Live from UXPA: Wear your words: wearables & speech design Kristen Deveau

Wear your words: wearables & speech design, by Kristen Deveau

UXPANuance Communications (where Kristen works) works with speech capability.

Today she’ll be talking about:

  • Speech capability within wearable devices
  • Speech as one of many solutions

Form factors – how do we interact with different devices? Start by thinking about tablets and phones (handset screens). They have screens. Wearables, however, have very small or even no screens.

How do we interact with very small screen devices?

  • User input
    • Tactile input if there’s a screen (tough to navigate between apps though)
    • Hardware button (if applicable)
    • Voice
  • System output
    • Second screen device (like a tethered companion)
    • Vibration patterns
    • LED lights
    • Non TTS (text to speech) tone with an ear connection
    • Appropriate action taken
    • Voice response
  • Example
    • User: I went for a run
    • TTS output: how long?
    • User input: a half hour

Transmodality: Speech as part of a holistic experience

  • Example of combination
    • Speech input (“I exercised yesterday”)
    • Visual output (images to choose one)
    • Tap input (choose one)
    • Vibration to confirm it has accepted the input
  • Example of offering options
    • Screen shows options
    • User can tap or say one of the options
  • What’s most important is to design a speech functionality that works.

Best practices

  • Core functionality: just because you can doesn’t mean you should
    • Functionality is key
    • Unique solution is nice, but not always necessary
    • Select a subset of features
    • Consider the use cases
  • What’s the role of speech within the app?
    • Are you creating a “natural language” app, where the user can use any terms and have a dialogue?
    • Are “structured commands” necessary?
  • Personality
    • Voice creates a personality.
    • Do you want voice output?
    • How does the voice sound?
    • Is it a friend or an assistant?
  • How do users engage speech?
    • We used to have “buttons” or microphone to show that the user should speak.
    • With a tiny screen the whole screen is button-sized. We need to create new standards.
  • Discoverability
    • How does the user know when he/she can speak?
    • Provide verbal cues “would you like me to send this?”
    • Introduce new features “I can send this for you.”
    • Handle errors
  • Situational considerations for TTS output
    • The TTS can be verbose when the user’s outside running, or driving a car
    • TTS should be muted when the user is at a conference or in the doctor’s office
    • Users need to have ability to change settings like this
    • OR the TTS can be smart enough to change settings based on how the user responds. If the user taps instead of speaks, the TTS might respond via vibration, rather than verbal confirmation.
  • Contextual awareness
    • Follow up commands (“text her” implies “text the last person whose name I referenced.”)
    • When we connect devices so as to create a complete experience, we need devices to recognize what other devices are doing.
  • Location awareness
    • Traffic/weather ahead – does that mean that the device does something in particular?
    • Tell the user, for example “weather ahead, so I’ve rerouted you.”
    • Or it might just be an fyi – “You’re out of milk”
  • Proactive speech output considerations
    • Public facing wearables and privacy.
    • If we’re telling a user about a prescription, we need to consider who might be around them.
    • An alarm to “take your medication” might be better as just a ding and an image, instead of a voice saying “it’s time to take your syphilis medication.” (example provided by Marli, not Kristen).
  • Personalize the experience
    • Make assumptions, and allow the user to opt out, if those assumptions are wrong.
    • Assume “John” means “John Smith” every time the user says “John,” until they say “no, John Jones.”
    • The output should mirror the input.
    • Provide proactive information.
    • Greet the user by name.
  • Error handling
    • We can help users by telling them what went wrong.
    • Offer suggestions to try again.
    • Provide examples of prompts that will be successful.
    • Refrain from an endless “try again” loop.
  • Technical considerations
    • Remember the battery consumption of voice recognition software and screen resolution!!

Leave a Reply

Your email address will not be published. Required fields are marked *