|
|
|
07-27-2021, 02:57 AM
|
#1
|
Mortal
Join Date: Jan 2006
Location: Wickenburg, Arizona
Posts: 14,051
|
Impulse responses, tone and latency
I was using matching EQ's to help make cabinet impulses today.
I noticed that every bit of length of the file before the first spike added to the perceived latency of the amp sim. I trimmed it to about 8 samples before the initial spike, the latency pretty much went away, and I don't *think* it changed the tone any.
Just how close can you trim the front and how tight can you trim the end before you really change the tone?
I was using the Pure impulse way (gave up on the other deconvolvers, I couldn't really get ReaVerb to get the tone quite right) as shown here https://www.youtube.com/watch?v=pGEE...nel=AndrewWade
This results in a sort of symmetrical impulse with a long run up and a long run back down. Can I just chop this in half?
|
|
|
07-27-2021, 08:19 AM
|
#2
|
Human being with feelings
Join Date: Jul 2009
Location: Oregon
Posts: 620
|
Find a good commercial IR that you like, load the wav file into reaper, and compare yours to it - that should give you some context for trimming the start and end, and also signal level.
|
|
|
07-27-2021, 11:24 AM
|
#3
|
Human being with feelings
Join Date: Jan 2017
Location: Los Angeles
Posts: 1,161
|
Quote:
Originally Posted by pipelineaudio
Just how close can you trim the front and how tight can you trim the end before you really change the tone?
|
Depends on the IR, but I've never been happy trimming the front end of a cab IR.
It can do "not nice" things to the spectrum even before you notice audibly.
And as you've found, it's not easy to know how far to go with the cut.
A much better way I found is to run the IR through Voxengo Deconvolver with "MP Transform" checked (minimum phase transformation).
Use a single sample click (on the first sample) as the test tone.
Then you can trim the IR length, only worrying about the tail. The front end will be as optimized as possible:
I trimmed and processed about 800 IR's last year this way. Very happy with how they turned out.
I'm in the cab-IR-as-EQ camp so I cut out all the ambience and aimed to trim them as short as I could.
In the end, most of the IR's wound up being 2048 samples, though some I cut as short as 1024, and some needed to be 4096.
Quote:
Originally Posted by pipelineaudio
|
That's the better way ITB.
Quote:
Originally Posted by pipelineaudio
This results in a sort of symmetrical impulse with a long run up and a long run back down. Can I just chop this in half?
|
What you're describing sound like linear phase, in which case you can't cut the IR in half without destroying it.
|
|
|
07-27-2021, 11:37 AM
|
#4
|
Mortal
Join Date: Jan 2006
Location: Wickenburg, Arizona
Posts: 14,051
|
andrew wade's makes a mirror image thru the eq as the result. I cut it to start right before the first spike and it nearly nulls, but I'm really interested in using the voxengo one as well.
It was such a blur last night that I don't know for sure if ReaVerb actually worked, but I think it didn't sound quite right
I'll check voxengo's deconvolver but its not cheap like it used to be
|
|
|
07-27-2021, 11:49 AM
|
#5
|
Human being with feelings
Join Date: Jan 2017
Location: Los Angeles
Posts: 1,161
|
Quote:
Originally Posted by pipelineaudio
andrew wade's makes a mirror image thru the eq as the result.
|
What EQ? If it's actually symmetrical, it's likely linear phase which may not be a good idea for a cab IR.
Quote:
Originally Posted by pipelineaudio
I'll check voxengo's deconvolver but its not cheap like it used to be
|
It's a little pricey (I still use the demo) and seemingly not updated for a long time.
Sai'ke linked to a Python library that will do the MP transform here:
https://forum.cockos.com/showthread.php?p=2357832
though I've never used it.
|
|
|
07-27-2021, 12:07 PM
|
#6
|
Human being with feelings
Join Date: Jan 2017
Location: Los Angeles
Posts: 1,161
|
This is how I determine how much to trim the IR:
https://www.youtube.com/watch?v=WfR3ucgSkAE
I go until the low-end starts to fall apart. I'm using an item fade like a sort of windowing function. You can see it smooths out the low-end rippling quite well.
|
|
|
07-27-2021, 12:13 PM
|
#7
|
Mortal
Join Date: Jan 2006
Location: Wickenburg, Arizona
Posts: 14,051
|
Quote:
Originally Posted by ErBird
What EQ? If it's actually symmetrical, it's likely linear phase which may not be a good idea for a cab IR.
|
Ozone matching
Even on his video, the result is a mirror image
|
|
|
07-27-2021, 12:14 PM
|
#8
|
Mortal
Join Date: Jan 2006
Location: Wickenburg, Arizona
Posts: 14,051
|
Quote:
Originally Posted by ErBird
This is how I determine how much to trim the IR:
https://www.youtube.com/watch?v=WfR3ucgSkAE
I go until the low-end starts to fall apart. I'm using an item fade like a sort of windowing function. You can see it smooths out the low-end rippling quite well.
|
Oh man that is PERFECT!!!!!!!!!!!! I was struggling to figure how it could be done in realtime, thank you!
|
|
|
07-27-2021, 12:24 PM
|
#9
|
Mortal
Join Date: Jan 2006
Location: Wickenburg, Arizona
Posts: 14,051
|
Edit nevermind, RTFM on me. Ok, seems like I need a different EQ if I want to be able to make lower latency impulses
I'm not quite understanding respectrum, it seemed to be stereo, but enabling the right channel seems to drop the left one
|
|
|
07-27-2021, 12:26 PM
|
#10
|
Human being with feelings
Join Date: Jan 2017
Location: Los Angeles
Posts: 1,161
|
Yeah, I wanted to trim a bunch of IR's so this is what I came up with.
I have Reaper looping a 16384 sample section, with ReSpectrum set to the same block size.
Watch the low-end, trim, fade (I like fade #4), render, done.
This is after transforming to minimum-phase, though you can avoid that step (at your peril).
|
|
|
07-27-2021, 12:30 PM
|
#11
|
Human being with feelings
Join Date: Jan 2017
Location: Los Angeles
Posts: 1,161
|
Quote:
Originally Posted by pipelineaudio
I'm not quite understanding respectrum, it seemed to be stereo, but enabling the right channel seems to drop the left one
|
I'm not sure, since I stripped my version down for this purpose. I can post a link briefly if you want it.
|
|
|
07-27-2021, 01:46 PM
|
#12
|
Mortal
Join Date: Jan 2006
Location: Wickenburg, Arizona
Posts: 14,051
|
I tried Pro Q3 as you can turn off Linear phase, but the final result is pretty craptastic compared to Ozone Match EQ. Is there another EQ that might do better?
|
|
|
07-27-2021, 01:54 PM
|
#13
|
Human being with feelings
Join Date: Jan 2017
Location: Los Angeles
Posts: 1,161
|
MFreeformEqualizer, but it's linear phase.
There's nothing wrong with what you were doing initially. You can use a linear phase matching EQ then convert the IR to minimum-phase and the latency will be removed.
|
|
|
07-27-2021, 01:57 PM
|
#14
|
Mortal
Join Date: Jan 2006
Location: Wickenburg, Arizona
Posts: 14,051
|
Can it only be converted to linear phase thru Voxengo Deconvolver or can ReaEQ do that somehow?
|
|
|
07-27-2021, 01:57 PM
|
#15
|
Mortal
Join Date: Jan 2006
Location: Wickenburg, Arizona
Posts: 14,051
|
I can't convert it using the single spike capture method right? It would have to be swept?
|
|
|
07-27-2021, 02:09 PM
|
#16
|
Human being with feelings
Join Date: Jan 2017
Location: Los Angeles
Posts: 1,161
|
Quote:
Originally Posted by pipelineaudio
I can't convert it using the single spike capture method right? It would have to be swept?
|
You can. Just capture the IR of the EQ (make sure you capture both sides of the symmetrical IR).
Then using Deconvolver, load a single-sample impulse as the "test tone" and the captured IR as the "file to process". Check MP Transform.
I just tested it with a linear-phase low-pass from Pro Q3:
Both IR's have the same amplitude response.
Forgive me if I missed something, but it should be that easy.
|
|
|
07-27-2021, 02:14 PM
|
#17
|
Human being with feelings
Join Date: Jan 2017
Location: Los Angeles
Posts: 1,161
|
Quote:
Originally Posted by pipelineaudio
Can it only be converted to linear phase thru Voxengo Deconvolver or can ReaEQ do that somehow?
|
Not with ReaEQ. Only other option I know of is the Python library I mentioned.
Mr. Voxengo talks about how he does it here:
https://www.kvraudio.com/forum/viewtopic.php?t=197881
but I don't understand all that yet.
|
|
|
07-27-2021, 02:29 PM
|
#18
|
Mortal
Join Date: Jan 2006
Location: Wickenburg, Arizona
Posts: 14,051
|
Whoa this is working REALLY well! Thank you!
|
|
|
07-27-2021, 07:31 PM
|
#19
|
Human being with feelings
Join Date: Jan 2017
Location: Los Angeles
Posts: 1,161
|
Quote:
Originally Posted by pipelineaudio
Whoa this is working REALLY well! Thank you!
|
NP. Glad to hear it.
|
|
|
07-28-2021, 12:43 PM
|
#20
|
Mortal
Join Date: Jan 2006
Location: Wickenburg, Arizona
Posts: 14,051
|
Thanks a ton, I really knuckled down and just kept pounding away and now I pretty much have my dream tone!
|
|
|
07-29-2021, 02:26 AM
|
#21
|
Human being with feelings
Join Date: May 2017
Posts: 3,204
|
Regarding latency:
If you click ZL and LL buttons on Reaverb before loading the impulse file, Reaper will compensate the latency.
Probably not useful for Live work though.
|
|
|
07-29-2021, 12:55 PM
|
#22
|
Human being with feelings
Join Date: Jan 2017
Location: Los Angeles
Posts: 1,161
|
Quote:
Originally Posted by pipelineaudio
Thanks a ton, I really knuckled down and just kept pounding away and now I pretty much have my dream tone!
|
May I ask what you were using the matching EQ on? Like what's the source of the cab tone?
|
|
|
07-30-2021, 09:12 AM
|
#23
|
Mortal
Join Date: Jan 2006
Location: Wickenburg, Arizona
Posts: 14,051
|
I'm trying to get my helix to sound like my favorite bias fx tone, its their version of a 5150 and then its using an impulse I bought from celestion.
In order to get close on the helix, I'm using an ubershall amp instead of their 5150.
Eventually, I'm sure I'll just end up learning to like some helix amps instead, but I wanted to start with something I'm familiar with
|
|
|
Thread Tools |
|
Display Modes |
Linear Mode
|
Posting Rules
|
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
HTML code is Off
|
|
|
All times are GMT -7. The time now is 09:54 PM.
|