I wanted to share some resources I've found useful for my projects.
Google Actions has a great overview of VUI design. The videos are a good starting point.
Skeuomorphism in voice interface design (for now)
Main takeaways and things to keep in mind:
1. Uncanny valley
The more something sounds like a human, the more we perceive it as a weird or off human and feel uneasy.
There’s only so much we can ingest from a response. A visual list of information we can go back and forth to is easier than trying to remember what we heard. We tire out trying to remember exactly what was said.
3. Recognition verses recall
What people can respond with should be easily available for recall. Don’t put the cognitive load on people by making them remember too many things, 3 things max.
4. Task-focused interactions
It’s better to have people focus on simple requests and tasks and be explicit about what can be done. Information exploration that’s possible in visual interfaces doesn’t translate as well to voice. The assistant needs to be clear about what function it's performing.
5. Complementary visual interfaces
Visual additions are great for extra information that wouldn’t be easily absorbed from a voice response.
Success depends on the device being able to work with differing speech patterns, high pitched and low voices, accents and speech impairments, soft voices and loud voices, etc.
The system should be transparent about its status and errors and what we can do to get unstuck.
Published by: reitmane in UX, product design