ποΈ Harness real-time multimodal AI for seamless "Vision-to-Voice" interactions using synchronized webcam and audio inputs with Gemini 2.5.