Tips for creating IRs from plugins or other digital sources?

[Progressive_disclosure]

I was assuming the white line was the white noise one.

orange line looks correct to me.

But if a sweep produced the white line, then it wasn't done properly.

Okay, thanks! The blue line is from a sine sweep and the orange is the white noise. White line is the diff between them perhaps? I’m new to this stuff, and the softwares especially, so thanks for bearing with me.

What would be the ideal sine sweep for guitar cab IRs? Different lengths seem to produce different results in Spectrum Thief, and Wave Arts IR Capture app has 5 seconds as default. I’ve tried 10, 5 and 1 second as well.

Thinking about it again, the smoothed out results from the Dirac file might be the closest? I’ll need to test again and can also try the Smoothing option in Frequency Thief. I’ll try to post some screens to follow up on this.

 

[Orvillain]

I can't speak specifically to Spectrum Thief. I've never used it. I do my IR's by reamping sine sweeps made with Voxengo Deconvolver, and 20 second sweeps. I reamp in my DAW and record back through mics. I don't use any specific helper utility other than that.

But what I can say is, for accuracy the sine sweep. By using a log sine sweep, you're maximizing signal to noise ratio. Dirac impulses and white noise doesn't do that, and they are susceptible to error.

I use Dirac impulses when I want to analyse or reverse engineer time domain effects in delays and reverbs. But I don't use them for cab IR's and honestly, nor do most of the professional IR makers out there.

 

[Progressive_disclosure]

I can't speak specifically to Spectrum Thief. I've never used it. I do my IR's by reamping sine sweeps made with Voxengo Deconvolver, and 20 second sweeps. I reamp in my DAW and record back through mics. I don't use any specific helper utility other than that.

But what I can say is, for accuracy the sine sweep. By using a log sine sweep, you're maximizing signal to noise ratio. Dirac impulses and white noise doesn't do that, and they are susceptible to error.

I use Dirac impulses when I want to analyse or reverse engineer time domain effects in delays and reverbs. But I don't use them for cab IR's and honestly, nor do most of the professional IR makers out there.

That's great info and I'd like to get some good results with that method!

I did a 20s log sweep now from 20hz to 20kHz with level at -12db but the results come out much more bass heavy than the original. Is there anything else that should be tweaked, or might it be some built-in "Hype" in Archetype Nolly X that does this?

 

[Sascha Franck]

If you want to capture IRs from digital sources and stay inside the digital domain all the time, there's no need to use sweeps. Diracs work just as well.
Regarding Spectrum Thief, it's a fantastic piece of kit, but it's a Match EQ. And as such, it ignores all timebased properties - and these *do* exist and might be important for IRs. Yes, even for short IRs of, say, 1024 samples. In the past, I was once wondering whether the time aspect would play a role in how the IR is perceived, so I've tried to "re-capture" an existing IR using a Match EQ, then created an IR out of that - and there's defenitely been some difference. Hasn't been a very scientific test - so you might want to try that on your own.

As far as capturing something with diracs goes (no idea whether it's explained the same way in the video):

- Record some guitar stuff (or monitor through your DAW in realtime, but I find prerecorded files to be easier to work with as you can use your hands for mangling parameters).

- The recorded file should contain everything but the cab.

- Fool around with whatever it might be.

- Place the dirac spike on the track where your cab-less recording was. In case the recording is using an amp plugin, bypass everything but the cab on that track.

- Bounce the portion where the dirac is on your track.

- Perhaps normalize and trim the bounced file.

=> There's your IR.

 

[Progressive_disclosure]

I’ve used the Wave Arts IR tool to clone DynaCabs for use with other modelers. It works very well.

I always do a control test by capturing something like a York of OwnHammer IR first. I load the original IR and the clone into Room EQ Wizard. If the two lines land on top of each other, I know my setup is good. Then I get to cloning DynaCabs or what have you.

I managed to get some good results with Wave Art IR Capture finally. They are very quiet though. What settings do you use in the application? I ran with the default 5s sweep at -6dB and then goosed up my captured IRs up in Logic but there must be a better way to normalise them.

 

[bikescene]

I managed to get some good results with Wave Art IR Capture finally. They are very quiet though. What settings do you use in the application? I ran with the default 5s sweep at -6dB and then goosed up my captured IRs up in Logic but there must be a better way to normalise them.

Try checking the Normalize Output box on the left, under Automatically Crop, in the Wave Arts plugin.

 

[aivoryuk]

I managed to get some good results with Wave Art IR Capture finally. They are very quiet though. What settings do you use in the application? I ran with the default 5s sweep at -6dB and then goosed up my captured IRs up in Logic but there must be a better way to normalise them.

How are you setting up wave art it captures to capture the Ir’s?

 

[Progressive_disclosure]

How are you setting up wave art it captures to capture the Ir’s?

Ah, should have added that I’m using Blackhole on Mac for audio routing between applications. I’ve got an aggregate device set up with my audio interface and BlachHole 2ch.

A problem though is that Wave Art IR Capture won’t let me use input 3 and output 4 for example which is needed to not cause loops. So I need to select 1 signal and 2 recordings. With that the Normalize feature won’t work since it normalizes between the files created.

Aaand now I realized I could perhaps go the other way and use one output but two inputs instead…