Convolution is the Next Big Thing

Questions and answers related to any Zoom Gear that doesn't have its own sub-forum yet. This includes any new gear that's been announced, and any old gear you've got. Please don't "post and run". Participate in the discussion. Thanks.
Post Reply
stubbsonic
dues paid
dues paid
Posts: 299
Joined: Wed Jun 21, 2017 2:24 pm

Convolution is the Next Big Thing

Post by stubbsonic » Wed Aug 15, 2018 6:36 pm

Convolution processing is an obvious choice for effects processing. It is often used for reverb, but can also be used for speaker emulation, resonator simulation, amp sim and many other things. However, it is very processor intensive. The convolution process takes an input signal and "convolves" it with an "Impulse Response" (IR) which is usually a short audio file that in some sense, contains the "essence" of a vibrational system (like a hall, a speaker, an instrument resonator, etc.). Creating a "formal" IR involves sending a full spectrum signal sweep or burst into a sonic "process" (like a room, or speaker, or resonator, etc), and then if necessary applying some post-processing to make it into a single "click" with a "tail" but really any audio file can be used as an IR. The longer the IR, the more CPU muscle it requires.

I also don't know how much standard convolution is used in Zoom and other gear-- I suspect it is part of at least some modeling processes. It would be great to see a true convolution processor in a small to medium format stomp or multi-effect. However, I don't know if processors are up to it.

One of the limitations of convolution processing is that it can lack some of the dynamic response of the original. Some vibrational "systems" behave differently according to dynamics and other factors. An IR does contain all the possible frequency content, but only under one level condition. That's not to say that it doesn't work well. Convolution reverb is stunning, and very convincing. It acts dynamically-- but while the reverb itself changes level along with the input source level, it doesn't change character. Still it works well.

I don't know if what I'm hinting at would be called "Dynamic Convolution" but some kind of more advance process that takes input level into account.

An iPad can do convolution, so I suspect it wouldn't be impossible to see a proper standard convolution processor in a stomper, but probably not soon.

It hasn't really hit the mainstream as a processor, but I suspect it will. It is relatively new, but super versatile. Weird it hasn't caught on yet, but it will.

EDIT: I did just find a product by Logidy called EPSi, which looks interesting http://www.logidy.com
Last edited by stubbsonic on Thu Aug 16, 2018 3:33 pm, edited 1 time in total.
0 x

stubbsonic
dues paid
dues paid
Posts: 299
Joined: Wed Jun 21, 2017 2:24 pm

Re: Convolution is the Next Big Thing

Post by stubbsonic » Thu Aug 16, 2018 10:00 am

If I understand correctly, the EPSi can be loaded with two different versions of the firmware. One version allows for longer IR files, up to 6 seconds (for longer reverbs) but the drawback is that there is more latency. The other firmware allows for low latency by only allowing shorter IRs of 1.5 seconds, like cab sims and room reverbs.

Ideally, it would be good to see a hybrid firmware that adjusts latency based on IR length, however, I can imagine they would have done that, if it was easy.
0 x

stubbsonic
dues paid
dues paid
Posts: 299
Joined: Wed Jun 21, 2017 2:24 pm

Re: Convolution is the Next Big Thing

Post by stubbsonic » Fri Aug 17, 2018 11:27 am

Once again, it would seem that the iOS solution is most cost effective (if you already have an available iPad).

AltiSpace and Mobile Convolution are two that allow you to load IRs. The former has more tweakable parameters. I haven't run any tests yet to see how they run on low buffer (low latency) settings.

The one worrying thing is that neither app appears to have been updated in a while, and it's unclear if they are still actively supported.

Another interesting one is Fiddlicator, which is designed for short IRs, like those to convolve stringed instrument resonators or cabs. It's not been updated in a long time either. If I have time I might test that as well.
0 x

stubbsonic
dues paid
dues paid
Posts: 299
Joined: Wed Jun 21, 2017 2:24 pm

Re: Convolution is the Next Big Thing

Post by stubbsonic » Sat Jun 22, 2019 10:44 am

Hotone recently released a new cab IR loader called Omni IR. It has a very nice feature set. You can load 3rd party cab IRs, had 4 band semi-parametric EQ, Aux in, Headphones out, and a separate XLR out (like a DI). I imagine it has a pretty strict IR length limit, so no fudging around with longer IRs- but the documentation doesn't say.

Another product in the same Omni series is the Acoustic processor. It also has the 4 band EQ, aux, cans, XLR. It shows acoustic bass and upright bass included in the models. There's a very short demo that doesn't try any of the bass tones.
0 x

Nhoj
senior member
senior member
Posts: 175
Joined: Wed Jun 20, 2018 8:28 am

Re: Convolution is the Next Big Thing

Post by Nhoj » Mon Jun 24, 2019 1:04 am

stubbsonic wrote:
Sat Jun 22, 2019 10:44 am
Hotone recently released a new cab IR loader called Omni IR. It has a very nice feature set. You can load 3rd party cab IRs, had 4 band semi-parametric EQ, Aux in, Headphones out, and a separate XLR out (like a DI). I imagine it has a pretty strict IR length limit, so no fudging around with longer IRs- but the documentation doesn't say.

Another product in the same Omni series is the Acoustic processor. It also has the 4 band EQ, aux, cans, XLR. It shows acoustic bass and upright bass included in the models. There's a very short demo that doesn't try any of the bass tones.
The question is if Zoom have new technology that can compete or surpass other brand competitors
0 x

stubbsonic
dues paid
dues paid
Posts: 299
Joined: Wed Jun 21, 2017 2:24 pm

Re: Convolution is the Next Big Thing

Post by stubbsonic » Mon Jun 24, 2019 11:08 am

Zoom targets a wide market of entry-level and intermediate players with versatile general purpose tools-- as opposed to "task-specific" devices. I think an IR loader can be VERY versatile (with a capacity for longer IR files). Such a device could have mass-appeal, but it is not easily explained or understood (by most). Then the task is to market it appropriately. Here's a fantasy example of such marketing text (just for fun)...

"Introducing the all new Zoom Impressionist. Splash your colors. Share your masterpiece."

Join the Convolution Revolution. Use high resolution recordings of actual spaces, devices, objects, and events to shape your sound. Put your guitar into any amp, speaker cabinet. Run it through any mic, or pre-amp. Place the sound in any space, or combine it with any other resonant sound. Make something familiar or something entirely new. Use our Impulse Responses, or yours. Combine two separate convolution paths with our entire zoom collection of effect models and you have an unparalleled color palette for any sound source- guitar, bass, vocals, acoustic instruments, synths, and anything you can feed into a 1/4" input."

There is at least one device out there that loads longer IRs, and it ain't cheap. But what is missing is the approach of having muli-purpose IR loader that takes amp, cab, reverb, resonator, and other FX IRs and combines them into a multi-effect pedal. It would be essential to put the convolution into the path at various points.The reason this tech is not currently available is that the CPU load is intense. It would probably require dual IR processing-- one for reverb/resonator FX, and one for amp/cab simulation. And it would have the limitation that the IR section couldn't have realtime modulation (perhaps other than wet/dry).

Probably what I'm asking for would require desktop level computing. But I'm assuming that because it can be done on current iPads, that it wouldn't be impossible for dedicated DSP chips. But what do I know? Seriously, what? Not much.
0 x

Post Reply

Who is online

Users browsing this forum: No registered users and 2 guests