Dimehead NAM Player

Are you familiar with Reamping?
After that it's simple enough to export an audio file, and run the trainer.
If you need more info, again, don't be a stranger and pop a question over at the NAM community and everyone can try to help.
 
Does the Dimehead play the ultra accuracy NAM captures?
You'd have to try them out (I assume you're talking about @Slammin Mofo's "oversized" profiles). In general, I still think the standard models provide a solid profiling quality and can null out pretty damn well against the source.

Oversized models make a lot of sense on profiles which are harder to reach the ESR threshold of 0.01 which is the widely accepted "great" quality standard for NAM profiles. Very hi-gain tones, with a lot of high-end, fuzzy etc might benefit from a larger architecture.

The training time (epochs) doesn't necessarily translate to improved accuracy because of how the training process works; it will eventually plateau because the neural network weights will not go through massive weight swings in order to settle on the final result.

What I'm trying to say is the ESR of the profile is what you're interested in if accuracy is feeding your OCD. If the mode's ESR is below the widely accepted 0.01 threshold, then just play & enjoy - you're not missing out.
 
I read the Dimehead blerb on their website and it says it will play all NAM profiles so I just wondered whether SMs ultra accuracy captures are included in that. For live performance I am sure the thing sounds fine either way.
 
I read the Dimehead blerb on their website and it says it will play all NAM profiles so I just wondered whether SMs ultra accuracy captures are included in that. For live performance I am sure the thing sounds fine either way.
There's only 1 way to find out.
 
It will play nams with standard architecture - I don't think nano will work for example, more would any bigger models
Why would it not be able to run the nano ones? They're still Wavenet and lighter in requirements. Legitimately curious (not challenging or confrontational).
 
I think the Dimehead should be able to run them. I had a recent pack put out with profiles that use a modified (slightly larger than standard) architecture because I wanted to go below the 0.01 ESR threshold.
One of the guys in the NAM Facebook group confirmed he ran these profiles on his Dimehead NAM player and they worked fine for him.
 
@Antipodes -

Dimehead NAM Player will run models made using the provided architectures in the unmodified trainer, eg. nano / feather / lite / standard.
It won't run models that have more complex architectures (higher than standard models), nor will it run anything over 48khz sample rate.

I think there are very few 'hyper' models available on Tonehunt, and you'll probably find what is there to be too costly to run in real time in most DAW sessions. They are probably much better for mixing, rather than tracking / jamming. I believe the author of those models also provides 'standard' architecture versions, for those use cases, and those will work in the Dimehead NAM Player.
 
Dimehead NAM Player will run models made using the provided architectures in the unmodified trainer, eg. nano / feather / lite / standard.

It won't run models that have more complex architectures (higher than standard models), nor will it run anything over 48khz sample rate.
It looks like it can run larger-than-standard models (snippet from the Facebook group):

1726820604796.png
 
That comment is under your post in the group, and as far as I can tell they should work because those are standard architecture / 48khz right?
That's correct but those models are slightly larger than standard; still 48Khz.

So, it looks like the Dime player *can* handle larger models but I guess it's up to the user to test / confirm.
 
Last edited:
Ah ok, they aren't alot more costly than standard, so probably will work, but those might not work if you use dual slots and a full fx chain. I haven't tested, but I'd assume that'd be pushing the limits.

The 'hyper accuracy' models made by Slammin' Mofo, are about twice the size again so probably aren't going to work.
 
Ah ok, they aren't alot more costly than standard, so probably will work, but those might not work if you use dual slots and a full fx chain. I haven't tested, but I'd assume that'd be pushing the limits.

The 'hyper accuracy' models made by Slammin' Mofo, are about twice the size again so probably aren't going to work.
Yeah - I experimented a bit to try and get something that won't go crazy on resources (probably 2%-3% more RTCPU when monitored in Reaper), train as fast as standard but can squeeze better, lower ESR values (target usually under the 0.01 threshold, usually anywhere between it and 0.005).

That said, another impressive feature of the Dimehead player is it can load larger-than-standard models but don't think anybody can guarantee what its "breakdown" threshold is; probably good material for a vid @PippPriss
:sofa
 
Yeah, it's definitely cool to know.
I've relayed that all to Dirk, so he's aware of it.
Perhaps it'd be a good thing, if you don't mind, to edit any details about that into your upload description on Tonehunt *and on facebook post?
Could be handy to know in future, especially if someone asks 'why is my cpu usage higher than normal?'. =)
 
@2dor Do you see yourself picking up the DH Player in future?

I'm asking because I value your opinion and I remain curious about this device.
 
@2dor Do you see yourself picking up the DH Player in future?

I'm asking because I value your opinion and I remain curious about this device.
Thank you for the kind words - that's flattering (for real).

I very well might. I'm actually watching the unit and its development updates closely.

Initially, the lack of an FX loop, proper input gain rating & file management was kind of a bummer but in the last firmware things took a turn for the better: that line out now can be used as an FX Send/Return and the ability to run 2 NAM profiles (stacked / in series) at the same time was impressive to see.

What I'd like to see (I've exchanged a few messages with Fabian from Dimehead earlier this year) is a "max input gain" rating for the guitar input & similar scaling much like we do on our audio interfaces with NAM.
That way, folks would be able to make use of that dBu rating some profile makers like Slammin Mofo, MirrorProfiles and others publish.
Dimehead hinted they're looking into this.
 
Last edited:
Back
Top