I thought the same... Tried GT1000 which was the king of latency. The sound and feel > latency. Trust me. Although I hate latency so much I cannot enjoy plugins for longer periods of time ;p
Thanks Leo. I hope these impressive results make the "big boys" ( Tonex, Quad Cortex, Fractal, Kemper, Helix ) urgently reassess and redesign their units.
@@johndo9648 don't criticise what you don't understand. This level can be noticeable. In my setup with near-field active studio monitors (which have their own a/d-d/a processing and thus latency) I can feel an additional 3ms, which is why I have tried but sold the named units but kept a Boss.
I get the feeling that we are going to see a bunch of NAM units in the future. That's a good thing for us. It may become the standard since it's open source and just getting started. I can see all the Mooer/Hotone/NUX/Valeton type companies putting this technology into sub $200 devices in the coming year or two.
Although they will NEVER achieve that grade of quality, be it in parts, audio or the software. Oh yeah, and let the copy cats port the NAM code to DSP, hahaha
Great work. When I saw that the graph you put went up to 3ms I thought oh no! But well done it kept me on my toes and didn't spoil it if you only had 1ms. Soon your graph may only need to be 0.1ms, 0.2ms etc
Leo delivers as always/ It would be nice to see how it handles a whole signal chain (pedals; front and fx loop) with real cab. Connecting it directly to FX loop of an an amp (using as a preamp and omitting the real one from the amp itself)
In my experience, complete NAM amp models also sound good through the FX return/power amp, as long as the power amp is not pushed too far to its limit.
It works absolutely fine, be it lineout w/o IR to power stage of amp (great on my tube amps actually), or with IR to headphones, recording, FR speaker or FOH. It also takes pedals in front very well to drive/push the NAM models!
The latency is not only due to horse power but some algorithm magic which all suppliers can do once they are forced to by competition. As of 4 weeks ago when I last spoke to the Dimehead guys the unit was limited to a subset of the NAM models, namely WaveNet, which is maybe fine for most but maybe not for the nerds. BTW all NAM models could still be substantially sped up by clever pruning, the average model structure is total overkill for the task, WaveNet for instance was invented for speech generation which is a lot more complex than a guitar signal.
We have automatically tested Nam Player with all models from Tonehunt. There was not a single file that was not wavenet based :) Less than 10 files failed, all of which appeared to have been created with a modified trainer.
@@dirkmeyer328 fair enough, but NAM is an open platform, it is just important for people to realize the limitations. The reason many models employ Wavenet is simply because it is the default setting which is because Steve doesn't like LSTMs for their learning problems which on the future could be maybe remedied within the learning cycle, and at a sudden NO Wavenet files would be used anymore.
For the past few days, I've been testing NAM Player and I can only say that my Quad Cortex is going up for sale :) NAM Player is very easy to use, made with top-notch components, and constantly updated with additional parameters, functions, and effects. The NAM technology itself is number one when it comes to capturing. You feel like you're playing on a real amplifier, one to one.
@@christianpucciarelli @LeoGibsonGtr - They say on the website that it's an Arm a72 processor with four real-time dsp cores. (If you look on the main page, and at the image with annotations).
Note to users: Expect to pay extra for a device with virtually no latency. Once the field is highly competitive about DSP the prices will come back into line.
It's not the DSP is beefed up in this thing but the actual code has been improved and optimized. So no, it's not the actual open source code anymore. And that's what you're paying for here.
@@fuzzaldryn9626 That is not correct. Much as your computer needs a video card, the sound processing can only reach virtually zero, as on this item, when a dedicated processor is used. If it was merely code, we wouldn’t see DSP solutions everywhere in the field of audio.
@@fuzzaldryn9626 A>D + D>A alone takes all of the .6 - .8ms delay here. And that’s with 96kHz conversions. The modeling cannot take place in zero time without a dedicated computer doing ONLY the processing. We call them DSPs. They are specialist CPUs that can cache an instruction set and electrically reconfigure to run the required 64 bit process in a handful of clock cycles. 15 years ago they were very expensive. Today they are widely available chips and not terribly pricey either. But it takes work to implement it with a host and we can finally see who’s not doing their job.
As a player who's sensitive to latency these are VERY interesting results. Keep up the good work Leo!
Thank you so much!!!!
I thought the same... Tried GT1000 which was the king of latency. The sound and feel > latency. Trust me. Although I hate latency so much I cannot enjoy plugins for longer periods of time ;p
This is really impressive. Thank you for the detailed test!
You are welcome!
Absolutely fantastic. Thank you so much for such a quick video. You're the best.
Thank you so much!!!!
Thanks Leo. I hope these impressive results make the "big boys" ( Tonex, Quad Cortex, Fractal, Kemper, Helix ) urgently reassess and redesign their units.
Because of 1 or 2millisecond difference. What nonsense.
@@johndo9648 don't criticise what you don't understand. This level can be noticeable. In my setup with near-field active studio monitors (which have their own a/d-d/a processing and thus latency) I can feel an additional 3ms, which is why I have tried but sold the named units but kept a Boss.
Impossible to feel 3ms latency.
If you didn't watch videos showing the different latencies you wouldn't feel the difference.
@@guitartoneSA that's your opinion. It is not fact.
I agree! When you start connecting this to digital mixers and digital in-ear monitoring...
The latency starts adding up!
What an incredible unit. Well done to Dimehead. I wish we could get it in Australia though.
Thank you for sharing!
We have been shipping to Australia for about a week now :)
Seems like a hot topic this one! Thanks Leo❤
You are welcome!!
I get the feeling that we are going to see a bunch of NAM units in the future. That's a good thing for us. It may become the standard since it's open source and just getting started. I can see all the Mooer/Hotone/NUX/Valeton type companies putting this technology into sub $200 devices in the coming year or two.
Yes, I do agree!
Although they will NEVER achieve that grade of quality, be it in parts, audio or the software. Oh yeah, and let the copy cats port the NAM code to DSP, hahaha
Great work. When I saw that the graph you put went up to 3ms I thought oh no! But well done it kept me on my toes and didn't spoil it if you only had 1ms. Soon your graph may only need to be 0.1ms, 0.2ms etc
😀 thank you, I'm happy that you appreciate the video!!!
Hey!! I named my Dog LEO and when the kids say LEEo i say No its LAYOH.
😀😀😀
cant wait for the full demo of that thing...i so wish its gapless
Thank you!
Leo delivers as always/ It would be nice to see how it handles a whole signal chain (pedals; front and fx loop) with real cab. Connecting it directly to FX loop of an an amp (using as a preamp and omitting the real one from the amp itself)
Thank you! Yes, it is possible but creating a profile of a preamp only...
In my experience, complete NAM amp models also sound good through the FX return/power amp, as long as the power amp is not pushed too far to its limit.
It works absolutely fine, be it lineout w/o IR to power stage of amp (great on my tube amps actually), or with IR to headphones, recording, FR speaker or FOH.
It also takes pedals in front very well to drive/push the NAM models!
Great video ❤
Thank you!
The latency is not only due to horse power but some algorithm magic which all suppliers can do once they are forced to by competition. As of 4 weeks ago when I last spoke to the Dimehead guys the unit was limited to a subset of the NAM models, namely WaveNet, which is maybe fine for most but maybe not for the nerds. BTW all NAM models could still be substantially sped up by clever pruning, the average model structure is total overkill for the task, WaveNet for instance was invented for speech generation which is a lot more complex than a guitar signal.
Hi, thank you for sharing!
interesting points...
We have automatically tested Nam Player with all models from Tonehunt. There was not a single file that was not wavenet based :) Less than 10 files failed, all of which appeared to have been created with a modified trainer.
@@dirkmeyer328 fair enough, but NAM is an open platform, it is just important for people to realize the limitations. The reason many models employ Wavenet is simply because it is the default setting which is because Steve doesn't like LSTMs for their learning problems which on the future could be maybe remedied within the learning cycle, and at a sudden NO Wavenet files would be used anymore.
As you can see, we offer firmware updates :) If a new architecture were to prevail, we would also implement it.
@@dirkmeyer328 👍👍👍
Can you remind us where this falls in the overall chart of available modelers?
This is unique...no other device can load complex NAM profiles. Furthermore the latency is first in class...
THanks!
Can't wait for your review and demo song! Sounds like the hype may be real. God bless and rock on 👍😎🎸
Yes, I'm working on the demo song....I can't wait to share it...it's pretty nice...😀
Hype? which hype?
It has been real since day one of the initial idea, knowing the inventors!
👍👍👍
Thank you!
For the past few days, I've been testing NAM Player and I can only say that my Quad Cortex is going up for sale :) NAM Player is very easy to use, made with top-notch components, and constantly updated with additional parameters, functions, and effects. The NAM technology itself is number one when it comes to capturing. You feel like you're playing on a real amplifier, one to one.
... Yeah, testing a new HP/LP feature for fine tuning the IR. That's a major leap in fidelity on different FR output devices.
Thank you for sharing!!
Sempre n1,riusciamo a capire che processore ha ?
Grazie!!! Provo a chiedere ... anche se credo vorranno tenere per se certe specifiche.
@@LeoGibsonGtr beh alla fine è un segreto di pulcinella,il primo smanettone con un cacciavite lo scopre 😜
@@christianpucciarelli @LeoGibsonGtr -
They say on the website that it's an Arm a72 processor with four real-time dsp cores. (If you look on the main page, and at the image with annotations).
@@DomMcsweeney-DoomPlugsuhm I don’t know that processor
I sold my Quadcortex, Nam sound so better that I capture the Nam profiles😂
Thank you for sharing!
Note to users: Expect to pay extra for a device with virtually no latency. Once the field is highly competitive about DSP the prices will come back into line.
It's not the DSP is beefed up in this thing but the actual code has been improved and optimized. So no, it's not the actual open source code anymore. And that's what you're paying for here.
@@fuzzaldryn9626 That is not correct. Much as your computer needs a video card, the sound processing can only reach virtually zero, as on this item, when a dedicated processor is used. If it was merely code, we wouldn’t see DSP solutions everywhere in the field of audio.
@@artysanmobile What if I tell there's no DSP in it doing the NAM modelling?
@@fuzzaldryn9626 You might wish to go to website. In giant letters: QUAD CORE DSP
@@fuzzaldryn9626 A>D + D>A alone takes all of the .6 - .8ms delay here. And that’s with 96kHz conversions. The modeling cannot take place in zero time without a dedicated computer doing ONLY the processing. We call them DSPs. They are specialist CPUs that can cache an instruction set and electrically reconfigure to run the required 64 bit process in a handful of clock cycles. 15 years ago they were very expensive. Today they are widely available chips and not terribly pricey either. But it takes work to implement it with a host and we can finally see who’s not doing their job.