shawn/op-pedal
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

What PiPedal Is

Browse Source
This commit is contained in:
Robin Davies
2024-10-10 08:09:44 -04:00
parent c02f3d7def
commit 3c99f9f221
6 changed files with 59 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/ChoosingAUsbAudioAdapter.md
+1 -1
View File
@@ -13,7 +13,7 @@ Cheap USB audio adapters, claim to to support 24-bit audio, and happily provide
You will get decent results; but not great results. Stepping up to a more expensive USB adapter dramatically improves the quality of signal you're going to get. With a decent USB audio adapters, some of the state-of-the art Machine Learning plugin produce results that are as good as or better the Helix or Kemperer amp emulations.
I think it's worthwhile, at this point, to make a brief discursus into the state of the art when it comes to machine learning models of guitar amps. There are two major Machine Learning amp simulators that are in the public domain. Jatin Chowdry produced some of the first really impressive Machine Learning amp simulators (I beleive) as part of his PhD thesis. Jatin Chowdry's ML amp simulation library (which was graciously provided under an open-source MIT license) forms the core of the ToobML plugin included with PiPedal. Steven Atkins' Neural Amp Modeler library (also provided under an MIT license) is the other major ML implementation, that implements a variety of Machine Learning algorithms that have been developed since Jatin Chowdry's initial publication of his results. TooB Neural Amp Modeler uses Steven Atkins' Neural Amp Modeler library. And both have large community-developed libaries of amp models. The quality of amp simulations produced by both of these libraries is breathtaking. As I said previously, as good as or better than Helix or Kemperer amp emulations.
I think it's worthwhile, at this point, to make a brief discursus into the state of the art when it comes to machine learning models of guitar amps. There are two major Machine Learning amp simulators that are in the public domain. Jatin Chowdhury produced some of the first really impressive Machine Learning amp simulators (I beleive) as part of his PhD thesis. Jatin Chowdhury's ML amp simulation library (which was graciously provided under an open-source MIT license) forms the core of the ToobML plugin included with PiPedal. Steven Atkins' Neural Amp Modeler library (also provided under an MIT license) is the other major ML implementation, that implements a variety of Machine Learning algorithms that have been developed since Jatin Chowdhury's initial publication of his results. TooB Neural Amp Modeler uses Steven Atkins' Neural Amp Modeler library. And both have large community-developed libaries of amp models. The quality of amp simulations produced by both of these libraries is breathtaking. As I said previously, as good as or better than Helix or Kemperer amp emulations.
But to get Helix-quality (or better) results, you need a good USB adapter. For amp simulators, particulary, every extra bit of input signal is precious.