Subvocal adds silent speech at 200+ WPM

The compelling idea is not faster typing—it is voice-speed computer input without making the room part of the conversation.

• –Silent input could make coding agents practical on trains, in offices, and during meetings
• –The 200+ WPM claim comes from prototypes; accuracy, calibration, and cross-user performance remain the real test
• –Subvocal says its sensing architecture is not yet public, so claims about radar should be treated cautiously
• –An input-layer approach could work across existing apps instead of locking users into a standalone assistant
• –The privacy benefit is obvious, but physiological signal handling will become a major trust and security issue