Gary Seath

AI Voice Builder & Hardware Engineer

I design and build practical, real-world AI voice systems with a focus on accessibility, reliability, and usability — particularly for elderly users.

Featured Project: Memory

Memory is an AI voice companion developed specifically for elderly users. It combines embedded hardware with real-time cloud AI to deliver natural, helpful interaction.

• Real-time bidirectional voice streaming using ESP32 + I2S
• Persistent medication reminder system with warm, patient dialogue
• Push-to-Talk (PTT) interface for deliberate and clear interaction
• Custom Python bridge with echo management and low-latency streaming
• Built from the ground up with Gemini Live (Zephyr voice)

I am currently seeking contract, freelance, or project-based opportunities in AI voice, conversational AI, and voice hardware integration — either remotely (UK/US) or on-site/hybrid in Johannesburg.