HushRounds
A silent dictation app for hospital clinicians and nurses that captures patient notes through a camera-based lip reader, keeping wards quiet and hands-free.
hospital clinicians on rounds who need to chart without typing or talking aloud
- Silent dictation that reads lips from a lanyard webcam into structured SOAP notes
- HIPAA-compliant local processing with no audio recorded
- EHR field auto-fill for Epic, Cerner, and Meditech templates
- Accent and quiet-mouth personalization tuned over a one-minute calibration
Recent breakthroughs show tongue and lip movements can be decoded with reasonable WER, and clinicians are desperate for silent input on quiet wards and during isolation care.
Nurses spend ~1/3 of a 12-hr shift on documentation and documentation is the top driver of physician burnout (Tebra 2025), with explicit nursing calls for silent input in multi-bed wards, ICUs, and isolation rooms (Jacques et al. 2025 via O'Hearn).Why EHR documentation is the leading cause of physician burnout ↗Silent Speech in Healthcare Documentation: Exploring Subvocal Voice-to-Text ↗
No clinical-grade camera-lip-reading dictation app found; existing players like LipReadPro target media/accessibility and Lip Reading AI is a consumer app, while clinical incumbents (Clinidict, ambient scribes like DAX/Abridge) rely on audible audio rather than visual lip decoding.Lip Reading AI | Visual Speech Recognition ↗Medical Dictation | Clinidict - Fast, Accurate Clinical Notes ↗Lip Reading AI - Apps on Google Play ↗
Hospitals clearly pay for documentation tools (ambient scribes sell at ~$1k+/clinician/yr) and EHR dictation is an established budget line, but a camera-based silent modality is unproven and would need HIPAA + likely FDA SaMD clearance before procurement.The Power of AI-Powered Voice Transcription in Clinical Documentation ↗
Quiet-ward and isolation-care needs persist, but ambient AI scribes are scaling fast (JMIR 2025) and audio ASR WER dwarfs visual lip-reading WER (VALLR/ICCV 2025 notes inherent viseme ambiguity), so the camera modality risks being leapfrogged.AI Scribes in Health Care: Balancing ... - JMIR Medical Informatics ↗VALLR: Visual ASR Language Model for Lip Reading ↗
Visual-only lip reading WER remains substantially worse than audio ASR due to viseme ambiguity, and a clinician-facing camera raises ergonomics, infection-control, and HIPAA concerns; medical deployment likely requires FDA SaMD review and tight EHR integration.VALLR: Visual ASR Language Model for Lip Reading ↗Automatic visual lip reading: A comparative review of machine-learning approaches ↗