Voice-Over Text Editing

Exploring Novel Interface Friction Points

Dorothy-Interface.png
 

Architecting the Multi-channel experience

When using a Voice User Interface (VUI), would people prefer numbered lines to unnumbered lines?

This is a simple question, but difficult to answer without a functional VUI. Instead of embarking on a multi-month adventure to create a functional prototype, we instead employed "Wizard of Oz" prototyping. Instead of a functional backend, a wizard (non-magical classmate) is the backend!

We first started off by designing the architecture of our interface, and how it might look. We wanted something that was simple so that we could build it easily, and the user would find it easy to use.

 
Basic system architecture.

Basic system architecture.

 
 

Sketching the Interfaces

After establishing our architecture, we moved into defining more details with the interfaces. Because the project time constraints, we skipped wireframing and moved straight into development.

 
Initial interface sketches.

Initial interface sketches.

 

Creating A functional interface

The Dorothy interface would be used by the unsuspecting user to format their text through their voice, not their hands. The Oz interface would allow hidden wizards to modify the user-specified text without requiring a fully developed speech recognition engine.

 
Dorothy user interface.

Dorothy user interface.

Oz user interface.

Oz user interface.

 

Testing the (fake) Voice interface

After finding five participants, we moved to test a simple hypothesis for this project: users would prefer line numbers over unnumbered lines.

Our prototype was wildly successful, and we did not reject our hypothesis. Additionally, we found a couple other findings:

  • Users would have appreciated a progress indicator to show the interface had received their command.
  • People said that they would not use the interface in non-private environments like work or a cafe.
 
"Dorothy" and the facilitator.

"Dorothy" and the facilitator.

The "wizards" actually doing the backend work.

The "wizards" actually doing the backend work.