VoiceMem Dual-Brain Architecture for Real-Time Conversational Memory

VoiceMem implements a dual-stream memory architecture using an informational left brain and an emotional right brain for speech language models. The system achieves top-5 retrieval accuracy nearly 30 points higher than Mem0 at top-200 levels while supporting long-horizon affective attribution.