Domenico Stefani research and coding website, Audio Programming page.

:::::::::   ::::::::  ::::::::: ::::::::::: :::::::::: ::::::::  :::        ::::::::::: ::::::::  
:+:    :+: :+:    :+: :+:    :+:    :+:     :+:       :+:    :+: :+:            :+:    :+:    :+: 
+:+    +:+ +:+    +:+ +:+    +:+    +:+     +:+       +:+    +:+ +:+            +:+    +:+    +:+ 
+#++:++#+  +#+    +:+ +#++:++#:     +#+     :#::+::#  +#+    +:+ +#+            +#+    +#+    +:+ 
+#+        +#+    +#+ +#+    +#+    +#+     +#+       +#+    +#+ +#+            +#+    +#+    +#+ 
#+#        #+#    #+# #+#    #+#    #+#     #+#       #+#    #+# #+#            #+#    #+#    #+# 
###         ########  ###    ###    ###     ###        ########  ########## ########### ########

Audio Programming Projects ↑ Back to the top

At the moment, I'm working on a multisource 6 Degrees of freedom convolution plugin to extend the SPARTA 6DoF plugin with the capabilities of the MCFX Convolver engine. You can find the repository here: github.com/domenicostefani/SPARTA_Multi6DoF_project

MCFX Convolver plugin with master gain control and metadata saving functions

I've previously worked at adding some features to the MCFX Convolver plugin, such as master gain control and functions to save parameters inside the metadata of WAV files. See the repository here: github.com/domenicostefani/mcfx-mod

In 2023 I developed a Deep learning audio plugin for emotion classification. The plugin can run on both regular computer and an embedded computer (Raspberry Pi 4 with Elk Audio OS).

The creation and training of the classification model was done by Johan Pauwels and the plugin was developed by me, which included input audio resampling, feature extraction (coherent with the Python training process) and model inference using TensorFlow Lite.
github.com/domenicostefani/music-emotion-classification-vst

A video of the system running on the embedded platform can be found here: youtu.be/MiHcn7VHEHA

Details about the emotion recognition systems are available in the following paper (in press): L.Turchet,D.Stefani,J.Pauwels "Musician-AI partnership mediated by emotionally-aware smart musical instruments", International Journal of Human-Computer Studies

Emotion classification plugin running on a laptop computer

Ultimately, I developed this static website from the ground up with just HTML and CSS (and the tinyest bit of javascript)

Hardware Projects ↑ Back to the top

2020 - Synthesizer game for two

In 2020 a colleague and I developed a "synthesizer game" for two players as a project for the master course "Multisensory Interactive Systems" at the University of Trento. The system was somehow inspired by the Dato Duo synth for two. We however designed a system where one player would play both melody and rhythm on a synthesizer interface, while the other player would have to learn to shake a game controller in time with the rhythm to control a video synthesis program.
We got to experience with:

Puredata (sound synthesis, sequencing, effects and DSP in general)
Raspberry Pi (Running headless Pd)
Open Sound Control (Communication between Video generation computer and synth part)
Teensy (Game controller data acquisition)
Processing 4 (video synthesis)

The interfaces were made of cardboard and components glued together, but we had a lot of fun and learned a lot in the process.

Synth Details

Components:

8-step sequencer, knob-controlled pitch
Sustain-length controlled via a ribbon (softpot)
4-way analog lever controlling a low-pass filter, delay effect and pitch shifter
Main tempo knob
Play/Stop buttons

Repository: github.com/ftrebo/mis1920

2021 - Photography Darkroom Timer

During 2020 I got into analog photography printing and set up a darkroom with black and white basic enlarger and chemicals. My enlarger didn't come with a timer, and I got quickly tired of using a stopwatch to time the exposure of my prints.
So I decided to build a timer myself, which turned out to be much more advanced than I expected, since darkroom timer technology has stopped evolving a while ago. The timer is based on an Arduino Nano, a 4-digit 7-segment display, a rotary encoder, a bunch of buttons and a relay to route AC mains to either the enlarger or the red safelight. The main functions were: It included a focus mode to manually toggle the enlarger on and off.

Count-down timer to precisely time a single exposure,
Count-up timer to experiment more freely with exposure times,
Contact sheet mode: Count-up timer with tunable beep intervals to automatically expose a contact sheet with a set of negatives,
Mask Modes: Count-up timer with precount and beeps to help with manual on-the-fly dodging and burning.

This was probably my hardware-diy project that I ended up using the most, and it was a great learning experience.

Sine icon by Icons8