Students Present at Houghton University’s Second Annual Undergraduate Research Day! 🎉

Exciting news from our research group! 🎓
Two of our students, Eric Waterhouse and Avery Belanger, presented their ongoing projects at Houghton University’s Second Annual Undergraduate Research Day, held on April 29, 2025. ✨

The event showcased innovative undergraduate research across disciplines, and our students contributed two technology-driven projects aimed at improving educational access and engagement:

Eric Waterhouse and Babafemi G. Sorinolu (2025).

“Voice-Enabled Approach for Thoughtful LLM Use in Education.”

Abstract
With the increasing use of LLMs such as ChatGPT among students and educators, it provides students with a supportive learning tool. However, with the rapid responses provided by LLMs, we risk diminishing the learning process by encouraging student dependence on instant solutions. We propose a voice-assistant that promotes the intelligent use of LLMs, preserving opportunities for deep, focused learning while still benefiting from their capabilities. We anticipate this device to promote thoughtful usage and learning while the student listens to the audio feedback. This device will be developed using a Raspberry Pi 4 and Python API as backend to process the audio input using OpenAI’s Whisper and Meta’s LLaMa. We expect this device to record classroom lectures, transcribe, summarize, and give responses to students’ queries on a broad variety of topics.

Avery Belanger and Babafemi G. Sorinolu (2025).

“A Low-Cost Solution for Accessible Audio Transcription and Language Translation.”

Abstract
Audio Transcription technology has recently become more accurate with improvements in the field of Natural language processing (NLP) and deep learning. This technology has provided access and better community integration for people with translation needs or those who are deaf and hard of hearing. However, available technology on phones and audio transcription devices today can be expensive and unaffordable for some.

We propose a low-cost prototype audio transcription and language translation device using a Raspberry Pi. This device will be able to process audio inputs using the SpeechRecognition API to achieve its multilingual audio transcription, utilize the Meta AI (LLaMA) Large Language Model to translate the text to the users choice language, and display the text on an accessible user interface. We expect this device to enhance communication among people of different languages or hearing impaired communities.

📸 Here are some highlights from the conference:

It was inspiring to see our students confidently present their creative ideas and early results to peers and faculty. 🌟
We’re excited to see how these projects evolve as they continue their research journey!