Designing for speech
summary
Voice interfaces beat typing speed (40 wpm vs ~120–150 spoken) and ASR reaches human parity (~5.1% error cited for Microsoft), so expect 50% of searches by voice and lots of smart speakers. Design advice includes writing sample dialogues, sketching flow, grouping utterances into intents and slots/entities, handling ambiguity with confidence thresholds and the N-best list, choosing visual vs voice confirmations, and preferring one focused capability over trying to do everything. Real risks: runaway learning (Tay), accidental purchases (child’s dollhouse) and search spoofing (Burger King/Wikipedia).
AI-written summary · about
chapters (7)
related talks
55mDesigning for speech
17mReimagining Customer Experiences Integrating Voice into Digital Strategy
50mSpeak To Me: Voice App Conversation Practices
57mDesigning Human-AI Experience
45mKeynote: Conversations with Machines
01h 04mBuilding Expertise Into Generated Conversations
13mStepWrite: Adaptive Planning for Speech-Driven Text Generation
54mConversational UX—Human Speech Patterns Meet UX