It's what I learned on, and there seems to be more info available out there about it (for now), but capacitive touch CYD would probably be better/more responsive. The resistive touch option seems to be a few dollars cheaper in general on Amazon.
That's really cool! Starred so I can find it later, might be challenging putting it on the esp32 but if there's an easy way to downsize the images then it should be do-able. Thanks!
A video about a fully open source guitar pedal with features similar to commercial high end digital effects. Discusses the project and demos various effects.
Flexibility and affordability. As a guitar player, if I walked into a room and on one side there’s a perfect digital emulation of a Marshall stack, and on the other side there’s and actual Marshall stack, I’d go for the real thing every time. But the reality for most musicians, myself included, is that I wouldn't be able to or wouldn’t be willing to pay that much for one. Now if I had a whole library of digital amps/pedals to try, then I could find something I like best and go get the real thing. I don’t see digital as replacing analog, only enhancing it. It’s also nice to be able to plug in headphones, and from a laptop or pedal that takes up much less space.
But as an engineer I just nerd out over all of it. Analog. Digital. If it makes good music it’s all cool to me.
But I will say that there is a project out there for time varying effects, but haven’t dug into it yet. I believe it’s a different method than the LSTM
It can’t do time based effects such as delay, reverb, flange, chorus, etc. The LSTM can model distortion, overdrive, compression to an extent, and amp circuits (including vacuum tubes).
As the creator of this project I can assure you that the audio used here is at least CD quality (44.1kHz 16bit). With the HiFiBerry hat the digital audio comes in at 24bit/192kHz. The NeuralPi DSP processes the audio at 44.1kHz with 32 bit floating point precision. No reason the sample rate can’t be higher though. Elk OS claims latency is less than 1ms, but I’d like to test and see exactly what the latency is running the plugin. As a guitarist, I can’t tell the difference between this and an analog effect.
Good questions, modeling the parameter space for different knobs will be a focus of future work. It will probably be more practical to learn one parameter space, like gain, and leave EQ for traditional algorithms.
No, the training is not accomplished on the Raspberry Pi, I used Google Colab, which has GPU runtimes. I had similar training times with local GPU. I haven't tried it yet, but I suspect any practical training will take too long on the Pi. I'll probably look into recording samples with the Pi/HiFiBerry hat, and then upload the samples to another computer (or the cloud/Google Colab), and then send the trained model back to the Pi. I believe I'll be able to automate that process at some point.
That is a really cool idea, I’m in a local band and we record from our phone if we want a quick way to remember how we played something. Your Band Studio app would add a whole new level to that, especially considering how hard it is to get everyone together in one place. If you wanted to apply a certain sound to the tracks you could use something like what I made on guitar/bass, or possibly a microphone model, in the same way you’d use in on a traditional DAW.
This project uses tensorflow/keras to emulate the sound of real guitar amplifiers and pedals on wav files. The model is trained on about three minutes of input/output guitar audio, and takes a few minutes to train. For example, you can record a $1000 amp and use this code to create a model to apply that sound to other wav files.
The purpose of this project is to improve on the previous WaveNet model built for the same task. Using LSTM is orders of magnitude faster and more accurate for emulating guitar signals than WaveNet. A real time guitar plugin exists for the WaveNet implementation, and a plugin for this the LSTM model is currently in work. Compare to Neural DSP’s Quad Cortex neural capture feature, which uses a similar technique on their $1600 floor modeler.
Nice, I’ll have to look into diffwave, have not heard of that. You may have difficulty training on reverb effects due to the nature of the WaveNet model. It works well for distortion, but for effects that extend past around 50 milliseconds like reverb or delay, it might not be able to capture it properly.
Wow, I am blown away by the feedback and discussion from everyone here. Just want to say that this is an extremely different experience from any other site I’ve been on, so thank you for that. Future goals for this project are to improve the audio quality and get community contributions for model training of different hardware. It’s very much still an experimental project, but I wanted to give an idea of what is possible using deep learning for audio applications.
The Nolly is awesome, tons of options in there for both clean and dirty sounds. The awesome thing is you can try all their plugins before buying to make sure it’s what you need.