PROJECTS

Speech Transmission Index (STI) Estimation for Large Rooms and Reverberant Spaces:

Cut Existing STI predictive models by 50%

Existing research provides a set of equations to estimate STI using the RT60 at 500 Hz, which perform well in small rooms (<350 m³). I implemented these equations in a custom MATLAB tool:

Click Here: STI Estimator

Click Here to See Demo Video: STI Estimator Demo Video

However, these models show a significant drop in accuracy for larger, reverberant spaces, with margins of error exceeding 0.2 STI units. To address this, I collected real-world acoustic data using Room EQ Wizard (REW), SPL meters, architectural blueprints, and MATLAB, then developed a Multi-Layer Perceptron (MLP) model to better estimate STI in large rooms. The code, analysis, statistics, write up of this journey can be found here:

Click Here to Read Thesis: Speech Transmission Index Estimation for Large Room Thesis

GitHub Link to Dataset: UM Concert Hall Room Impulse Response Dataset

Plugins: Compressor / Parametric EQ / Reverb / Delay / More…

Built Realtime C++ Audio Plugins

The bulk of my coursework at the University of Miami centered on building a strong theoretical and practical foundation in digital audio / signal processing. This involved implementing DSP concepts in MATLAB first, for clarity and prototyping, and then porting them into C++ using the JUCE framework for real-time audio plugin development. Below are a few of my favorite DSP projects:

GitHub Link: Linear Interpolation Circular Buffer Delay

GitHub Link: Multi-Effect Parametric EQ with Tremolo

GitHub Link: EQ | Compressor | Reverb Plugin

Inspired by Hack Audio by Eric Tarr, one of my favorite DSP resources, I’m working on translating core projects from the book (originally written in MATLAB) into modern C++ with JUCE. This is an ongoing open-source project, if you are interested in collaborating on this project feel free to reach out to me, my information can be found on my contact page.

GitHub Link: Hack Audio Translation to Real Time C++

Microcontroller Projects:

Participated in a Hackathon to create a musical instrument that used shadows to play notes in an arpeggio.

The following project was a team Hackathon project for the 2024 Synthux-Academy Annual Hackathon. The concept of the 2024 Hackathon was music and human connection, so we built a light controlled arpeggiator powered by an Arduino and coded in PureData.

Acoustics and Transducer Projects:

These two projects showcase my love for hands on work, below is an Acoustic treatment plan and a speaker build I made myself using wood and parts from PartsExpress:

Acoustic Treatment Plan:

Created a treatment plan for a small studio that reduced prominent low frequency room modes long RT60 time.

Click Here to Read the Acoustic Report: Weeks 108 Acoustic Treatment Plan

Low budget acoustic treatment plan that reduced reverb time in low frequencies by 0.5 seconds in the Weeks Studio Psychology Lab on campus at the University of Miami.

Transducer Build:

Improved the Bass response from a PartsExpress speaker by more than 5 dB.

Click Here to Read the Transducer Report: Speaker Build

Custom enclosure for a Parts Express driver. Modeled in WinISD and measured in REW. About 10 dB more low frequency output.