NNMPLIFIER
Individual work (academic)
Bob Tianqi Wei
Tutor: Pierluigi Dalla Rosa
2024.1-2024.2
Nnmplifier is an assignment for an old appliance remodeling project. I purchased a guitar amplifier at a thrift store and remodeled it to allow the use of nn~ to run sound related neural network models used to process audio in real time as an effector. For this project I used a microcontroller and Max MSP and 3D printed the product case.
In the video, I used a neural network model called wheel that can describe sounds in words and recite them phonetically. There are five wheels running locally via nn~ and three knobs that control the overall volume, the model output volume, and the number of models running simultaneously.
nn~ provides the basis for running many neural network models in this project, so I named this project Nnplifier.
This is a guitar amp I purchased on theused market. In the DET class, our goal is to use emerging technologies totransform used appliances. I hope to transform this guitar amplifier.
The disassembled structure is shown. Isoldered new potentiometers and installed new jacks on them, using amicrocontroller to send their serial data to the computer for the next step.
Ultimately I decided to make this guitaramplifier capable of running neural network models used to process sound. Iused an external called nn~ on the max for running the modellocally in real time to process sounds. Here I used a model called wheel thatrecognizes the sounds and describes them in text, and the described text I readthem out immediately using machine speech. Interestingly, what they read aloudis not even a language, I processed it as a sound effect.
I also modeled and printed an enclosurefor mounting this hardware. In the end, it looks great.
