TextFlow: Screenless Access to Non-Visual Smart Messaging
summary
TextFlow generates short, entirely auditory message options from a 4M Weibo corpus using RoBERTa retrieval, ranks them by entailment probability and serves them through a finger-worn device with five tapping gestures. A user model fuses location, time, activity and calendar commitments; a rule-based reasoning model maps contexts to topics. In studies (20 interviews to elicit 17 topics; 10 BVI evaluators, 15 tasks) the system achieves an overall 88.6% task success rate and shows gains in privacy and hands-free efficiency.
AI-written summary · about
chapters (4)
related talks
09mVoice and Touch Based Error-tolerant Multimodal Text Editing and Correction for Smartphones
12mTyping Haptically: Towards Enabling Non-auditory Smartphone Text Entry with Haptic Feedback for B...
08miText: Hands-free Text Entry on an Imaginary Keyboard for Augmented Reality Systems
20mScene text access: a comparison of mobile OCR modalities for blind users
34mAutomating Full-Screen TalkBack for a Government App with Designer Collaboration and AI Agents — a Practical Experience
20mNew User Interface - Part 1
55mDigital Modernization Through Accessibility
19mBSpeak: An Accessible Voice-based Crowdsourcing Marketplace for Low-Income Blind People