Designing a Discrete Speaker Recognition App for Conversational Support on Smartwatches
For my MRes, I investigated a conversational speaker-recognition system that infers the speaker’s identity from their voice to help those who find recalling names in conversation difficult or those with prosopagnosia. I developed Pwy (Welsh for “Who” as in “Pwy ydych chi?” or “Who are you?”), an iOS and watchOS app that can identify who is speaking based on a speaker’s “vocal fingerprint”. I also realised during my research that individuals who suffer from social anxiety would find Pwy helpful in instigating and sustaining conversations.
I ran three stages of research: a design session with students in a user experience course, participatory design workshops with expert designers, people who find socialising difficult and people with a traumatic brain injury. The findings at each stage were then used to inform the next phase of the design process.
I ran three stages of research: a design session with students in a user experience course, participatory design workshops with expert designers, people who find socialising difficult and people with a traumatic brain injury. The findings at each stage were then used to inform the next phase of the design process. I carried out design theatre, scenario cards, and a design critique in the participatory design workshops. I found that each design group had their own distinct needs, with each group having a different attitude to failure, privacy and features. For example, expert designers and those who found socialising difficult were concerned with failure. In contrast, those with traumatic brain injury would be happy to make their judgment and point toward the app’s failure. Further, I ran a Wizard of OZ study with participants in a scenario where they did not know who they were speaking to but felt they did know them. Participants met five actors each, who acted as if they knew the individual. To emulate the experience, I modified Pwy to allow me to control it from my phone. There was no existing prototyping suite we had access to at the time for the Apple Watch that met my requirements. Participants were under the illusion that the app worked. I found delays impacted the quality of conversation, the usefulness of the prompts and the watch.