I have looked everywhere for what format does the Quad Cortex use so I can load some of my York IRs. York has the choice of 44.1k for kemper, 48k for Fractal and Line6 etc. and 96k for Strymon. Which one am I supposed to use??? There is no information anywhere? Thank you for any and all help.
3 Likes
I have used 44.1k kemper ir seemed ok
1 Like
I tried the 96k and it seems to be working fine. I am just curious as to which is best suited for the QC.
I have been using 48k ML, York and OwnHammer, no issues 
1 Like
Thanks yāall. Sounds like it can do any format it wants. Does anyone know which format has the highest fidelity?
It probably resamples but I believe the native bitrate is 48KHz
2 Likes
Bump - Looks like we never got a proper answer for what IR format is ideal for the QC. Canāt find anything on it, even in the latest documentation. Anyone?
2 Likes
The internal rate is 48k so thatās what I would go with.
Thanks, that is almost certainly correct and has been already stated in this topic. I was hoping to see an āofficialā recommendation from Neural. Maybe @MP_Mod can get an answer for us. Makes me wonder if maybe the ideal resolution/length and sample rate specs might be in flux right now as the firmware continues to evolve. From other comments here though it would seem that the QC is fairly accommodating as to what formats it will accept. Even if anything greater than a 48khz sample rate at 1024 resolution is essentially converted and the additional data ignored. This spec is something I would have expected to see in the documentation.
1 Like
yeah, Iād like to see some official info too. I think it only references ācompatible formatsā in the manual, but doesnāt go into detail as to what those are. For now, I think itās a matter of 'upload it and see what happens."
From the user manual: āAny compatible WAV file can be uploaded to the Cloud no matter its length. The files will be resized to 1024 samples after being uploaded (21 milliseconds approx)ā. This means 48kHz sampling frequency. I donāt know, though, what is the difference between Single and Double IR files.
1 Like
As of November 2024 I am still struggling to find a clear answer regarding recommended IR formats for QC. It has come up a few times on this forum. I did see the quote in the user manual that @luizneto mentions above: āAny compatible WAV fileā¦ Files will be resized to 1024 samples after being uploaded (~21 milliseconds).ā
I believe most IR loaders simply truncate the WAV file if it is beyond the desired number of samples. I would be surprised if Neural is actually re-sampling the files to convert them to 48kHz, but I suppose it is possible. Doing some quick calculations in Excel, here are the implications if they are truncating higher-resolution files down to 1024 samples (see image). A 48 kHz IR would represent 21.3 ms of cab response (in line with the user manual), while a 96 kHz IR would only contain the first 10.7 ms of the response when reduced to 1024.
That also brings up an often debated question of how much time (milliseconds) should be represented in an IR file. Some guidance I found online suggests that 200 ms works for general purpose, and 500ms is best for the most accurate representation. I think an IR of that length would create delays in a realtime DSP application, and it probably goes way beyond the initial cab response and starts to capture the room qualities. That may be what you want in some applications, but for Quad Cortex, there is a separate function in the IR loader (unless you select the āLITEā version) to handle room response.
I found a video on YT where Pete Thorn does some live testing of various IR lengths. He concludes that 18-20 ms is the sweet spot, and listening to the examples in the video, I would agree.
I did my own test and loaded three versions of the same IR file from one of my OwnHammer packs: 44.1 kHz, 48 kHz, and 96 kHz. In the QC IR loader, you can use the arrow keys while auditioning IR files to quickly move between them. To be honest, I could not convince myself that there was any difference. Based on that, maybe Neural really is resampling the 96 Hhz IRs and not simply truncating them?
TL;DR: Safest bet is to stick with 48 kHz 24-bit WAV format for your IRs.
1 Like
Iāve done similar tests, plus comparing IRs to captures of IRs and apparently at those sample rates and lengths the differences are infinitesimal.
Iām pretty certain the higher rate IRs get resampled down as well as truncated. Iāll see if I can dig up official confirmation on Discord. Or we could just ask Support, sometimes they know things (not always!)
1 Like
I noticed that most of the purchased IRs I have (OwnHammer, SeaCow) are 200 ms long, even the simple cab IRs. I am guessing they do that for consistency and simplicity. The Amalgam IRs that came with my QC cab captures are also longer, ranging from ~100-200 ms. I was curious if there was any substantial data out that far, so I pulled some representative IRs into RX10 and plotted them.
From the plots below, there is not much energy beyond 20 ms in the simple cab IR. The room IR goes out further, tapering off around 100-120 ms (beyond the window shown). The last example is a hall reverb which has substantial energy beyond 1200 ms.
When I tried these in a patch on my QC, the basic cab IR sounded great. The Room IR was missing all the low end and sounded like it was piped through a telephone. The large hall reverb IR was weird, like a super-gainy room IR. Interesting, but not in a good way. Since all of these were already at 48 kHz, the QC likely truncated them, creating the strange results with the last two.
1 Like