NEAMPMOD - The Tweed - Initial Release

neampmod

Newbie
Messages
3
Hi,

I have been working on a plugin, The Tweed, for the last few months in an attempt to model a Fender Tweed Deluxe 5e3, primarily as a test plugin for an audio / DSP engine I am working on as a hobby.

the_tweed.png


I have quite a lengthy release note in the GitHub repo which provides mode details (goes into detail on the engine functionality as well).

I have put some NAM captures up of the last few versions, v1.4.15 (with a Magnatone IR from Origin Effects) should give you a decent demo before you decide if you want to bother with the plugin.

The plugin is available in VST3 and CLAP formats.

Please note I have not tested the Windows VST3 and CLAP plugins, only the Linux plugins (in Ardour and Reaper). I currently do not have a Windows-based system available, so if there are any issues please let me know.

I would appreciate any feedback, suggestions or advice.

Cheers,

Dan.
 
Thanks for the SPAM!

@mods How about introducing a ‘no bloody self-promotion in new users' first post’ rule?
Sorry if it comes across that way. Maybe I should have made a few posts before this.

As for the concerns with the GitHub release page I can replace the zips with the .vst3 and .clap files directly if that is better. (I have added the vst3 and clap files to the release page.)

Let me have a think if I want to release the source code, there isn't much beyond the signal chain since the DSP is in the engine which I don't want to release just yet, but I think that'd at least have the benefit of letting people see what is in the vst3/ clap files.
 
Last edited:
I'd be interested in knowing what modelling techniques you're using. This is also something I wanted to dig into eventually. I've been reading up on wave digital filters, but haven't yet done anything in practice.

First post seems fine with me, welcome to the forum!
 
Let me have a think if I want to release the source code, there isn't much beyond the signal chain since the DSP is in the engine which I don't want to release just yet, but I think that'd at least have the benefit of letting people see what is in the vst3/ clap files.
I just assumed that the source was available on Github, and whilst I do understand why you might not want to release it, I think providing it will gain you an awful lot of goodwill!
 
The plugin runs perfectly on my PC. I also gave it a quick test with pluginval and didn't see anything wrong. The only small thing is that the GUI is a bit small. Maybe think about adding a scale option.

I'm really curious about something in the release notes, though. It looks like you're using both ADAA and normal oversampling. Is there a special reason for doing that? Usually, 8x oversampling is more than enough to keep aliasing really low.

As for the plugin not working on older hardware, that's often just an instruction set issue. This is super common if you use newer ones like AVX2. You could try making a version with older instruction sets to support those older machines. Hope this helps!
 
I've migrated my code into the github.com public repository, so if you are interested you can look at src/lib.rs to see the signal chain layout; You'll see a lot of functions from the neampmod-engine Rust crate. gui.rs has the user interface and main.rs is a dummy file (maybe I'll add a standalone option.)

Thanks for the responses so far and confirming that it does run on other machines, that is good to know :)

I will add details later on today regarding the modelling techniques, and about ADAA + oversampling as I would love to get some thoughts on what I've implemented, how it's connected and if there are better methods availabl. I'm still learning this stuff so I will definitely have made some mistakes.

I did have some issues with power tube modelling generating a lot of audio artifacts when I accidentally skipped the oversampling and just thought I'd need them both enabled from then on. Now that I think about this it may have actually be down to the threshold I had around ADAA that would cause a bypass and thus no anti-aliasing at all, so I probably don't need both running. I'll do some testing.

The modelling isn't anything as modern as the black-box neural wavenet / machine learning stuff, and it isn't WDF, at least I don't think it is that. It's really just multi-dimensional (different voltage levels, etc) lookup tables derived from Koren specs for the tube non-linearities, and GLF for the output transformer, with MNA for the tonestacks but that has given me quite a lot of problems. After doing some research it seemed like a lot of the techniques haven't changed much in the last few years but it's more about how components interact, so I have tried to focus heavily on that aspect, again, probably with a lot of misconceptions / mistakes, but the code does work so it's a starting point for sure.
 
I am going to try it. It might be a week before I actually get around to it. As far as using Windows I don’t blame you. I preferred Linux for everyday use but since most modern software is attached to windows I use it.
I hate everything about 11. I got a eBay machine with the correct hardware and 11 loaded. It refused to allow my l6 driver to work no matter how I tried to work around it. I down graded to 10.
 
Back
Top