possible (and done now).
- Add goniometer More places?
- Finished optimizing performance Auto EQ etc. DONE, 9.5 times faster than in version 8.52!
- Finished optimizing performance of AGC
- Optimized performance of Multiband 1, Multiband 2 and Singleband in RMS mode (factor 3).
- Buffer issue Triple
- SCA was broken - only for threads > 1
- Gray screen Wolf
- Optimized Auto EQ and new AGC performance (partially done) -> Output can be slightly different with 'Low CPU' enabled (possibly even better!)
- Added MagicRDS remoting support (Pira program)
- Auto EQ was missing in main page
- Increase NAN detection thresholds (test with new AGC with insane input levels)
- Omnia.9 return clipped audio via callback
- Omnia.9 logging remove time and \n
- Scheduler killed new AGC setting!
- MicroMPX decoder web interface could hang on right click on tray icon
- Buffer issue Generic plugin - needs to be tested.
- SST increased contrast
- SST wrong word in MicroMPX settings

TODO:
- SST update installer
- SST & ST update Matt's preset (Dense 3)
- Check merge use_ [1] from 8.54 to 8.53?
- Dequantizer effect on lossy codecs? -> 0.25 dB improvement on 16-bit audio without noisy sounds, if not using Noise Gate.
- MicroMPX Bits shifted -> Seems OK now, merge back into 8.53
- Run SST 8.53 release build

- Check compressor levels further -> Improved "Low CPU" mode: Much closer to normal mode now
- Octavafm's crash -> Happened once, didn't kill the audio (beside at the moment itself), recovered, message could be closed. Happened while changing parametric eq settings, but it crashed in Absolute Highs. Nothing suspicious according to Valgrind.... - Solved, was triggered by a NAN which caused a calculation to give an unexpected result, which caused a table lookup outside of allocated memory. Error has been there for almost 2 years, but would only occur if you have a NAN issue and Absolute Highs is enabled.
- Matt's MB1 NAN behavior (does not reproduce here, more info needed) --> Code added, long test running, all good so far
- Would a times 32 version be useful? (Or at least for ARM-based?) (Even more so now that the behavior is much closer!)
- WatchCat website
- Set maximum to L-R asymmetry?

- Would multiple where_i_am start points improve attack? YES!!!! Think about how to implement it...
- Extra AGC?
- Compressor/limiter on input before AGC to handle extreme sounds?
- uMPX 24x2ch interface
- One bass page
- New bass effect
- Revisit new Hard Limit / Final Limiter with Matt
- Report of Linux Jack version crashing - since 8.52

- GUI: Hide disabled stuff, far less confusing / much cleaner interface.
- Talk to chip
- 2 jaar
- Fardau
- Remote control via web interface doesn't work anymore for some people due to SID.
- -24 dB side chain range?
- Dummy data sturen om initialisatie te doen voor geluidskaart openen.
- 2nd AGC with higher threshold for voice tracks
- Bigger font on 4K screens

http://www12.zippyshare.com/v/gr4Zv6CP/file.html

Very good song for sound and preset testing. It has loud tones and can be a test track

EQ is Always before. "Between" is also before MB1.

_________________
control point
control point2

What happened to that Pre-limiter ?
And what i can use instead of it , so the effect to be the very same ?

It's still there but you shouldn't need it anymore - it was mainly needed before Leif and I worked on the clipper together and improved it.

dj_szpajda Just for clarity, is this related to the new Low CPU mode, or is this something that has been the case since version 8.40 where the new RMS compressor mode was introduced?

I'm not seeing big differences between Low CPU and normal mode, and that's what I was testing (because everything else has been there for months). It could very well be that "Attack Shape" and "Release Shape" affect each other (they are a bit 'magical' anyway, I have to internally compensate for both and the compensation might be off, especially if you use very strange values).

If you do see a difference that's related to "Low CPU" mode, please send me the exact settings that you're using, because I've tested it a lot and I don't see any big differences (I see relatively small differences, but that's impossible to completely fix, and probably not a real problem - once you've chosen either normal or Low CPU mode you can just make your preset and make sure it sounds and behaves as you want.).

I can`t see it under Advanced clipper in 8.40 for example.
If it`s on another place, then please guide me there.

Enable "Show deprecated settings" under Configuration.

But, why do you still need it?