Stereo Tool

So many adjustments, so little time. That got it Hans...many thanks.

I like to chain normal Stereo to Mono conversion afterwards when doing mono down mixes or audio touch ups from crappy commercial recordings that are in mono but have their phase issues. I usually chain stereo tool before the normal Stereo to Mono conversion. You can than duplicate the new mono file to both channels before burning on to a CD. Creates a perfect mono CD. Usually I only do it with mono recordings anyway in that case. Stereo Image manipulator sometimes leaves a little in the side channel (MID/SIDE Check) which probably doesn't matter with a normal mono AM or FM broadcast.

STAND ALONE 32 BIT: http://www.stereotool.com/download/ster ... 43-014.exe
STAND ALONE 64 BIT: http://www.stereotool.com/download/ster ... 43-014.exe

CHANGED:
- Add CPU load (kinda) display.
- Show popup window if sound card cannot be opened.
- Sound card restart fails sometimes, needs to be fixed. -> Opening non-ASIO cards first now, which seems to work fine on XP. Still need to test if this doesn't cause issues on other OS's or with other sound cards.

TO DO:
- Added Downmix to Mono button in AM panel
- AM C-QUAM clipping/limiting was happening with a twice as loud input level as it should have. Fixed.
- Add language files
- Fix Compressor DIFF
- Fix crash reported by bojcha. I might already have fixed it, not sure...
- Show Lowpass filter freq in AM and FM panel? Maybe also Downmix to Mono in other outputs (FM, stream)?

OLDER CHANGES:
- Add more logging in case of crashes.
- On top of this, make a built-in continue-on-crash behavior which logs the error and just resumes processing. If more crashes occur in a short amount of time (I'm setting it at 10 per 10 minutes), go to the restart behavior. Reason: A crash is either recoverable (in which case no further crashes will occur, maybe one in some cases), or it's not and then it will keep crashing continuously.
- Fix crash sometimes followed by hangup reported by bhelbush (it's in the AddNextBuffer function when called in the AudioRecorder thread) - seems to occur only when switching from VLC to a sound card.
- C-QUAM: Adjust HARD LIMIT. C-QUAM signal is ok now!
- Added C-QUAM clipper; HARD LIMIT still needs to be adjusted to remove some peaks that are left in L and R (to the negative side). Should be usable for small scale tests now.
- Added Event FM License (lower price, works 3 days at a time)
- Added auto-start with Windows option
- Added tests for exception handling
Improved limiter behavior. The limiter (in all compressors) now works much tighter so it causes less pumping and at the same time it causes less distortion.
- Support ASIO as well in the watchdog
- Improve restart behavior (make sure that the old instance is really closed).
- Memory usage of stand alone version reduced.
- Merged in older changes in a separate branch. should not have any noticeable effects.
- Lower CPU load when using streaming with low pre-emphasis values. "Hole protection" must be set to LQ mode if it's disabled for non-preemphasized audio.
- Make SCA frequencies configurable for HAM radio
- Real watchdog behavior: Regularly check for sound card input & output
- Added restart-on-crash behavior (enable 'Watchdog' in the Configuration panel).
- Fixed one weird hangup in VLC output when passing bad parameters
- Added option to pass extra parameters to input VLC (just type them after the URL).
- Carrier at +/- 19khz on the AM band. Just turning the FM off doesn't kill the stereo pilot. (fixed in 004)
- I again will say here that there is something wrong with Calibration section . Three options are identical with what they do : Test right silence , Test SSB lower and Test SSB higher. Nevermind what option you check, they causes same display behavour and same result as audio. --> NO, they are very different if you look at the actual output. Display is generated before calibration. Actual output does contain SSB or LSB audio.
- At the new Normal output settings there is a problem with "Input without processing" option . Sound is gagged/strange / I'm not at my Stereotool pc at the moment but when using a pipeline input and normal & fm output (192 kHz) , I noticed a strange effect when switching the normal output to bypass. Sounded like it could be a sampling rate thing? - Worked fine in some cases but not in others. I think it's fixed now.
- I think I also noticed a small typo - check the AM transmitter window.
- On the thorium preset it plays fine and then gets garbled and sounding like a 45rpm song playing on 33 rpm. cpu load is just 26% the rest of the presets work fine. (waiting for feedback, unable to reproduce it). --> Multipath Clipper is enabled in this preset. That explains the higher CPU load.
- VLC output still hangs on outputs that don't work (for example an invalid ShoutCast address). Probably because no data is sent through the output path, which causes the same stop hangup again. --> If Status is STOPPED, ERROR or NOTHINGSPECIAL, I don't call STOP or RELEASE. This seems to cause a small memory leak (a few kB probably) every time it happens, so it's not ideal. But better than a complete hangup. Still waiting for feedback on a question in VLC's forum.
- Add VLC output error detection & re-connect behavior (if possible without resetting the other inputs/outputs).
- VLC input: But something is strange sometimes plays like 1 or more samples delay between channels, occurs highs cancelation when switch to mono. I checked everything including exact mono sound input. -> Output was already running before connection was established. That's ok, but the start sample must be left channel -> fixed.
- VLC input with Low Quality Low Latency output hangs -> Solved by disabling LQLL output when VLC is used for input (which makes it impossible to reach low latencies anyway)
- VLC MP3 output. No room in current settings window... -> You can now add more options in the SOUT string (space separated)
4. Turn AM stuff off if FM is enabled (?) but what about asymmetrical clipping/limiting for FM? -> Extra switch in Expert mode
5. Add pre-emphasize output button for AM pre-emphasis
Fixed the mentioned and some other bugs in the previous beta. All combinations of settings seem to work now.
- Normal Output can now be partially separately processed for FM output. So you can generate, for example, a composite clipped 75 us pre-emphasized FM signal and a not or far less pre-emphasized signal for streaming, or a separate FM signal without composite clipping to feed remote transmitters, in the same Stereo Tool instance - no more need to run a separate copy for the stream! Only the clipping and limiting is done separately, so the CPU load does not go up much. Note that the lowpass filter can not be specified separately (lowpass filtering happens much earlier in the sequence and using different values would require doing almost all the processing twice).
- AM section now contains an 'AM Transmitter' Enabled switch.
- Headroom slider added to the AM section. No more need to abuse the Post Amp slider to create headroom for asymmetrical output. (Note: It's also not possible anymore, so you need to update your settings if you use AM).
- AM pre-emphasis checkbox added in the AM section, so you don't have to abuse the FM pre-emphasis anymore.
- AM pre-emphasis curve (which is slightly different from the FM curve) is now correctly implemented.
- AM signal can now go through the FM sound card output (it's now basically an FM/AM sound card), including tilt correction etc.

This watchdog behavior works as follows: If you enable the check, it will act if no audio passes through an enabled audio input or output for more than 1 minute. If so, it will trigger a 'RESTART SOUND CARDS' button click. It will attempt this 4 times, if NO DATA passes through the sound card AT ALL during all these attempts, if the 'Restart' watchdog setting is enabled Stereo Tool will close and a new instance will be started.

I'd like to ask that the third view window (the one that displays composite on FM) on the Input and Output screen remain active for AM. While not being useful for verifying NRSC compliance in the USA, it will at least confirm that at the audio processing output, the mask is being met. What would make it even more useful on AM would be to have the NRSC mask selectable to be displayed on this screen including the number of db down each frequency range needs to be for compliance. For those of us who want to run the low pass filter as high as possible without exceeding the mask, this would be quite a benefit.

Compile a debug version with runtime checks enabled, for starters. Yes, a debug version might run slower (or faster) and differently, so it may not be found. This harkens back to proper exception handling, not exception swallowing. Present a "layman's terminology" message, and include the capability of reporting the call stack to you by prompting the user to "send crash report to developer", or something like that.

I always test with debug versions - but have never seen this crash that Bojcha had. The debug version runs at about 1/10th of the speed of the release version so it's completely unusable (at least when using a complete preset), you cannot run it in real time. There are lots of asserts in the code to check all kinds of things that could go wrong and might not be noticed immediately. I have also analyzed both the debug and the release version with Valgrind on Linux and DrMemory on Windows, both show no warnings at all (except during startup or when opening a file, and those warnings are in Windows code, for example the open file dialog).

Reporting the call stack seems to be impossible in release version, but if an exception occurs I do log the exception code, in which filter and thread it happens, the memory location of the crash and if it's a memory error, the location of memory that it tried to read from or write to. So at least I will know where to look.

I'll also look at static analysis tools (just found something called cppcheck that I will test).
Edit: Hm, useful tool! It did find a few things - nothing major though. But it allowed me to remove some code that was not doing anything, and some other minor things.
Edit: O one major finding - but only when you have the RF display enabled (which is normally only the case during testing, since it increases the CPU load and has no effect on audio).

CPU(kinda) display is measuring with delay - may be it`s the buffer time delay . Am i right ?

True. Too much work to fix that for now (and there will be a delay anyway because it averages over time).

STAND ALONE 32 BIT: http://www.stereotool.com/download/ster ... 43-015.exe
WINAMP DSP: http://www.stereotool.com/download/dsp_ ... 43-015.exe
VST 32 BIT: http://www.stereotool.com/download/vst_ ... 43-015.dll

CHANGED:
- AM C-QUAM clipping/limiting was happening with a twice as loud input level as it should have. Fixed.
- AM Transmitter: Added Downmix to Mono button checkbox to AM panel.
- Added translations (Dutch, Spanish, Russian, Hungarian).
- Fixed issues reported by CppCheck.

TO DO:
- Fix some more (smaller) issues reported by CppCheck.
- Try PVS-Studio for static code analysis. DONE, didn't really find anything important.
- Fix Compressor DIFF
- Show better sound card warnings (4 = IN USE etc.)
- Fix crash reported by bojcha. I might already have fixed it, not sure...
- Show Lowpass filter freq in AM and FM panel? Maybe also Downmix to Mono in other outputs (FM, stream)?

OLDER CHANGES:
- Added separate processing for FM and Normal (or plugin host) output.
- Streaming output: Added option to send output to a stream using VLC.
- Streaming input: Added option to pass extra parameters to input VLC (just type them after the URL).
- Limiters: Greatly improved limiter behavior. Limiter (in all compressors) now works much tighter, causing less pumping and less distortion.
- Lower CPU load when using streaming with low pre-emphasis values, by using LQ Hole Protection mode.
- AM Transmitter: Added AM Transmitter panel.
- AM Transmitter: Added NRSC AM pre-emphasis curve (slightly different from the FM 75 μs curve).
- AM Transmitter: Added C-QUAM AM Stereo compliant clipping and limiting.
- AM Transmitter: Headroom slider added; no more need to abuse Post Amp to make headroom for asymmetry.
- AM Transmitter: Added Downmix to Mono button checkbox to AM panel.
- AM Transmitter audio now uses FM sound card settings (including tilt correction).
- Added auto-start with Windows option.
- Added Watchdog, regularly checks that sound card inputs and outputs are working. If no audio runs for a longer period, sound cards and later Stereo Tool gets restarted.
- Added Exception handling: Exceptions should never occur, but if they do, they are now handled gracefully: They are logged and shown to the user, an attempt is made to continue processing, if that fails (exceptions keep occurring) Stereo Tool can be restarted automatically.
- Added Event FM License (lower price, works 3 days at a time).
- Added CPU load (actually, percentage of time busy with processing) display.
- Added an error popup window if sound cards cannot be opened.
- Added translations (Dutch, Spanish, Russian, Hungarian).
- SCA frequencies are now configurable for HAM radio.
- Reduced memory usage of stand alone version.
- Merged in older changes from a separate project, should not have any noticeable effects for end users.
- Bug fix: Fixed crash that could occur when switching input from sound card to VLC and back to sound card.
- Bug fix: VLC input channels could get swapped. Fixed.
- Bug fix: VLC input with Low Quality Low Latency output hangs. Output is now disabled when VLC input is used.
- Bug fix: Opening sound cards could fail sometimes if ASIO and non-ASIO were combined using the same sound card. Fixed by opening the non-ASIO connections first, which appears to solve the issue.

Windows Error Reporting services (WER).

To do the fun stuff, you need to get a valid digital certificate.