revxstd_demonstrations

Amp Head Capture

NAM

10

11

rgsson

6 months ago

Description

Demonstration of REVxSTD * 1pass_only does not have any pre_emph applied. * 2pass has been trained a second time and with pre_emph that controls higher frequencies better. Less harsh. [Development, TheGearPage] https://www.thegearpage.net/board/index.php?threads/probably-the-best-nam-quality-in-the-world-revxstd.2727799/ [LINK] https://mega.nz/folder/B0kHjYQZ#fLI0_70-QFXrf4cCO7EERA Colab training file (with instructions and steps in it), samples, etc [Github]https://github.com/38github/neural-amp-modeler I have started to implement REVxSTD and REVxLITE 1-pass training with mild pre_emph and hopefully it works. Please comment here or create an ISSUE on Github if something isn't working. Thank you goes out to SlamminMofo, 2dor, Dan Brown...

Makes and Models

REVxSTD Demonstration

Tags

namrevxliterevxstd

Preview

Model
IR
Input

Powered by

Models

Each model captures the gear at specific knob settings.

A2-FullESR: 0.0007

A2-LiteESR: 0.0049

500 Epochs

Sweep

Captured on

Tone3000 - NAM Captures and IRs

A2-FullESR: 0.0028

A2-LiteESR: 0.011

500 Epochs

Sweep

Captured on

Tone3000 - NAM Captures and IRs

A2-FullESR: 0.0014

A2-LiteESR: 0.0086

500 Epochs

Sweep

Captured on

Tone3000 - NAM Captures and IRs

A2-FullESR: 0.0153

A2-LiteESR: 0.0491

500 Epochs

Sweep

Captured on

Tone3000 - NAM Captures and IRs

A2-FullESR: 0.0136

A2-LiteESR: 0.0503

500 Epochs

Sweep

Captured on

Tone3000 - NAM Captures and IRs

A2-FullESR: 0.0066

A2-LiteESR: 0.0218

500 Epochs

Sweep

Captured on

Tone3000 - NAM Captures and IRs

A2-FullESR: 0.005

A2-LiteESR: 0.0196

500 Epochs

Sweep

Captured on

Tone3000 - NAM Captures and IRs

A2-FullESR: 0.0029

A2-LiteESR: 0.0112

500 Epochs

Sweep

Captured on

Tone3000 - NAM Captures and IRs

A2-FullESR: 0.0017

A2-LiteESR: 0.0071

500 Epochs

Sweep

Captured on

Tone3000 - NAM Captures and IRs

A2-FullESR: 0.0023

A2-LiteESR: 0.0086

500 Epochs

Sweep

Captured on

Tone3000 - NAM Captures and IRs

11 Comments

Share your thoughts...
@username to mention • Ctrl+Enter to submit
furakutaruwamugennikurikaesaremasu's avatar
furakutaruwamugennikurikaesaremasu
Hi! Nice job! Did you try your arch with super input (forcing model to autocorrect aliasing while training?)
1
furakutaruwamugennikurikaesaremasu's avatar
furakutaruwamugennikurikaesaremasu
Also I just checked your null testing and their aren’t ood (out of distribution) To make a proper null testing and compare with std and xstd please remake them with an original DI track (also avoiding sine sweeps in that part can better reflect real life cases) But good news I suspect your results would be even better. Do you have an email to talk about that? Mine is yannick.guitarist@gmail.com
0
metsi
Great to see more experimentation with the training! What are the amp models? And what are the differences/improvements over xstd?
0

License: T3K

Users may download and use the data file in software and publish the resulting outputs without royalties or restrictions. However, they may not upload, republish, or distribute the data file without the author's permission.

Use Your Tones

Load NAM files and IR's from TONE3000 onto a free plugin or compatible pedal.

Plugins

NAM
Recommended (free)

Pedals

Chaos Audio
Chaos Audio
Anagram
Darkglass Electronics
Dimehead
Dimehead
TONE3000 Digital Models

NEW TO TONE3000?

Discover hyper-realistic digital models of iconic gear, including the Fender Twin Reverb, Vox AC30, Marshall JCM 800, and thousands more. Getting started is easy—and free!

Discover TONE3000