VocalPoint, a portable assistive listening device designed to improve speech clarity in noisy, multi-speaker environments for hearing aid users, was my university capstone project.
VocalPoint uses a microphone array and real-time signal processing to isolate and amplify a target speaker's voice while suppressing background noise. The system is built around an ESP32 microcontroller and a Raspberry Pi, utilizing Altium-designed custom PCBs for audio input and output. The device processes audio using digital signal processing (DSP) techniques, including beamforming and noise reduction algorithms, implemented in C++ and Python. Audio data is transmitted via I2S to the Raspberry Pi for processing, and the final output is sent to Bluetooth headphones for a seamless listening experience.
I designed parts of the electrical schematic, and implemented the PCB layout in Altium for a Raspberry Pi CM5 carrier PCB, incportating USB-C Power Delivery, a 3V3 buck converter, XIAO ESP32 for communication with a mobile application, multi-microphone audio sources, and PCIe connectors for impedance controlled, high-speed peripheral communication to a Hailo-8 neural network accelerator.
-
Links
GitHub Repository
-
Date
Sept. 2025 - Apr. 2026