Stereo Tool

Stereo Tool 7.71 BETA012

Note: CPU load is slightly higher because I have enabled some extra checks to find bugs (NAN checks).

Windows 32 bit:
Windows stand alone: http://www.stereotool.com/download/ster ... 71-012.exe
Winamp DSP: http://www.stereotool.com/download/dsp_ ... 71-012.exe
Windows 32 bit VST: http://www.stereotool.com/download/vst_ ... 71-012.dll

CHANGES:
- Fixed the lowpass bug that was introduced in 009. Although it has probably always been there! As a side effect, the output is also slightly louder (about 0.02 dB) with no negative effects.

OLD CHANGES:
- Fixed distortion introduced in 009/010. Extra highs are still there though! See new 'Sparkling highs' section under Advanced Clipper -> Sound
- Fix optimizations
- Fix code #warnings
- Improved highs filtering in Clipper (both Composite and Normal); sound has more brilliance and suffers much less from ducking on loud sounds.
- Fix bug: NAN's can occur at certain input sample rates (40960 Hz among others)
- Fixed downsampling/resampling issues.
- Made several changes to the Clipper that affect the sound! Sound should now be more open en brighter (but it might have more distortion in some cases, I hope not though).
- CPU: Added 'Ignore high frequencies above'.
- Big Composite Clipper CPU load improvement: Added a new slider ('Oversampling Quality' or something like that); CPU load reduction is HUGE and effect on peak control (for values above 4) seems very small.
- Solved a number of warnings that I added in the last few days
- Some small optimizations
- New Highs Gap Protection improved, now similar in behavior to the old one. I had to add 1 extra step (slightly higher CPU load).
- Fixed problems in 771-004. Some optimizations are disabled - I need to fix that.
- Fix CPU measurement for high loads (jumps to 95% for unknown reason). -> Showing real average instead of median now. Median in dark red.
- Difference in measured CPU load is bigger than expected after CPU load measurement update - check why - DONE, is indeed big! Separate slider added to show both processing and total CPU usage.
- CPU: Clipper: Reduced default number of Bass Clipper steps from 4 to 1
- CPU: Clipper: Implemented Leif's mirrorring method for Stokkemask and optimized steps using Omnia interface, much fewer steps for same result.
- Added option to add a text in the title bar to easily recognize multiple instances.
- EXCEPTION OCCURRED IN HttpServer: Got unknown widget ID.
- CPU: Clipper: Highs gap protection: Use 1 step for non-FM.
- Check if COMMON_STATIC still improves performance, if not remove it -> Does not, removed.
- Redesigned parameter classes and moved everything around for integration with other products.
- Remove Stokkemask mode for non-composite FM clipper (it's probably not working correctly and more heavy than the composite clipper)
- Fixed small bug in Composite clipper FM overhead level compensation calculation. Might make the same setting very marginally louder (less than 0.05 dB)
- Fix crash reported in previous post
- Improved CPU load measurement (small part of code wasn't included, it is now).
- CPU: Composite clipper: Optimize Stokkemask: Skipping outer parts of window
- CPU: Composite clipper: Optimize Stokkemask: Fixed 'Skip smoothing' mode to be compliant, all other modes removed.
- CPU: Composite clipper: Optimize Highs Gap Protection (trick from Leif Claesson that he uses in the Omnia.9 - nearly no CPU load needed anymore!)
- CPU: Added some #pragma's to make compiler generate more efficient code
- RDS: Added some character conversions (Euro -> E, ` -> '). (not yet tested)
- GUI: Allow drawing scopes in half-height to have more headroom above and below.

TODO 7.71:
- Attempt to reduce effect on volume of using Leif's clipping method.
- Fix DJ Doggy's reported errors (19 kHz, junk around tones).
- Optimize Chain function: Move stuff to ProcessingChain if possible.
- Fix downsampling and Quality slider problems: Lower Quality setting
- Turning "Bandpass" Highpass filter OFF doesn't really turn it off (!)
- Remove "Take some shortcuts" under CPU, probably broken! And has not much effect anyway. -> Hm, no, must have been in use for the tests. Ugh... TEST to check what is better: ON (which will probably allow more sounds through) or OFF.
- Ignore frequencies above a specific frequency -> Should lower TOTAL CPU load by another 10-20% -> Started working on this
- Synchronization is broken in 008 (was ok in 007)
- Another bug in latest beta - ST doesn't remember FM settings - presets from the fm list start up with stereo and rds encoding off
- Fix MPX display sample rate issue
- Fix Stokkemask in Leif's mode with Quality < 3
- VST multiple instances fails
- Make VST window bigger
- Fix Mac VST version (not recognized apparently)
- Fix Linux version (someone mentioned that it crashes after a few minutes, on Ubuntu 14.04)
- Clipper highs improvement
- GUI integration for Multiband?
- Instability in compressor behavior [WesKeene, Bojcha]

TODO:
- Noise learning Micha
- Change default preset (it still uses Classic compressor etc!)
- EQ with 'standard' sliders.
- Make BS412-like graph for levels
- Add 'warm' noise before clipper (requested by Andrew Palmquist, might help with PPM as well)
- CPU: Clipper: Optimize Advanced Bass Distortion Protection

TO DO WITH LEIF:
- Clipper issues Leif / Wolf Stahl (probably the same thing)

Hans,
looks like you can`t reproduce the Sparkling issue ?
I`ll make a video for it.

Clipping is the same . Advanced clipper still distorts compared to the last stable version.

This is not from low pass ....Tested.

Here is the video for Sparkling
If not run from site , then download it.

http://www7.zippyshare.com/v/Rr4LpQmI/file.html

I hadn't tested it yet with your preset - this definitely looks like a bug. Will try to fix it asap.

I have done a lot of measurements to compare the previous clipper implementation with the new one, and differences appear to be very small (which is good news). Leif's optimization makes the overall output about 0.02 dB quieter. The new Highs Gap protection is slightly less effective, which causes the lows and mids to come out a bit softer (but the highs slightly louder). The mids clipper has been changed to clip at 0 dB instead of at -1 dB, which causes the mids to be slightly (0.2 dB) louder, at the cost of slightly softer highs (0.05 dB) and bass.

Can't wait for the new build. Will you be able to disable NaN checks this go around?

In case you don`t understand for what reason i put these screenshots
Put in a "white noise" to ST, take a look at the DeEsser indicator (tune it to 0.00dB or 87000) and see the little "grey" peaks that appear on the right side (not the yellow fill).

These peaks are unwanted.

I have screnshots with white noise from latest beta and from old version.
If needed i`ll upload them.

@DJ_Doggy: My previous post wasn't about that issue that you reported, I didn't test the '19 kHz' setting, just the clipper itself. And with normal audio, not with tones. There is some fine-tuning left to fix those issues, but I first had to make sure that there were no other differences - such as a lower output level, more audible distortion or other artifacts etc.

I basically had already checked all those things, but the 'Sparkling Highs' setting had such a big effect on the highs (even with only a 0.15 dB difference in measurement) that I wanted to make sure that there was nothing in the new code that was making things worse.

Yes , i understand but it`s not about the 19 Khz . It`s about the down-right side of the L+R signal ,respectively both sides of L-R . Thereis unwanted signal after the value of highpass filter (mostly 15000 Hz) growing up to -80/-75 dB . This signal not presented in older versions. May be this fine tunning you talking about, will eliminate it? Will wait for next betas and test (report).


I test it with music too (it`s clear on scfreenshot). Other issue with Sparkling and Ignore freqs is out of this screenshots and are another issues.