Convolution Processor Test

May 23, 2008
176
0
16
51
Hey everybody,

I put together a little test of 5 convolution processors to see if there was a difference in sound from one to another. I recorded a few simple chords and am repeating them each with different convolution software for the speaker simulation all using the same impulse.

The first convolution processor used is the RIR module within ReValver. The next is Voxengo's Boogex with all the amp settings set to zero and the EQ flattened out as much as possible. The next is Sir2 (which is making the beeps in the entire recording). The last is Voxengo's Pristine Space. EDIT: I just added a fifth test incorporating keFIR.

I think they all sound similar but it was interesting to hear the slight and sometimes not so slight differences between them.

Here's the recording...
http://www.matt-steele.com/convolution_test.mp3

Which one is the best? Which is the worst? Let the discussion begin! :rock:
 
Well Boogex, SIR2 and Pristine Space obviously sounds better than RIR, and Im probably imagining this but Pristine Space seems to add just a liiitle more high (almost not noticable). I'm listening with AKG K271.
 
Thanks for your ears guys. I agree with your comments completely. I'll add after working with each convolution processor, ReValvers RIR sounds thin and lacks the most depth. Boogex is a great free option but I wish I could disable the EQ curve completely. I'd say Boogex and Pristine Space sound very similar. Sir2 has a softness and a low end response not present in any other processor, not to mention its zero latency makes Sir2 my favorite out of the bunch over Pristine Space's latency. keFIR sounds better than I expected it to, but in comparison to Boogex and Pristine Space, it sounds to me as if some high end is being shaved off.
 
I did a test between Altiverb and SIR 2. There was barely any difference between the two from a quality POV. It's the impulses themselves that constitute the vast majority of the sound, for sure. For simplicity's sake I like SIR 2. Altiverb is a bit too bloated for me.
 
Instead of putting music through them, you should just put an impulse through and compare the output to the impulse response you were using. That way you could get some numerical results on why they sound different. It could be that some do resampling (that would cut some high end, likely), or some differences in how rounding errors add up, or something like that.

Mathematically convolution in all of the cases should work out exactly the same and none of these should be subject to things like "warmth" and "depth." Of course in practice, they probably implement things slightly differently which could cause some errors.
 
Mathematically convolution in all of the cases should work out exactly the same and none of these should be subject to things like "warmth" and "depth." Of course in practice, they probably implement things slightly differently which could cause some errors.

I couldn't agree more, but that's what this test was about, and I was able to find differences between the different processors, which actually scared me a bit. After hours of study, I'm now leaning towards Voxengo's Boogen simply because of the built in curve EQ. Cutting some frequencies is just what the doctor ordered for guitar. I don't think I would use Sir2, Altiverb or Pristine Space just on a guitar track without some post EQ. Their frequency range is wide open which make them ideal for full mixes.
 
I just had a listen and SIR 2 sounds best to me, which is great, and totally in line with my own observations. It's a very simple app, but great functionality and quality. The tone seems to have this edgy part on top which doesn't come out on the others.

EDIT: I just worked it out... it brings out the extra notes in the chord, you can hear the harmonics clearer for some reason. Different takes?

The other ones all have some harshness to them, apart from keFIR which just sounds dull and small.