Designing and building VUIs (Voice User Interfaces) has its challenges - especially when they are built for events. I created the concepts for the Amazon Alexa skills for the conferences re:publica and Tech Open Air and give an overview over the pros and cons of VUIs in general, the concept process of the event skills and decisions on the feature set.
4. Who am I?
• Nelli Hergenröther, MD of
Evenly, a software consultancy in
Berlin-Kreuzberg
• Focus on mobile, TV, voice
software and wearables
• Background in project and
product management
• nelli@evenly.io & @no_vem_ber
5. What’s this all about?
• We built an Amazon Alexa
skill for the conferences
re:publica and the Tech Open
Air
• I created the concept and
decided on the feature scope
for the first version together
with developers
• Voice UIs for live events have
their own specific challenges
10. Amazon Alexa conference skill
A conference skill is a great
use case for a Voice UI:
• Stationary device
• No privacy sensitive
information
• Use case: Getting, not
giving infos
• Short infos (locations, dates
etc.) needed
• Play audio while doing
something else
12. The three states
Before conference
• Infos about conference and session dates, location
• No recordings or live sessions yet
During conference
• Dates and times of sessions
• Live session streams
• Already a few recordings available
After conference
• Infos about location or session dates are not relevant
anymore
• Recordings are available
15. Confirmation strategies
User needs to know that
• Task was successful
• Correct task will be done
To consider
• Cost of getting it wrong
• Available modalities for
feedback
Confirmation method
16. Explicit and implicit
confirmation
• Explicit confirmation
Forces the user to confirm the information. For example, “I
think you want to play the live stream of session X. Is that
correct?”
• Implicit confirmation
Lets the user know what was understood, but do not ask
them to confirm. For example, “Playing live stream of
session X". (In this case user should have possibility to
cancel or go back a step)
17. Error handling
Intent not understood
• „I couldn’t understand you“
• User can try again (implied)
Information not found
• „I couldn’t find session X“
• Info is missing or intent was not understood correctly
• User can try again (implied)
Wrong information found
• Plays session by Anna instead on Hannah
• Implicit confirmation - user notices and can cancel
20. Apps
Localization & Review Process
One app, localized in multiple
languages = one review
process
Voice Skills
Every language of the skill is
handled as an independent
skill in review process
21. Selecting the feature scope
• We focused on a very limited feature set for MVP
• More features (user requested) to be added later
• Feature scope limited by available data
• Possible new features:
• Add favorites
• Recommendations
• Reminders (Reminders API)
• Notifications (ProactiveEvents API)
22. Summing up VUIs for events
• Event skill is great use case for VUI: stationary device, no
privacy sensitive information, short infos, getting, not giving
input, can be used as background audio
• Keep in mind the 3 states: before, during and after
conference
• Make skill relevant for users in all 3 states
• Work with user stories and move on to mapping dialogs
• Use implicit confirmation for most features (cost of failing is
not very high)
• In case of error: prompt user to try again
• Languages are hard, especially when spoken, plan time for it
25. More infos and reading
• Create and publish your own skills with templates: https://
blueprints.amazon.de
—
• Cathy Pearl: Designing Voice User Interfaces: Principles of
Conversational Experiences
• Clifford Nass: Wired for Speech: How Voice Activates and
Advances the Human-Computer Relationship
• James P. Giangola, Jennifer Balogh: Voice User Interface
Design
27. Let’s discuss!
• Do you have experiences with designing/building Voice UIs?
• What were the challenges?
• What solutions did you choose?
• What would be other great use cases for Voice UIs?
• What could be interesting features?