Speaker Profiles Will Replace IRs

Speaker Profiles Will Replace IRs
In this video I check out the Quantum Speaker from Neutron Studios. The QS is a collection of speaker profiles that are 99.6% accurate to the original source essentially making it identical to micing the same guitar speaker in the same way. This QS pack features a Celestian Greenback Speaker and various microphones and placements.
These speaker profiles use the Neural Amp Modeler. Similar to how we have amp profiles, speaker profiles are just the other side of the chain.
Check out the Quantum Speaker here:
www.neutronstu...
Check out the QS Deep Dive by Neutron Studios here:
• Neutron Quantum Speake...
Please leave any questions or feedback in the comments!
Follow me:
Instagram (@stevenkambach) - / stevenkambach
Tiktok (@stevenkambach) - / stevenkambach
Facebook - www.facebook.c...
(NOT "Steven Kambach Music" that one was stolen from me)
KZbin (@stevenkambach) - / @stevenkambach
Contact me: [e-mail]
More from me:
linktr.ee/stev...
MY GEAR (Amazon Affiliate Links):
Laptop: linktw.in/JVgAEL
Camera #1: linktw.in/tuBHJO
Camera #2: linktw.in/FJDtqL
Camera Lens: linktw.in/EJmVfg
Interface: linktw.in/hTzkfd
Microphone: linktw.in/gSOjrz
Desk: linktw.in/FvEaTb
Guitar Strap: linktw.in/cqRudT
Strap Lock: linktw.in/FhmJag
In Ear Monitors: linktw.in/IERWNt
Guitars -
Epiphone Les Paul: linktw.in/iBRnsn (I have an old version from 96')
Fesley FLP350: linktw.in/qQTNhk
Blackstar Travel Guitar: linktw.in/gzSDFi
Squier Stratocaster: linktw.in/AnVzBf
Pedals -
Tuner: linktw.in/OqCVQn
Blue Overdrive: linktw.in/exZytB
Green Overdrive: linktw.in/vwQZNQ
Amps -
Blackstar Fly 3: linktw.in/xNwfbw

[@JasonAnthony2222]

I play Kemper and QC through a KSR power amp and KSR cab. I ripped the QC through the (v30s) in the room and one of the sub and it feels and sounds awesome. I'm glad I saw this posted on FB. This beats the hell out of option paralysis and 100,000 IRs and I can play quiet and still get "the sound". Thanks for sharing.

[@stevenkambach]

That sounds like an awesome rig 🙏

[@ReductioAdAbsurdum]

A comparison with IRs would have been useful. Also, some mention of CPU cost. IRs are linear, so you can take 50 of them and combine them to form one IR. That's obviously not going to work for this.

[@stevenkambach]

Def correct on the CPU aspect 🙏 that is something that will need to improve as these become more common. I chose not to compare because there are so many different options for IRs that there is no "standard" to compare against. Any IR I would have picked, someone would have said "should have used this one" not to mention IRs vary widely in how they were captured. Even if it is the same speaker, there are more variables between different people and signal chains used capture that make a difference. TL:DR - I didn't think it was important to compare to an IR because there is no standard IR that everyone would agree was a good representation of IRs

[@djabthrash]

@@stevenkambach UNLESS ... the dudes from Neutron Audio make IRs while they do the capture (so that the captures and the IRs are of the exact same setup and can be compared directly), and show the differences in a comparison video, but i doubt they will do that

[@neutron_audio]

Thank you for your continued feedback! I’d like to address a couple of points you made about about IRs and our approach with Quantum Speaker profiles. Purpose of Quantum Speaker: Quantum Speaker profiles are designed to accurately replicate the real-world, non-linear behaviour of a speaker, capturing nuances like distortion, harmonic interactions, and other characteristics that define a speaker’s true sound. IRs cannot capture any of this behaviour, they are just static EQ curves with some added convolution (ambience). For this reason, they are not relevant at all to the discussion around realistic speaker simulation, they are just fundamentally flawed in that regard. Linear vs Non-Linear: While IRs are incredibly convenient in many situations, they are linear representations (non non-linear as you stated), meaning they capture a static snapshot of the system’s response. In contrast, Quantum Speaker profiles focus on the non-linear dynamics-things like speaker distortion and changing behaviour with volume and frequency-that are crucial for an accurate, real-world simulation of a speaker’s performance. Combining Profiles: One important difference is that while combining multiple IRs (or any linear phase EQ curves) as you suggest often leads to serious phase, frequency peak and signal degradation issues, Quantum Speaker profiles are designed to be combined, allowing for realistic and natural results that stay true to the speaker’s behaviour. Again this is the fundamental purpose of the product, and since you cannot do that safely with linear phase filters (like IRs) they are again not part of the discussion as far as we are concerned. Try it for Yourself: We encourage you to compare Quantum Speaker profiles to your favourite IR of the same speaker if that is something that interests you and decide for yourself. In the end, it’s all about what sounds best to you. To make it easy, we provide free samples, with more coming soon, and over 3 hours of deep-dive content to explain and demonstrate the Quantum Speaker concept.

[@djabthrash]

@@neutron_audio There are advanced IR engines/ecosystems/plugins (like Two Notes DynIR, ML Soundlab stuff) where for the same cab, all the mic positions are in phase with each other.

[@void_snw]

Forgive my lack of understanding, but I thought non-linear meant it would react different to different input volumes, which an IR does not do afaik. So what's up with that?

[@whawha9016]

Happy New Year. A real interesting experiment would be to hear a normal IR of the speaker VS speaker profile of the same speaker! As others have said, comparison of the CPU-use would also be interesting!

[@stevenkambach]

Good idea 💡 NAM uses mote CPU than and IR loader 👍

[@ReductioAdAbsurdum]

*"Based on null testing get you 99.6% there"* There's no way to tell that from a null test, so anyone making that claim is (1) technically illiterate or (2) lying. If you run an identical signal through a real physical cabinet twice, it won't null. If we were to take a non-zero null as an indication of audible difference, we'd have to conclude that the cabinet sounds different from itself. That obviously makes no sense. In fact, if you copy a sine wave and shift its phase by 30 degrees, when you invert the polarity on one of them, they won't null, but they are in fact that same signal and they'll sound identical. A perfect null can only guarantees the signals are the same and in phase. A non-perfect null does not guarantee that the signals don't sound the same (or even that they're not literally the same). It could be argued that a deeper null means it's more likely that the signals can't be audibly distinguished. However, we can't necessarily say even that. See the sine wave example. Dave Rat has a video on his channel (kzbin.info/www/bejne/hofbpaxspLaHpJo) where he shows that his Behringer X32 console is 10 times noisier than his Yamaha CL1 console on a scope, yet the CL1 sounds twice as noisy, because the noise of the CL1 is in frequencies we're more sensitive to. So profile X might have a null remainder that's twice that of profile Y, but actually sound closer to the real thing, because the differences are in less psychoacoustically important regions. That's the purpose of dithering in converters, or in deciding what information is thrown out in perceptual codecs like MP3. Null tests have their usage, but the conclusion "better null = better profile" is not supportable.

[@stevenkambach]

Take it up with Neutron Studios 👍

[@djabthrash]

Very interesting take !

[@neutron_audio]

Thank you for your thoughtful comment. You’ve raised some important points about null testing, and we appreciate the chance to clarify. A null test is indeed just one tool we use to assess how closely our Quantum Speaker profiles match the original mic’d signal. We agree that perfect nulling guarantees identical signals in phase, but a non-perfect null doesn’t automatically imply audible differences, as psychoacoustics and frequency sensitivity certainly play a role. Your example of phase-shifted sine waves and console noise highlights this well. Regarding the 99.6% match we mention, it comes from a controlled test specifically designed to establish a highly accurate comparison. In this test, we ensured perfect temporal and amplitude alignment between the original mic’d signal and the Quantum Speaker profile to eliminate any timing or level discrepancies. This precise alignment allows us to accurately measure the delta between the two signals and establish the level of residual difference. Here’s the technical data from the null test: Quantum Speaker Profile: The Quantum Speaker profile null test resulted in a -24 dB score (i.e., the null was 24 dB quieter than the original signal). This means that only 0.398% of the original signal’s power remained, corresponding to a 99.602% match with the real mic’d signal. While we don’t claim that null tests alone are the sole way to judge audio quality, a -24 dB residual signal is still very low-well below the threshold of perceptibility in most real-world listening environments. The remainder of the signal that did not null is minimal, and the 99.6% match demonstrates that any audible differences would be imperceptible to most listeners in practical settings. We believe this is a fair and accurate description of the accuracy our customers should expect from the product, and as you can imagine, there are not many easy ways to explain it. Controlled Test and Temporal/Amplitude Alignment: It’s important to note that this test was conducted under controlled conditions. We used strict protocols to ensure that both the Quantum Speaker profile and the original mic’d signal were in perfect temporal and amplitude alignment before running the null test. This ensures that the comparison was valid and the measured delta accurately reflects any differences between the two signals. Psychoacoustics and Listening Tests: You’re absolutely right that psychoacoustics play a key role in what we actually hear, not just what a null test measures. This is why, in addition to null testing, we conduct extensive listening tests to ensure that our profiles sound indistinguishable from the original in real-world scenarios. While a deeper null suggests a closer technical match, it’s the combination of technical accuracy and listening feedback that confirms the profile’s fidelity. Additionally, we ensure that any residual differences, where they exist, fall outside the regions that matter most in psychoacoustic terms. For the Quantum Speaker profile, the result is a signal that has been found to be virtually indistinguishable from the original mic’d cabinet in typical listening environments. Conclusion: The 99.6% match from the null test reflects the extremely small deviation between the Quantum Speaker profile and the real mic’d signal, as established through a carefully controlled test with perfect temporal and amplitude alignment. While we recognize that null tests are not the only measure of quality, they offer a reliable insight into the accuracy of our profiles. When combined with real-world listening tests, we are confident that the Quantum Speaker profile delivers near-perfect fidelity in both technical and practical terms. We appreciate the conversation and your thoughtful insight. Thanks again for sharing the Dave Rat video-it’s great to dive deeper into these topics with those who care about sound quality. If you want to know more, come join us on our FB group and I will be happy to supply you the files from the test to analyse for yourself to validate our numbers.

[@ReductioAdAbsurdum]

@@neutron_audio *"we ensured perfect temporal and amplitude alignment between the original mic’d signal and the Quantum Speaker profile"* That means you did the null test correctly. It doesn't say anything about what conclusions you can draw from a null test, which is what we're talking about. *"this precise alignment allows us to accurately measure the delta between the two signals and establish the level of residual difference.* You're measuring a dB difference. You can "accurately measure the delta between the two signals and establish the level of residual difference" with a dB meter. The careful use of language doesn't change what we're talking about: doing a null test and measuring the dB level of the difference. The question is not _how_ are you measuring this difference -- we'll take for granted that know how to do a null test and know how to measure dB levels -- the question is about what _conclusions can you draw from it._ *"a -24 dB residual signal is still very low-well below the threshold of perceptibility in most real-world listening environments"* dB is a relative measure. Is that 24dB below the original signal? Peak? LUFS? Do mean -24 dBFS? 24 db is ~4x quieter to the human ear, which will be more or less audible depending on where in the signal it is. But that's not being quantified. It's fair to claim "the difference is not audible in most real-world listening environments". But the claim that it's a "99.6% match" just doesn't make sense semantically. The level of "match" is not being quantified. And how are you getting from -24 dB to the value 99.6? In any case, Leo Gibson is doing the same thing on his channel, asserting that a better null = better profile. It's equally unsupportable there. A real amp won't null with itself. Identical pink noise won't null with itself if it's offset even a little bit, you can't correlate the troughs and peaks in the output of processing that includes effectively random elements, no matter how perfect the "temporal and amplitude alignment" of the input is. *the result is a signal that has been found to be virtually indistinguishable from the original mic’d cabinet in typical listening environments* That's a fair claim. *the 99.6% match* That makes no sense. It reminds me of an old face cream commercial where a woman in a white lab coat from "Ponds Institute" asserts that their beautify cream reduces the appearance of aging by 37.3%, as if they even attempted to quantify that. It's using numbers to create the impression of precision where it doesn't really exists. It immediately sets of snake oil red flags, which is not what you want for your product. Dan Worral covers what conclusions you can draw from null tests at the start of this video :kzbin.info/www/bejne/Z3qrZ4FsaKaLq5Y I'm going to ping him on this one, now that it's more than just Leo Gibson doing it. He lives for this kinda stuff.

[@yikelu]

@@neutron_audio Personally I'd rather you give the dB score on the null rather than the percentage. When the numbers get this small (due to the nonlinear scaling), percentages become misleading. It would also be fair to give both numbers as you did here. Also, it would be helpful to give the null test scores of standard IRs for comparison.

[@yikelu]

This immediately reminds me of why I prefer ribbons for guitar cabs in a rough and tumble recording situation. It's so much easier to just set the ribbon on speaker center and a couple inches back vs moving a dynamic around to find that sweet spot. So much more forgiving.

[@stevenkambach]

I've never used a ribbon mic, but these profiles of the ribbon on the speaker are my favorite 👍

[@yikelu]

@@stevenkambach I saw some shootouts with ribbons and even cheap ribbons sound great on guitar cab. Fast forward a few months and I'm recording my band and the dynamic mics I was using sound absolutely terrible -- incredibly harsh even though I was somewhat off center. Then I remember what I heard with the ribbon shootout. Next session I swap all the cab mics to cheap ribbons (I kid you not, $100 MXLs) and problem SOLVED. They're especially useful when you don't have all day to audition every mic position. Set to speaker center, back off 3-6 inches and you're good to go.

[@stevenkambach]

👍

[@GingerLeftyGuitar]

Have both the V1 and V2 versions of the QS greenback. They sound super. Love them. Paired with my Bluguitar Amp1, I have everything I need.

[@stevenkambach]

That's awesome 🙏 the QS is great 👍 looking forward to the V30 set

[@xlilxillx]

we need more speakers recorded with a flat response mic in a sound chamber so we can emulate the sound of a speaker in a cab in the room, not the sound of a speaker being mic'd by a sm57 on the cone or whatever. i think that's what modelling is missing and why so many people say it doesn't feel the same as a tube amp

[@stevenkambach]

Good idea 👍

[@ivanvicente9633]

Man, I was always gaslighted when I told mixers/producers that I really preferred micing cabs and that I got much better results no matter the speaker amp combination I used

[@stevenkambach]

Been on both sides. The struggle is real. I always lean towards the talent's preference, but when I was the talent I always just went with what was easier for the techs 😄

[@yikelu]

I'd argue that a full rig capture essentially bakes in the speaker dynamic response as well. It's not separable from the amp capture of course, so that's a major downside, but it gets you the dynamism.

[@agirotto1]

That's my thought as well. They seem to work better for me than the amp capture + IR setup.

[@johnbowman3630]

Agreed, but the issue here is that _any_ component within a dude's rig who is making full captures is baked into the engineering he did through the cab; in other words, if he has a lot of HI-end in his guitar>pickup>amp stage, he very well may dial all of the 'sizzle' out of the capture, and thus you're unable to access the speaker in its full fidelity. This approach seems _way_ better in theory for being able to offer a next-gen solution for cabinet/speaker emulation, look forward to trying it. Hope the CPU hit isn't insane!

[@yikelu]

@@johnbowman3630 Agreed, it is significantly better, even if the end user is the capturer, as you could swap out cabs and mics without having to re-capture.

[@davidfaustino4476]

Yeah man.. idk. Ive never gotten tones that bright with a real mic on a cab. And I primarily use a Sennheiser e906 pointed right at the cone of a V30

[@Fernando-ry5qt]

What? really? I've definitively got ice picky bright tones from a SM57 straight into the center of a 10 inch speaker

[@rudycastillo6369]

do you got a link for the plugin company

[@stevenkambach]

www.neuralampmodeler.com/

[@meagainokay7673]

Great video. I don't like the UI though as I'm a visual person. I guess that's why I use Amp Room. You can drag the mics around and see what's going on. Maybe that can be phase 2 for the company😊

[@stevenkambach]

The UI and the profiles are two different companies. That type of tech isn't developed for profiles just yet. I believe some are working on liquid profiles that will put together a bunch of profiles and have AI fill in the gaps to have a similar experience that you are looking for, but that is still a ways out as far as profiling goes

[@maurizioviera]

Sounds great! Just a noob question. I see you have your Amp DI on one track and the mics (NAMs) on individual tracks, how do you route all your tracks in your DAW? Thanks!

[@stevenkambach]

I have the guitar armed on one track with the NAM amp profiles. I then don't send that output to the master track, but send that to the tracks with the NAM speaker profiles. If you do one amp and one speaker you can do it on one track, but if you want multiple speakers you have to route it similar to this so you don't have one speaker profile feeding into the next one. Does that make sense?

[@philippgrunert8776]

So... Where is the data? We need a comparison of regular IR, IR with these algorithms Bogren and other companies use to get the dynamic effect of speakers and these profiles. Does is sound better in a mix? Are there significant differences. Is it worth the extra cost performance wise? It is interesting for sure, and I am genuinely interested

[@stevenkambach]

Check out the deep dive (part 1 of 3) linked in the description for an in depth look at these. Neutron also responded to one of the comments with a book about the accuracy measured from their null testing if that suits your fancy 👌

[@philippgrunert8776]

I am watching the content on their side. Thank you for pointing this out and sharing the info. Looks very promising indeed! Now I would need info on how this was done, so I could make profiles from my speakers and speaker combinations that suit my needs.

[@stevenkambach]

👍

[@InTheSh8]

Interesting. If you use full-rig profiles you kinda have that, too. Of course you cannot combine or swap, but I would say it is as realistic as it gets and the AI is gonna get even better at understanding how a full-rig behaves dynamically.

[@stevenkambach]

You do! I never enjoyed the full rig stuff for some reason tho. It only goes up from here 👍

[@JeramyRichter]

Quantum Speaker Profiles are LEGIT

[@stevenkambach]

Right? It took me a minute to warm up to them, but good golly they are great 👍 everything being in phase with each other takes them to the next level 🚀

[@PASHKULI]

terrible 3:40 min. crackles and dist. harsh regardless of "filters"… something is wrong

[@stevenkambach]

Clipping on the recording. My bad 🙏

[@IvanEinstein]

How could you use these with modelers live? Like a helix for example. Or is it only possible with the dime head nam device?

[@stevenkambach]

At the moment, nothing runs them effectively outside of a computer. I think the DimeHead has space for one NAM profile at a time so you would have to pick between an amp profile or speaker profile (I could be wrong. Maybe there is a slot for a pedal profile and an amp profile you could use). These are a bit ahead of their time. Someone on FB said they have made capture of NAM captures so they can run them on their quad cortex. The helix doesn't use profile technology at all as far as I know

[@IvanEinstein]

@@stevenkambach gotcha! Well sounds like dimehead with speaker profile could be a paired option for any modeler. Not to mention a daw option on a laptop for live

[@stuartbradbury944]

@@stevenkambach actually the dimehead now can run 2x full NAM profiles.. they updated it :)

[@dominikpohli2381]

is this a profile of a celestion greenback

[@stevenkambach]

Yes, with various mics and positions

[@RobbDizzl]

Dynamic IRs or speaker captures are an interesting concept I've been thinking about for a while. But I also think that IRs are already 98% there. So is it worth the squeeze? Most people, and by that I mean guitar players, won't be able to spot the difference. Maybe it just feels a little different/better, which could be good for the player.

[@stevenkambach]

IRs I think come out to 80% ish based on the same testing. If that works for toy, then great! If you want some more accurate or you don't think the IRs cut it, then these may be an option. I agree that the audience will probably never know 👍

[@djabthrash]

03:30 sounds like there is something wrong in your session here... It is crackling all over the place... CPU issue ? Clipping ?

[@stevenkambach]

Little but of clipping 👍 my bad 🙏

[@mattg082]

Looking at getting an external IR after this year, already have load boxes…but 40 tracks deep I hate doing overdubs monitoring in a DAW. I much rather use a hardware loader with quality AD/DA going into the preamp vs software after with an amp. That way I get direct monitoring, and less nudging or playing with buffer size, freezing tracks if I want to overdub later. Any hardware that will load these like you can with IR hardware? Or is if in just VST currently? I like using this stuff on demos with amp captures…but still like using my real amp on a load box either way via hardware loader for direct monitoring no latency worries.

[@stevenkambach]

I can only think of one hardware unit that loads NAM files at the moment, and that is not really what you are looking for (I think). Its the DimeHead NAM player. It is similar in function to a ToneX pedal, but for NAM users. I get wanting some hardware though! I miss having that option once in a while

[@AnimeAsLeaders998]

about the blending. I never know that 1 NAM amp can be fed through to 2 NAM profiles. was it placed parallel to the speaker profiles or in series? I am not sure how to do this on cakewalk

[@stevenkambach]

I am not sure how to do it in Cakewalk, but I will describe how I did it in Reaper. It is a little bit complicated, so I apologize in advance 👍 I have 3 tracks for this. The first has the guitar input and a NAM with the amp profile. The other 2 have NAMs with speaker profiles. The guitar track is armed and routed so the audio goes to the speaker profile tracks, but not the master track. All I am hearing are the speaker profile tracks. You can make as many speaker tracks as you want this way and can adjust the panning, volume, or add other plugins as needed. Hope this makes sense 🙏

[@timrichard]

If the DAW isn't helpful for running these profiles in parallel, it might be worth looking at Blue Cat Audio's PatchWork plugin. Their Black Friday deals aren't far away... DDMF MetaPlugin is an alternative. Of course if you are open to switching DAWs, then Ableton makes this ridiculously easy with Audio Racks, which can have multiple chains with a built-in mixer. You could arrange your NAM instances in series or parallel with the blend set up, save as a preset, and just click to insert that rack later as an insert plugin on any track (although the same should also be possible with PatchWork and MetaPlugin).

[@stevenkambach]

All good ideas. There is no one right way to do it. Just do what is best for you 👍

[@AnimeAsLeaders998]

@@stevenkambach I managed to sort this out. I had to create separate buses for each speaker profile and mute the origina signal. it works fine.

[@2Funky4me]

It looks similar to the Torpedo Wall of Sound.

[@stevenkambach]

Maybe?!? I haven't used that. I believe that came out before all the profiling hype so I don't know if they are anything beyond IRs

[@djabthrash]

It is way different. The torpedo thing is slightly advanced IRs where you can move the mic position, blend mics, etc. This is a capture, reacting to volume and supposedly more accurate, and with a fixed mic position for each nam file.

[@xsonicassassinx]

A lot of that sounded really really bad. Micing my cab never sounded that cartoonishly bright or dark. Maybe that professional engineer who made these is going deaf

[@stevenkambach]

Not for you then 👍 looking forward to checking out your amazing tones 🙏

[@jeffallen3382]

I agree... Once he added that 1st amp with the crunch channel that didn't sound very good. I don't understand why he is getting so defensive...

[@Bryt25]

Try to relax and leave some breathing space maybe, you sound manic or coked up :-) It is distracting.

[@stevenkambach]

Watch somebody else 👍

[@21stcenturymetal31]

No it won't it just add a new flavor but IR's are alive and well and that's not going to change any time soon.

[@stevenkambach]

👍