Stereo Tool

Hm, I've made a hack-version where you can enable and disable some things and also try out a different reconstruction algorithm. If you want to test, you can find it here. Otherwise, if you can send me some recordings (or links to them) and tell me where you hear it, I can try it out myself.

Windows stand alone: http://www.stereotool.com/download/ster ... 85-046.exe
Winamp DSP: http://www.stereotool.com/download/dsp_ ... 85-046.exe
Windows VST 32 bit: http://www.stereotool.com/download/vst_ ... 85-046.dll

I'm suspecting 2 things that could be causing it:
- The reconstruction algorithm. So, now you can choose another one
- The algorithm that detects and fills MP3 holes might switch on and off too quickly. You can bypass it, that will add audio everywhere.

Thanks Hans, I am trying your hack version beta right now. Check your pm for two links to videos which you can try for yourself.

Another quick listen: the 'no protection' hack seems to make artifacts worse, while I'm not too sure about the 'alternate reconstruction' yet?

Might be worthwhile doing some measurements.

Ok, great. Effect is very clear on this. I do prefer the sound with the Delossifier enabled, as long as the top part isn't reconstructed. I've also tried using Absolute Highs to restore the highs, but it still sounds really bad, probably because the source material is so horrible.

What I will test: Dynamic detection of peak frequency in the MP3 file, and restoring upto that frequency. I don' t know if that's a good idea but I'll add a setting for it.

The old restoration method seems better than the new one. But I need to do more tests to make sure. Also, for the new one, Strength needs to be set lower (!) about 3 dB lower, so for the default 141%, you need 100% for the new method.


Edit: I see some other things that aren't working exactly as they should. Will try another way of determining how much reconstruction is needed as well.

Dear hvz,

When do you think to have the software compliant with 384Khz sample rate?
Im using 67Khz and 92Khz SCA subcarriers in my system and the 92Khz subcarrier with the 192Khz sample rate gets very distorded.
Thanks.

really like it very much. if the pitches can not be reconstructed now from the cut edge, then it sounds really good. But the alternative algorithm I find not so good! but the hack with the "no protection" I also find interesting. just in bad mp3 sounds I find very good.

thing I have occurred overnight: you could in the "absolute highs" exactly contrary, so add the high frequencies? So the automatic determination of the frequency, at which the pitch is to be completed? then the two filters would super cooperate.

@near05: I'm working toward that now. Next build will detect max freq and lower it quickly after a spike. Only disadvantage is that after a spike the highs stay there for a moment. On not extremely low bitrate files I like what it does though. It does fill holes on top but if they are big it dials down, which seems to avoid sounding distorted a bit. Building now!

7.85 BETA047

Windows 32 bit:
Windows stand alone: http://www.stereotool.com/download/ster ... 85-047.exe
Winamp DSP: http://www.stereotool.com/download/dsp_ ... 85-047.exe
Windows VST 32 bit: http://www.stereotool.com/download/vst_ ... 85-047.dll

CHANGED:
- Added several options to change how highs are reconstructed in the Delossifier: Not at all, or a bit after a frequency spike. Finally, there's an option to immediately jump up on a spike or go up more slowly, which might reduce distortion-like effects on sounds like piano's.

OLD CHANGES:
- Added option to NOT restore highest part of the spectrum.
- Simple mode: Natural Dynamics button too small?
- Add OIRT/Polar Stereo ("You probably don't want to use this") warning
- Fixed NAN message caused by Delossifier on startup.
- Fixed crash in 040 stand alone version
- Make Simple mode single overview screen?? - Waiting for feedback, is this usable?
- SPA remove monthly popup
- MP3/MPEG2 input: Added Delossifier Spectral Hole Filler!!! http://www.welt.de/wissenschaft/article ... nervt.html
- Fixed crash in Dequantizer caused by the optimizations added in BETA038. So far, didn't see any other issues caused by those optimizations.
- Test "looper" optimization on Raspberry Pi 3 -> No effect
- Command line version doens't like long name (Antenne V. - see mail 2-3-2016 22:24) -> Was already fixed.
- Removed optimization that caused crashes
- Added other optimizations
- Voice protection in classic MB is broken. Check what changed. (Code should not have changed at all). - Hm, more has changed. Check in which version! -> Waiting for reply to question -> Compared again, and the distortion is caused by the singleband compressor that's used afterwards. Which does indeed get a slightly louder input because a (wrong!) filter has been fixed since that version.
- MB: Make Overview window for new Multiband! For main settings. Hide other windows when possible, and also make it for MB1
- Exciter: Improve start of filter (less steep!)
- Fix crash in BETA036.
- Bug in 034 should be fixed again.
- Exciter: Make it possible to choose where in the chain to run it
- Hide Stereo Tool name and THIMEO in help window in specific versions.
- VST plugin level should be ok again
- Did some optimizations for the Raspberry Pi version, not sure yet how much effect they will have (so far only build on Windows).
- Exciter: Make 2 kHz block size configurable, or automatic, or combination. Might work best if it's constant?? --> This should fix the too bright sounds in some cases without introducing distortion in other cases. I hope...
- Exciter: Optimize for LPF setting
- Plugin -> Output level doesn't work in VST version
- MB2: Make Overview window for new Multiband! For main settings.
- Woops, wrong flag set for Exciter, dynamic Squeak mode didn't work. It does now.
- "Input without processing" doesn't work in stand alone version.
- Exciter: Add Volume slider
- Exciter: Add DIFF mode
- Exciter: Try to improve tone behavior (should also reduce whole area around a spike I think - maybe configurable?)
- Add another AM test tone, sawtooth with cut off bottom
- Add 'Forcibly remove' to AM settings? Or is this a problem with the transmitter that the tester on Radioforum uses? -> Regardless, I've added it now.
- Fixed AM sawtooth test tone, adjusted level of other test tones when using AM.
- Fixed bug in HARD LIMIT introduced in BETA027 (caused crashes).
- Added slider to Exciter to choose between a more natural spectrum which however sounds more distorted, vs a very clean sound but with shrill tones in it. Will try to get rid of this slider and automatically choose the correct setting soon!
- Polar stereo: K filter added, but output is not strictly clipped anymore. Waiting for feedback.
- Add AM test tone: Sawtooth with pause at 0, asymmetrical. LPF needed?
- Fixed crackle at startup caused by Ignore high frequencies (even if it was disabled!)
- Loading old preset from before Analog mode was added apparently loads in Analog mode! -> Works now, just no idea why??
- Exciter: Crash should be fixed.
- Exciter: Sometimes you see darker bars in the spectrum, changing the overlap back and forth removes those, so somehow some levels aren't always calculated when a setting is changed.
- Fixed crash when closing
- Exciter: Moved to separate page, for now called Absolute Highs.
- Exciter: Added GUI element to show Absolute Highs effect.
- Exciter: Broken at latency 512
- GUI: Fixed hangups caused by tooltips.
- GUI: Remove dots in changed settings for some buttons on top of the screen.
- Exciter: Made frequency range configurable!!! (Old default numbers were 8000-9000 Hz)
- Blog post pre-processing, MP3 and declipping
- GUI: Added tooltips!
- Added exciter (rest of list removed)
- Added (not yet fully ok) Polar Stereo support for OIRT FM stations.
- Error (crash) on some systems should be solved
- Show something in settings that have been changed (for example a dot next to the name of the setting).
- Add license check function: extern "C" bool stereoTool_IsLicenseValid(gStereoTool* st_instance);
- Make scope skinnable
- Bug: i found one malfunction in the BS412 Section: If your switch off the FM Option and leave the BS412 Limiter in On-Position the Advanced Clipper only works in a reduced way: You can´t change the Clipper drive even if you pull the handle. The BS412 Limiter is greyed out instead so maybe someone doesn´t find the failure. If You switch the BS412 to off everything works fine again! Just a small detail, but maybe someone will have problems… -> I think I have fixed it (but there are a lot of different combinations of settings that can impact the result, I hope I haven't missed any). The next version shouldn't have this problem anymore, also, when BS412 is used I have now disabled the Advanced Clipper Drive slider.
I've been busy hiding a lot more deprecated features (if they are set to their default value). This should make the GUI a bit simpler. Stuff that's now hidden (still there and still functional if you really want to use it):
- "Bass AGC (Old)" panel
- "Classic Multiband Compressor" panel
- "Classic Singleband" panel
- "Pre-limiter" panel (in Advanced Clipper)
- "Leif's clipper efficiency modification" checkbox
- "Composite clipping Strictness (CPU)" checkbox
- Phase rotation: ""At start of processing" checkbox
- AGC "Compatibility & behavior" header
- All Singleband and Multiband "Compressor type" (Analog/Digital) pull-downs
- Multiband and Natural Dynamics "Compatibility mode (bad)" for Flat Frequency Response checkbox
- "Bass protection (Deprecated)" header in Advanced Clipper -> Bass
- "Highs gap protection: *" in Advanced Clipper -> Highs
- "Smooth slope" in Advanced Clipper -> Distortion
- "Upsampled highs clipping (useless)" in Advanced Clipper -> Distortion
- Stokkemask: "Force Stokkemask even if not using Composite Clipping (bad for audio)"
- BS412: ""De-basser non-ITU.1770 compatibility mode (bad)"
- Right clicking now restores default again, mid or double right clicking restores last loaded preset (or startup) value.
Bug: If you use FM and Normal output, with separately processed audio for streaming for Normal output, and for FM output enable pre-emphasis but *not* pre-emphasize output, then Normal output is not processed separately!
- What is the pink line in the BS412 display? Maybe remove it. http://i.imgur.com/70mWvqp.png -> Ah no, it's the adjustment of the compressor thresholds. Needs to be added to Help page.
- BS412 has VERY narrow jump on de-basser from 240 to 250 Hz. (Jump of multiple dB's). -> No, brown line jumps but actual output is smooth, so no problem.
- BS412 bass filter doesn't use ITU1770 as it should but some weird other shape that goes up to about 250 Hz.
- Need to support 2 default values and a RESET TO LAST LOADED VALUE reset option.
- Hide not selected deprecated stuff - partially working!
- Move Copyright stuff etc. also to command line version (maybe just copy it?)

TODO:
- Added mode where for slowly rising the highs above the current level are not reconstructed. This reduces 'fluttering' highs (not sure how to describe those) that can otherwise be present if small bursts of audio above the current maximum frequency get reconstructed.
- MP3: See discussion with Near05 on the next page. Delossifier highs reconstruction if it's not a gap that needs to be filled sounds much less good than using Absolute Highs there. So stop regenerating top part! (At least add an option for it!)
- Add Polar stereo viewtopic.php?f=16&t=6233&p=33031#p33031 - Feedback so far is that it works fine, I'll just assume that it's ok until I hear otherwise.
- Simple mode: How to handle clicking on things at bottom?
- Delossifier: Add *very* conservative mode (minimum of rolling level and current level)
- MP3 Propaganda - Duel way too much highs generated on default settings!
- SPA "This sound is processed by"?
- Exciter: Try to reduce pre-ringing (maybe by doubling overlap? Or measure on bigger block size, then go back to smaller?)
- Exciter: Add auto-adjust HPF frequency
- Put in Ben Liebrand settings for vinyl cleanup? Asked question on vinyl forum.
- License check
- set Encryption : impossible to view or modify all the processing settings
- Preset Demo : an encrypted preset with an audio message and beeps when not registred (same as your plugins)
- Preset Licensing : encrypted preset without beeps unlocked by a license
- Resampler: Predict effect and compensate for that.


LATER:
- Check Dirty Voices bass IMD problem. Maybe also rename to Phat Bass?
- Try adding phase delay compensation
- Also show tooltips for blocks of parameters.
- RDS encoder: Use Mathijs' method with interpolated 0's. Should give a nicer signal! Also I saw a weird jump bug in the raw (unfiltered) RDS output. Shouldn't matter (it gets filtered out) but still better to fix it.
- MB: Add voice volume protection, don't allow mid bands to drop below surrounding (to some extent), protectes voices and it might help to replicate the sound of certain other processors
- Continue with finding parameters to hide
- O skin

Hm, not really looked into that at all yet. It's not a trivial change, and I don't think there are a lot of people using SCA2, who also have a 384 kHz capable sound card...

Last time when I compared the difference appeared to be in the singleband compressor that you were using at the end. Can you try what happens if you disable that?

What's that all about, particularly the first point?