Try It Yourself
Try It Yourself
Hands-on exercises to practice voice interface design:
Test Intent Recognition
Build a simple voice command parser:
Define one intent: “Turn on lights” with 10 phrase variations (“lights on”, “turn the lights on”, “I need light”, etc.). Implement keyword matching vs. intent-based classification. Test with 20 natural phrases from friends/family. Measure: What percentage of natural speech is recognized?
What to observe: Measure how many of your 20 test phrases keyword matching catches; record failures before trying intent patterns. Compare the revised parser on the same 20 phrases; report its observed recognition rate and new failure modes. Notice which phrases break your system—users say “it’s dark” not “activate illumination.”
Audio Feedback for Hearing Loss
Record a confirmation message (“Okay, turning on lights”):
- Version A: High-pitched voice (300-400 Hz female voice)
- Version B: Low-pitched voice (180-220 Hz male voice)
- Test with older adults (60+ years) in noisy environment (TV on at 60 dB)
- Measure comprehension: Can they repeat what the system said?
What to observe: Treat 40–50% missed by Version A as an illustrative test outcome to verify with participants who have age-related hearing loss. Treat 85–90% understood for Version B as an illustrative test outcome to verify in the same cohort. Lower frequencies may help in some settings; test with the intended audience.
Memory Support vs Exact Syntax
Design two error recovery approaches:
- Exact syntax: “Invalid command. Please say ‘set temperature to [number] degrees Fahrenheit’”
- Memory support: “I didn’t catch that. You can say things like ‘too cold’ or ‘warmer’”
Test with users with mild cognitive decline:
What to observe: Treat 70% exact-syntax failure and 90% example-aided success as hypothetical results to measure in the design exercise. Notice how many users try the exact example phrase—that’s why examples matter.