A smiling man with short dark hair, wearing a purple patterned shirt, is shown in a close-up portrait. The background features a digital or artistic framed design with shades of brown and gold.

I like to keep things fun. I hope this format gives a good impression of my personality.

Please enjoy!

-Ian Coleman

I am a graduate student at Boston University pursuing a Master’s in Engineering in Electrical and Computer Engineering.

My primary interests are in machine learning, artificial intelligence, statistics, signal processing, and software engineering. I enjoy making numbers have meaning and have interest in both statistical analysis of data and quality analysis of data or capture methods.

Most recent project: Transformer Based Text-To-Speech Model

I have found text-to-speech synthesis to be a rewarding project. It has forced me to broach new topics and fortify my understanding across programming practices, signal processing, machine learning, and statistics. I find a great thrill in being able to extract numerical representations of real data. Additionally, I’ve been surrounded by music, recording equipment, and ham radios throughout my life, so this a personally fulfilling project

Plot of training history for the text to speech model

My model is based off of “Neural Speech Synthesis with Transformer Network”. My model utilizes positional encoding by injecting sine and cosine values into the sequence, some simple pre-nets that should not be compromised by the length of input sequence, and a post-net that smooths the decoder output.

This is currently a work in progress and I am troubleshooting some problems. Please visit my GitHub Repo for more information.