Project CapisciAMe
summary
CapisciAMe applies machine learning to build highly personalized speech recognition for people with dysarthria, training neural-network models on voice samples collected from each user. The project provides a free smartphone app to guide non‑tiring recording of limited-word corpora (Italian and English), and integrates the trained on‑device models with OpenHab and Google Assistant for smart‑home control; extensions include embedded real‑time edge deployment for games/robots and research into speech‑therapy applications.
AI-written summary · about
chapters (5)
- 0:00Project CapisciAMe introduction
- 27:46Neural model development
- 32:52Voice donation appeal
- 37:33Linguistic challenges and future apps
- 42:52Acknowledgments and conclusion
related talks
36mVoice Enabling Mobile Apps and Custom Hardware Devices
26mProject Voice - Building Voices for All
23mDesigning for Accessibility
06mSensing to Hear: Speech Enhancement for Mobile Devices Using Acoustic Signals
01h 06mAI as a Force for Innovation
20mLip-Interact: Improving Mobile Device Interaction with Silent Speech Commands
42mdata.bythebay.io: Shubho Sengupta,
45mSpeechMate: A System for Dictating Mathematics