Prototype roadmap
A flagship system, broken down into what a serious first prototype can actually prove.
System architecture
The launch form stays fixed: four ultrasonic emitters, one sensing node, and one edge compute core. The geometry can start on desktop stands or a ceiling frame, but the control logic stays the same.
Inside-bubble clarity
Engineering target: measurable target-speaker SNR lift versus room baseline.
Outside-bubble intelligibility drop
Engineering target: outside listeners lose sentence intelligibility before raw audibility.
Re-lock after motion
Engineering target: the system reacquires the pair quickly after head shifts or posture changes.
Safe power fallback
Engineering target: explicit low-output fallback when tracking, calibration, or exposure certainty degrades.
Testable first milestone
Build a constrained two-seat setup, calibrate the emitters, enroll two voices, and validate that the pair hears more useful speech than nearby listeners do. Then pressure-test tracking loss and exposure fallback.
Prototype now
01- Desktop or ceiling-mounted ultrasonic nodes with constrained seating geometry
- Target-speaker extraction for two enrolled users
- Tracking-aware power reduction and explicit confidence UI
Mid-term productization
02- Better room calibration and lower leakage through array optimization
- More resilient re-lock behavior in motion and reverberation
- Industrial acoustic safety measurement and certification workflow
Far-term research
03- Moving from constrained seating to freer crowd navigation
- Multi-bubble orchestration in one room without unstable interference
- Shrinking the hardware footprint toward furniture-scale or wearable adjuncts