| Stereo Tool https://forums.stereotool.com/ |
|
| Stereo Tool 7.52 BETA https://forums.stereotool.com/viewtopic.php?t=5722 |
Page 18 of 33 |
| Author: | hvz [ Mon Dec 15, 2014 10:17 pm ] |
| Post subject: | Re: Stereo Tool 7.52 BETA |
Windows 32 bit: Windows stand alone: http://www.stereotool.com/download/ster ... 52-017.exe Winamp DSP: http://www.stereotool.com/download/dsp_ ... 52-017.exe VST: http://www.stereotool.com/download/vst_ ... 52-017.dll CHANGES: - Optimize band coupling in MB compressor; ignore bands with coupling < 0.5%. (CPU load improvement) - REMOVED BAND COUPLING PANEL. I think it didn't do what it should do so for now I have removed it completely... - Added band linking in the way that Wes Keene described. Now 4 links: 2->1, 3->2, N-2->N-1, N-1->N. QUESTION: Band linking behavior currently behaves different from what you might expect. With 50% linking, bands can still get more than 50% apart. I can change that but I'm not sure which behavior is better, the current (if the difference is really big it can still be really big in the output) or the 50% = max 6 dB (huge differences will get much smaller). TODO: - Re-record 'This sound is processed by' thingy. It sounds annoying and I hear it on too many stations now... (even big ones with millions of listeners lol... they should really get a license!) - Restore Band Coupling Matrix after discussion with Sjoerd Wijdogen. - Stop saving .sts.pnr file if there's no PNR data. -> Reset button didn't work when PNR was disabled. - Bug fix: On program close, the parameters were removed before the HTTP server was switched off. This might cause an exception if the HTTP server is processing a request. - Add interface for media group I talked to at IBC for 50 FM and 50 web stations. - Improve password protection. Prepare to later also support password protected presets (not now). Mail Henk M. Maybe disable for now. - Crash in BETA014 stand alone http://i.imgur.com/Fpz7Jmr.png LATER: - Check Multiband settings Bojcha in Skype - ok, seen it, but what should I do with it? - Check Multiband meters when using band coupling. Something is weird. - Split limiters from coupling. - Insane bass effect: https://www.facebook.com/l.php?u=https% ... =RAQHC9vQO PREVIOUS: - Band linking 2->1 and N-1->N. Added to Multiband main window. Currently only works if "Band Coupling" slider is set to at least 1%, I will fix that later. But it can be tested now! - Fix GUI crash at certain display sizes - Test Declipper after PNR - Test voice vibration methods 0/1/2 - I don't hear it, waiting for feedback... --> Just go for "0". - Clear log in box when logged in. - Made CPU affinity more user friendly (pulldown and masks as backup) - Check if CPU load didn't increase for 3 Highs Gap Protection steps in FM HQ mode instead of 2. Both for with and without composite clipper. About 1% increase, including voice vibration protection. That's ok. But in Composite mode the difference is much bigger! Check; if needed go back to 2 steps. Or, if I'm using vibration protection in all steps, turn it off in the 1st 3 steps. --> Went back to 2.3 steps (last step is done 1/3rd), CPU load is now roughly the same as before. Highs gap protection seems to be nearly unaffected. - Check pilot generation code (57 kHz harmonic) -> Pilot is ok (generated output does NOT contain this tone), this is only a GUI (frequency analysis for displaying) issue. Ignore. - GUI: Weird jumps at 100 ms buffer size reported by DJ_DOGGY -> Must have occurred at both 100 and 200 ms. Should be fixed now, there were 2 different values used to switch between quick updates with fading and scrolling without fading. Now both use the same value. - Hide panels in password protected mode or something Temporarily done, now waiting for feedback. * - Voice vibrations: Choose one of 3 methods (010, 011, 011A). Added Method slider. See viewtopic.php?f=3&t=5734&start=47 Waiting for feedback to finish this. ** - ABDP: Mid tone intermodulation distortion: Make new filter at start of ABDP to control mids clipping level. Clip lows and mids separately, then together, measure volume loss for each and then use that as new clipping levels. Only in case where there are loud tones present (voices) with loud bass. Bojcha: Phase Tornado fixes it. Problem seems te be caused by combination of: 1. Loud bass below 100 Hz (maybe even lower); 2. Loud mids consisting of a constant tone with no or very little other sounds. But apparently only if this constant tone has a lot of harmonics - a sine wave comes out nearly ok. Ok. Fix almost there now. Todo: Improve detection (bigger frequency range? Add highs!!!). Todo: Measure at 75% clipping as well. TODO: Harmonics turnoff should not affect bass! - Sloppy causes crackling sound in piano's. Should probably be switched off is there are very little highs. - Highs Gap Protection: Turn 3rd step off for non-FM mode. - Highs Gap Protection: Forgot to make changes in Composite Clipper mode. - Fix declipper + PNR. Does it cause problems if it's used as intended? MUST BE TESTED TO MAKE SURE IT DOESN'T DAMAGE THE SOUND - Fix crash when clicking on volume bars. - Highs Gap Protection: 2nd step does almost nothing - why? - Correct, but seems ok this way... - Highs Gap Protection: power_X must not be used (sounds better without), but it must be used for something else because if power_X is 0, the current filter is insufficient. --> Now not using power_X anymore, but 3 steps instead of 2 in HQ gap mode. This seems to be very close to the optimal result. Need feedback on this.... it appears to be ok but I'm not 100% sure that the problem is completely gone. And I still need to test if it doesn't increase highs intermodulation distortion (probably not though). - Don't use Sloppy value when ABDP is disabled. - Highs Gap Protection caused a jump in the spectrum between 3.5-3.8 kHz for loud highs. This is now smoothed over a bigger area (2400-4800 Hz). - Bad gap in ABDP - causes highs to sound less full. Fix ASAP! Only in Composite mode! Was introduced in one of the previous beta's - Make Configuration, Repair and Processing settings saveable separately. - Split filters in Restoration / Processing. - At Band Coupling '0' the values in the GUI weren't recalculated properly. - Default value for Sloppy was 0% (-99 dB), should be 100%. - ABDP: Added new "Sloppy" slider. Effect: Slightly louder, 0.5 dB more highs! And probably less pumping due to highs. (*) - Band coupling settings don't work if Band Coupling slider in main Multiband window is set to 0%. - Declipper result is very bad if PNR is used before it. NOT SURE YET IF IT'S REALLY FIXED. - PNR can also be done AFTER declipping. - Display FM lowpass frequency at bottom when FM output is used. - Show warning when PNR is disabled due to changed settings. - Added Multiband compressor Band Coupling matrix. You can use this to overrule the standard band coupling settings. Only available in Extreme Tweaker mode, awaiting responses to see how useful it it (if it is useful I'll move it to Expert mode). - Advanced Clipper: Advanced Bass Distortion Protection (ABDP): Drop in output when strongly clipping around 2600-4000 Hz is reduced. - Advanced Clipper: ABDP: "Smooth Slide" slider added to make transition between deep bass (clipped) and less deep bass smoother. There used to be a drop upto the 2nd configured bass frequency, with this switch enabled the drop is smoothed (maximum at 1st configured bass freq, gone at 2nd). Result: Warmer, more open, less 'restrained' sound. But probably also a bit more intermodulation distortion. (which might be fixable by changing the frequencies a bit). (*) - Added PNR Noise & Hum - PNR analysis data is saved now - Clip highs stricter when highs cause volume drops. -> Doesn't seem necessary - ABDP: Calvin Harris ft. John Newman - Blame has distortion in lower highs; Highs distortion protection doesn't seem to handle it properly. Except for that I'm now getting a good sound for almost all tracks with bass clipping set to 100% (!) and Highs Priority to 20%, Sloppy at maximum. Cause: Highs in this track are at around 4 kHz and I'm not filtering highs for harmonics there. Very unusual track, ignore for now. |
|
| Author: | weskeene [ Mon Dec 15, 2014 10:31 pm ] |
| Post subject: | Re: Stereo Tool 7.52 BETA |
I wouldn't worry too much about the fact that it is possible for bands' coupling to not match EXACTLY in real life, compared to how they're set. A little background, Orban has publicly said that coupling is achieved by feeding the leading band's gain into the control loop of the following band (at the selected percentage). They never actually force the bands to have identical gains directly. I don't know what your implementation is, but it's working very well (at least at 100%, where I'm testing right now). In the implementation we're basing this on, the bands could easily have different gains than what was specified because every band always follows its own time constants. Furthermore, their own user manual specifies that you should in some cases only set coupling to 80% or so, because their method could actually allow the coupled band to unnaturally take on *more* gain than the leading band, because its own gain is in the control signal as well. This side trip down memory lane is noteworthy in Band 2 > Band 1 coupling where band 1 is likely to have a very slow attack time. So even though it may get all of band 2's gain in its control loop, it may not have time to respond and match the gain of band 2 before band 2's gain drops again. Of course, they did this to get a more musical sound. So what if your implementation doesn't guarantee an exact match to the settings. Perhaps your implementation also allows some freedom that might prove to be more musical than a precise match. |
|
| Author: | hvz [ Mon Dec 15, 2014 11:31 pm ] |
| Post subject: | Re: Stereo Tool 7.52 BETA |
Hm... Currently I just force band 1 to be below band 2 regardless of timings etc. Not sure if that's better or worse... The behavior that I was talking about is the behavior at <100% linking settings. |
|
| Author: | MLM [ Tue Dec 16, 2014 5:51 pm ] |
| Post subject: | Re: Stereo Tool 7.52 BETA |
Noted that since version 7:51 normal output changes when connecting the fm transmitter, the signal in the normal output responds to filter the FM output. In the version 7:50 it does not. |
|
| Author: | REA [ Tue Dec 16, 2014 9:21 pm ] |
| Post subject: | Re: Stereo Tool 7.52 BETA |
I noticed that if in normal output the signal selection is set from "de-enphasized version of FM output" to "separately output stream" the cpu load changes from 55% to 66%. (ST 7:52 beta17). Sorry for my bad english. |
|
| Author: | weskeene [ Tue Dec 16, 2014 9:25 pm ] |
| Post subject: | Re: Stereo Tool 7.52 BETA |
I'll play with 50% for a while and see if the results make sense to me. Quote: Hm... Currently I just force band 1 to be below band 2 regardless of timings etc. Not sure if that's better or worse... The behavior that I was talking about is the behavior at <100% linking settings.
|
|
| Author: | hvz [ Wed Dec 17, 2014 3:02 am ] |
| Post subject: | Re: Stereo Tool 7.52 BETA |
Quote: Noted that since version 7:51 normal output changes when connecting the fm transmitter, the signal in the normal output responds to filter the FM output. In the version 7:50 it does not.
You're saying it changes when you *connect* a transmitter? That makes no sense... unless something is wrong in the connections or the sound card. You're not talking about turning FM on and off? Because then it could very well be.
|
|
| Author: | hvz [ Wed Dec 17, 2014 3:02 am ] |
| Post subject: | Re: Stereo Tool 7.52 BETA |
Quote: I noticed that if in normal output the signal selection is set from "de-enphasized version of FM output" to "separately output stream" the cpu load changes from 55% to 66%. (ST 7:52 beta17).
It needs to clip twice in that situation... difference should be lower when I actually release this version (because currently all the builds contain some extra checks which I will turn off in the release build).
Sorry for my bad english. |
|
| Author: | weskeene [ Wed Dec 17, 2014 5:23 am ] |
| Post subject: | Re: Stereo Tool 7.52 BETA |
I used 50% coupling on 2 > 1 and N-1 > N for a decent amount of time today and I see no obvious problems. It seems to pretty much obey the rules. Perhaps, if I had a little more background on what the problem might be, I could look for it in particular. I have yet to observe a time when the amount of variance between the affected bands was greater than 50%. |
|
| Author: | 2Sense [ Wed Dec 17, 2014 5:48 am ] |
| Post subject: | Re: Stereo Tool 7.52 BETA |
Really like the sound that can be achieved using the band coupling/linking in the new beta/s. |
|
| Page 18 of 33 | All times are UTC+02:00 |
| Powered by phpBB® Forum Software © phpBB Limited https://www.phpbb.com/ |
|
