Old 05-03-2016, 08:09 PM   #81
The Telenator
Banned
 
Join Date: Dec 2011
Location: Oud West, NL
Posts: 2,335
Default

Quote:
Originally Posted by Softsynth View Post
Duly noted. Folks do not recommend convolution reverb to guys with a tendency towards high levels of OCD, less they spend hours of their life looking for that elusive perfect free IR
Essentially, I totally LOVE the whole idea of convolution reverbs. It totally makes sense, too. And I know how fine they can sound, even though I'm happy enough with a good algorithm most days of the week.

I guess, though, that the various issues, especially the latency sometimes, and maybe the extra effort got to me after a while. The way the whole world seems to be now, I'm lucky if I can find time to eat properly some days, and lots of people are busier than I am, constantly trying to multi-task and always on the run. This digital wonderland was supposed to make our music lives so much easier and hassle-free. 10,000 pages of manuals and complicated-but-ground-breaking new recording software appearing almost every other week!
The Telenator is offline   Reply With Quote
Old 05-03-2016, 08:39 PM   #82
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

Quote:
Originally Posted by Colox View Post
That's not delay doing that. What is heard there is ambience.
Respectfully, no it isn't. :-)

I think this further example will convince you: this is a dirac spike as the source audio, same IR as before:

http://lacinato.com/pub/reaper/ir1de...verb_dirac.wav
http://lacinato.com/pub/reaper/ir1delay/ir1_dirac.wav

You can more clearly hear the delay with this example. The reaverb one is right in time, but the IR1 version has a tiny gap between dry and wet. You can even see it:



...note the intial dirac spike is in the same place for both, but the IR1 verb is delayed 16ms.
clepsydrae is offline   Reply With Quote
Old 05-03-2016, 09:13 PM   #83
Colox
Human being with feelings
 
Join Date: Feb 2012
Location: Sweden
Posts: 1,206
Default

Quote:
Originally Posted by clepsydrae View Post
Respectfully, no it isn't. :-)
Well, since I opened my big mouth I had to verify my own claim, to see if I was off :P And I'm glad to say I was right:

This is the original, and a duplicate of it displaced +11,6ms, played together
This is the result of phase flipping one of the above
This is the result of phase flipping your original against your reaverb render (no 11,6 delay between them)
And this is the result of phase flipping your original against your IR1 render (no 11,6 delay between them)

Neither the Reaverb or IR1 render sounds even close to the result of adding a 11,6ms delayed duplicate to the original. Delay wasn't what caused the effect. Ambience was. There is no audible delay between your renders and your original here. The effect that got applied by the IR1, is very close to that of the Reaverb but not identical. Those two won't cancel each other out 100% either.

But, this is .. academic, heading off-topic. I just needed to redeem opening my big mouth I guess. And yes, I see your graphic measurements, and I'm not going to take any side about it. Just addressed my claim about the former render test
__________________
There are only two kinds of people in the world: those who entertain, and those who are absurd.
- Britney Spears

Last edited by Colox; 05-03-2016 at 09:53 PM.
Colox is offline   Reply With Quote
Old 05-04-2016, 12:40 AM   #84
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

The issue with IR1 isn't that it adds a duplicate of itself at 11.6ms delay. It's that the reverb that is added is not convolved in time with the original because the plugin does not report PDC to the host (it also does not align the dry signal with the verb if any dry signal is output from the plugin itself; hence the configurable latency setting in the GUI.)

So: the dry signal passes to the master from the main track as normal. It is also sent to the reverb track; the reverb generated from IR1 is spit out with a delay. This isn't noticeable with most IRs because they often have considerable pre-delay anyhow. But with a short-duration verb like this one, it's obvious. (Which is why i think it's a little suspicious that it's not possible for me to get IR1 to play short IRs -- they reveal the flaw in IR1).

So, I haven't listened to your first two examples, because they don't bear on the issue, afaict. There's no reason a copy of the delayed dry signal over itself should sound like anything I posted.

As to the second two, they conveniently prove the point. (Incidentally, 11.6ms was not the delay of the reverb; 11.6 is what they say in the manual, but the tested delay is actually different, and variable depending on algorithm, as mentioned above; about 16ms in 44.1k in mono/stereo mode.)

If you phase flip the dry, it will mix with the reverb-wet version and leave just the reverb. This is true for both files. The result will just sound like the ambiance that was added by each plugin, just as expected. However, one will be ~16ms offset from the other. They will of course sound identical when played back (identical within the small variation of how each plugin sounds).

You can easily see this delay if you pull up the latter two files that you posted -- the time selection in the following picture is about 16ms, exactly as tested before:



...so you can see from that that the ambiance added by IR1 is clearly added with a delay.

These aren't just graphical measurements. The second example I gave above with the dirac spike is very obvious: you can hear that the ReaVerb example is a single click, and the IR1 example is two clicks, right? (It helps me to hear the two clicks if I turn it up.) It shouldn't be two clicks! It should be one click with some ambiance, but it's not, because IR1 is adding the reverb with a delay, causing two perceptible clicks.

I've provided a few clear examples, the manual says it's the case, their support says it's the case, and your own example files demonstrate the delay. If you need more than that I'll have to give up here. :-)
clepsydrae is offline   Reply With Quote
Old 05-05-2016, 11:11 AM   #85
Colox
Human being with feelings
 
Join Date: Feb 2012
Location: Sweden
Posts: 1,206
Default

Quote:
Originally Posted by clepsydrae View Post
Just confirmed with Waves support that their plugin does add ~11ms of pre-delay to all reverbs due to not reporting PDC to the host. I wonder if this has something to do with what you describe above!

I also wonder how many other convo plugins do that.
Quote:
Originally Posted by Colox View Post
Yeah, I wouldn't think this affects the sound.
Perhaps it opens up for misinterpretation, when I say "there is no delay there". This means "There is no discernable delay there, audibly". What was heard there, was ambience. Delay was not heard, not at that last test. That's why the reverbed signal out of the IR1 sounded radically different than adding a delay to the dry signal, and also wouldn't cancel out.

I feel a delayed reverb signal, will cause the sensation of a delay/phasing, and it could cause phase cancellations and comb filtering alterations at different points in the time/frequency domain, compared to a reverb unit that doesn't introduce a delay. But this would only be present if the dry signal is heard to some degree as well. On 100% wet, it wouldn't matter.

Another perspective, is if you use pretty much any type of response with the IR1, and phase flip the reverb against the original dry sound, you won't get any more difference in result than you'd get from most modulating reverb units. In some cases you get an improvement by doing this, depending mostly on the character of the response file.

So the delay - which is true for IR1 native in Reaper 5 today, but perhaps not in other DAWs of either yesterday or today, or using IR1 in a DSP usage (I'm not updated) - isn't the source for the IR1s sound character. You can't cancel the reverbed signals of reaverb and IR1 out against each other even when you compensate for the delay. They just sound different.

I think you hear this character of the IR1, sense it over time when you listen to the reverbs. I remember it became much clearer when I made side-by-side listening comparisons of the same exact response-file using several different convolution reverbs. The IR1 sounds almost like a bit of tube saturation effect, to me.

A very graphical estimation shows me an average about ~17,5ms delay. Slightly different sometimes. Waves Inphase, Melda AutoAlign and Sound Radix AutoAlign all seems to agree with this. If Waves says there's an 11,6ms difference at 44.1, one wonders where did the extra 7.5ms come from? :P
So what was confusing back in the day is true: The Waves admission isn't overall applicable, may even differ between DAWs and versions of DAWs. The manual says 11,6ms for both 44,1 and 48kHz, but the IR1 interface reports 10ms in 44,1 and 11ms in 48, and measurements by you shows 66% more. Add to it, that this delay isn't really discernable, and measurements differed between DAWs, you see the foundation for a Jerry Springer type debate about things, back then.

With all respect though, remember to be very very careful with trusting the eyes rather than ears. Very easy to get fooled, convinced about something that isn't true.
For example, the wet signal's graphic profile always looks very different from the original, and reverb needs time to rise up after being induced by a dry signal. Therefore the reverb signal will graphically rise up at a later point in time compared to the original. This might look like the whole signal is delayed, might convince you that it is, while it really isn't.

Also, since the reverb signal is different in shape, one might find a similar-looking graphical dip in both of them, and be lead to believe that this dip represents the same thing, same point in time in both signals, and should therefore be 100% in (or out of) phase, while it really shouldn’t.

The two signals are also different, and they are at different overall levels too. Reaper displays graphs differently depending on signal strength and zooming. If 1 signal is weaker, a rising up reverb response won't be visible until a later point on the timeline. So we risk thinking that it is different in timing, while in truth it is only visible at a later point on the timeline, due to its lower level. Two signals that are the same, but at different levels will easily mislead.

These is a few of many reasons why I tend to listen more than look. Not as easy to be deceived then, and you can also dismiss of things that looks undesirable but sound desirable.

But for short, no, I feel certain this isn't the reason for the character of the IR1. I believe it's poor transparency in IR1s rendering - for better or worse. And I don’t know of any other convolution reverb that does this. Manufacturers usually update and change things over time, but to my knowledge the IR1 has remain essentially the same since 2004.
__________________
There are only two kinds of people in the world: those who entertain, and those who are absurd.
- Britney Spears
Colox is offline   Reply With Quote
Old 05-05-2016, 12:50 PM   #86
Softsynth
Human being with feelings
 
Join Date: Jun 2015
Posts: 8,696
Default

Quote:
Originally Posted by The Telenator View Post
Essentially, I totally LOVE the whole idea of convolution reverbs. It totally makes sense, too. And I know how fine they can sound, even though I'm happy enough with a good algorithm most days of the week.

I guess, though, that the various issues, especially the latency sometimes, and maybe the extra effort got to me after a while. The way the whole world seems to be now, I'm lucky if I can find time to eat properly some days, and lots of people are busier than I am, constantly trying to multi-task and always on the run. This digital wonderland was supposed to make our music lives so much easier and hassle-free. 10,000 pages of manuals and complicated-but-ground-breaking new recording software appearing almost every other week!

Telenator,
I've said this already in the thread- NI Reflektor (and some others) are designed to be zero latency. It doesn't matter what settings you use in Reflektor in practice it works exactly like an algo reverb i.e a real-time reverb, you can listen to as you play your instrument, no different in that respect. CPU usage is about the same as Tsar-1r algo reverb (which I feel is excellent too).

Also Reflektor has a hidden menu, you can use the basic level stuff or choose to fiddle with it by clicking on the extended menu. I have algo reverbs that are much more complex in terms of settings. No doubt other convolution reverbs offer similar flexibility.

Reflektor (like other newer commercial efforts) also comes with a more than adequate bank of it's own reverbs (I choose to play with others out of curiosity more than necessity).
Softsynth is offline   Reply With Quote
Old 05-05-2016, 05:03 PM   #87
The Telenator
Banned
 
Join Date: Dec 2011
Location: Oud West, NL
Posts: 2,335
Default

Quote:
Originally Posted by Softsynth View Post
Telenator,
I've said this already in the thread- NI Reflektor (and some others) are designed to be zero latency. It doesn't matter what settings you use in Reflektor in practice it works exactly like an algo reverb i.e a real-time reverb, you can listen to as you play your instrument, no different in that respect. CPU usage is about the same as Tsar-1r algo reverb (which I feel is excellent too).
That pretty impressive. I don't suppose Reflektor would run and work within the NI Reaktor framework, would it? I like Monark a lot, but another excuse to install Reaktor would be simply grand. I have it installed on my laptop, but can't really stand to use any of its funky outdated plugins that come in its library. I've had a good look -- just can't justify the zillions of registry entries and programme files, not to mention the PITA I had with the whole registration process. I vowed it would never get put on the main recording PC. NI drives me nuts -- if I could just have something much more minor, like Reaktor, and let me choose a dozen or two 'best plugins' and VSTi. Hell, I'd even be willing to pay extra for a lessor setup and to keep the rest of their second-rate stuff away from my machine!
The Telenator is offline   Reply With Quote
Old 05-05-2016, 05:13 PM   #88
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

Quote:
Originally Posted by The Telenator View Post
That pretty impressive.
If you're interested in a zero-latency convo reverb, is there a reason you don't use ReaVerb with ZL enabled?
clepsydrae is offline   Reply With Quote
Old 05-05-2016, 05:31 PM   #89
The Telenator
Banned
 
Join Date: Dec 2011
Location: Oud West, NL
Posts: 2,335
Default

Quote:
Originally Posted by clepsydrae View Post
If you're interested in a zero-latency convo reverb, is there a reason you don't use ReaVerb with ZL enabled?
Yes, I'm otherwise happy enough with algo plugins and have some good ones. I did get to know ReaVerb a few years ago when learning the benefits of the ReaPlugs collection. I still use ReaPitch and sometimes still like to play with ReaTune to run mono guitar lines into MIDI (and watch the little squiggles go across its screen). I do remember instantiating ReaVerb the first couple of times and without the ZL and marvelled at the latency! Wow!

There's no denying that the ReaPlugs are about as good as any for sound quality and for how they work. I'm always surprised that more people aren't aware of this. I think many just look at them and go, "Hmm, not shiny" and just move on.
The Telenator is offline   Reply With Quote
Old 05-05-2016, 05:32 PM   #90
Softsynth
Human being with feelings
 
Join Date: Jun 2015
Posts: 8,696
Default

Quote:
Originally Posted by The Telenator View Post
That pretty impressive. I don't suppose Reflektor would run and work within the NI Reaktor framework, would it? I like Monark a lot, but another excuse to install Reaktor would be simply grand. I have it installed on my laptop, but can't really stand to use any of its funky outdated plugins that come in its library. I've had a good look -- just can't justify the zillions of registry entries and programme files, not to mention the PITA I had with the whole registration process. I vowed it would never get put on the main recording PC. NI drives me nuts -- if I could just have something much more minor, like Reaktor, and let me choose a dozen or two 'best plugins' and VSTi. Hell, I'd even be willing to pay extra for a lessor setup and to keep the rest of their second-rate stuff away from my machine!

It's used within NI Guitar Rig Pro 5.
I got it as part of NI Komplete 8 which has full Guitar Rig Pro 5 with Reflektor.

Make sure you switch to stereo mode for any stereo sources as guitar rig defaults to mono operation, unless told to run stereo.

2011 technology:
http://www.native-instruments.com/en...demo-versions/

http://www.native-instruments.com/en...ayer/download/
Softsynth is offline   Reply With Quote
Old 05-19-2016, 04:11 AM   #91
G-Sun
Human being with feelings
 
G-Sun's Avatar
 
Join Date: May 2010
Location: Norway
Posts: 7,318
Default

I'm looking for a free convo reverb with true stereo.
There is one under development at KVR, right?
Anybody has a link?
__________________
Reaper x64, win 11
Composer, text-writer, producer
Bandcamp
G-Sun is offline   Reply With Quote
Old 05-19-2016, 04:14 AM   #92
noise_construct
Human being with feelings
 
Join Date: Nov 2015
Posts: 1,566
Default

Quote:
Originally Posted by G-Sun View Post
I'm looking for a free convo reverb with true stereo.
There is one under development at KVR, right?
Anybody has a link?
IRDust by mystran:
http://www.kvraudio.com/forum/viewto...f=246&t=403082


He is an extremely competent DSP wizard, and has several other excellent free plugins here: http://www.signaldust.com/plugins.html
noise_construct is offline   Reply With Quote
Old 05-19-2016, 04:31 AM   #93
G-Sun
Human being with feelings
 
G-Sun's Avatar
 
Join Date: May 2010
Location: Norway
Posts: 7,318
Default

Quote:
Originally Posted by noise_construct View Post
IRDust by mystran:
http://www.kvraudio.com/forum/viewto...f=246&t=403082


He is an extremely competent DSP wizard, and has several other excellent free plugins here: http://www.signaldust.com/plugins.html
Thank you!
Sounding very good
__________________
Reaper x64, win 11
Composer, text-writer, producer
Bandcamp
G-Sun is offline   Reply With Quote
Old 04-13-2019, 11:35 PM   #94
kodebode2
Human being with feelings
 
Join Date: Oct 2009
Posts: 164
Default

Could you kindly share your copy of IRDust with me?

Can't find it online anywhere

Thanks
kodebode2 is offline   Reply With Quote
Old 04-14-2019, 09:29 AM   #95
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

The developer apparently pulled it: https://www.kvraudio.com/forum/viewt...1769&p=7341994
clepsydrae is offline   Reply With Quote
Old 04-14-2019, 10:05 AM   #96
Softsynth
Human being with feelings
 
Join Date: Jun 2015
Posts: 8,696
Default

Free convolution VST (parallel stereo only):
https://www.liquidsonics.com/software/reverberate-le/

If you want "True stereo" rather than regular stereo in convolution he also offers that at $39:
https://www.liquidsonics.com/software/reverberate-core/

True Stereo described here:
https://www.liquidsonics.com/knowled...ng-topologies/

Last edited by Softsynth; 04-14-2019 at 10:11 AM.
Softsynth is offline   Reply With Quote
Old 04-14-2019, 10:15 AM   #97
kodebode2
Human being with feelings
 
Join Date: Oct 2009
Posts: 164
Default

Quote:
Originally Posted by clepsydrae View Post
The developer apparently pulled it: https://www.kvraudio.com/forum/viewt...1769&p=7341994
Unfortunate. No probs.

One of our key challenges for this wonderful technology of convolution, is that there are very few resources to indicate how good a convolution plugin is. Difficult to compare them, and especially in magazine reviews, focus is given far more to the impulse collections that are bundled with them.

So we end up having to try them and listen, and in some cases, two instances of the same plugin, with identical settings, using the same impulse file, do not null, if one of the instances has polarity reversed..

Difficult to know what's going on in these plugins, which is what informs my search, to find out which is the most accurate.

Based on your most informed experience, which convolution plugins would you consider the most accurate?

Why?

I'd like to create impulses, (for personal use only at this time) and having created them I would like to know the best convolution plugin to test these impulses, so I can improve my impulse creation process.. Which informs my quest for accuracy..

Any thoughts would help. Thanks
kodebode2 is offline   Reply With Quote
Old 04-14-2019, 10:26 AM   #98
Softsynth
Human being with feelings
 
Join Date: Jun 2015
Posts: 8,696
Default

Liquidsonics/Reverb Foundry Matthew Hill intentionally adds variability into the sound of his Fusion based convolution reverbs (Reverberate 2 and on) with multi layered IRs. Also chorus, EQ, LFOs, delays and stretching effects etc.

He has moved into created self generated IRs with the latest products, combining this with what he calls Fusion.
So even if we still believe there are benefits to convolution will we need real IRs gobbling up drive space at all?
Softsynth is offline   Reply With Quote
Old 04-14-2019, 10:27 AM   #99
kodebode2
Human being with feelings
 
Join Date: Oct 2009
Posts: 164
Default

Quote:
Originally Posted by Softsynth View Post
Free convolution VST (parallel stereo only):
https://www.liquidsonics.com/software/reverberate-le/

If you want "True stereo" rather than regular stereo in convolution he also offers that at $39:
https://www.liquidsonics.com/software/reverberate-core/

True Stereo described here:
https://www.liquidsonics.com/knowled...ng-topologies/

Thanks for your suggestions.

At this time, I only need regular stereo.

I have used liquidsonics-le, quite a bit, but in comparison - listening tests only, with Reapers RaeVerb, I found it colored and grainy in the result.

It processes audio via max of 32 bits, while in Reaper ReaVerb does this @ 64 bits - i,e what I observe in the audio path using SChwa Bitter plugin, which may explain its comparatively cloudier result.

My other convolution tools have been Waves IR-1 and IR-L, which are ok for typical room convolution, adequate - enough for room simulation - cos these are somewhat more complex, but not good enough for reproducing impulses of e.g a hardware device, like a hardware equaliser setting..... or DI inject box, which does not have much deviation from the original signal, so the utmost accuracy is needed by me for the convolution plugin to reproduce the subtlest change captured in the impulse file, as accurately as possible, when applied to the target audio.
kodebode2 is offline   Reply With Quote
Old 04-14-2019, 10:34 AM   #100
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

Quote:
Originally Posted by kodebode2 View Post
Based on your most informed experience, which convolution plugins would you consider the most accurate?
ReaVerb.

Quote:
Why?
Ignorant blind trust, honestly. :-) But also some experimentation and comparison. There may be others that are accurate, too.

Calculating an impulse response is not a procedure that has any room for variation, as far as I understand it. The plugins that you describe that don't null with the same settings are doing extra stuff to the signal besides the impulse calculation (e.g. Waves IR1). My understanding is that ReaVerb (set up correctly) is not adding extra stuff in there.

Don't use Waves IR1, IMO: https://forum.cockos.com/showthread....64#post1677164 EDIT: I see that you have discovered this already yourself. :-)

If you want true stereo, it's not too painful to set that up in reaper, AFAIK. EDIT: I see that you don't need this anyhow.
clepsydrae is offline   Reply With Quote
Old 04-14-2019, 10:39 AM   #101
kodebode2
Human being with feelings
 
Join Date: Oct 2009
Posts: 164
Default

Quote:
Originally Posted by Softsynth View Post
Liquidsonics Matthew Hill intentionally add variability into the sound of their Fusion based convolution reverbs (Reverberate 2 and on) with multi layered IRs. Also chorus, EQ, LFOs, delays and stretching effects etc.

He has moved into created self generated IRs with the latest products, combining this with what he calls Fusion.
So even if we still believe there are benefits to convolution will we need real IRs gobbling up drive space at all?
This aspect of some convolution plugins, I discovered the hard way, analysing the merged result, after reversing polarity on one of two plugins with identical settings and the same input audio.

Actually wrote Liquidsonics about this, but their higher end paid tools also have this constraint, as you have so rightly indicated. I guess with room impulses these minor variations are not relevantly audible. If only the end use was given choice in the matter, to enable or disable the variability, which can only be based on some randomisation, which causes two instances of the plugin to generate slightly different results.

Purely from a pedantic perspective, which may not be audibly relevant, that means these "variable" plugins, generate a slightly different result, every time you do a mix down.

Guess the market for their intellect is in creating rooms, so a bit of randomness similar to the air molecules in any room, which are never constant from one instance to another, is acceptable....

For my use case - see earlier post, I need the utmost precision, if I can have this.
kodebode2 is offline   Reply With Quote
Old 04-14-2019, 10:41 AM   #102
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

Quote:
Originally Posted by kodebode2 View Post
For my use case - see earlier post, I need the utmost precision, if I can have this.
I'm pretty sure ReaVerb at default settings with an impulse loaded and dry to -inf is just computing the basic impulse matrix with no special sauce added.
clepsydrae is offline   Reply With Quote
Old 04-14-2019, 10:44 AM   #103
kodebode2
Human being with feelings
 
Join Date: Oct 2009
Posts: 164
Default

Quote:
Originally Posted by clepsydrae View Post
ReaVerb.



Ignorant blind trust, honestly. :-) But also some experimentation and comparison. There may be others that are accurate, too.

Calculating an impulse response is not a procedure that has any room for variation, as far as I understand it. The plugins that you describe that don't null with the same settings are doing extra stuff to the signal besides the impulse calculation (e.g. Waves IR1). My understanding is that ReaVerb (set up correctly) is not adding extra stuff in there.

Don't use Waves IR1, IMO: https://forum.cockos.com/showthread....64#post1677164 EDIT: I see that you have discovered this already yourself. :-)

If you want true stereo, it's not too painful to set that up in reaper, AFAIK. EDIT: I see that you don't need this anyhow.
ReaVerb is really good in this aspect - absolutely perfect null from two instances..reversing polarity of one... i,e bit perfect identical processing...
kodebode2 is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 12:23 AM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.