With 8 billion voice assistants in use globally by 2023, the future will be voice-first, but that future doesn’t include people with Down syndrome. Voice technology’s algorithm often doesn’t understand the community’s unique speech patterns, leaving them behind in the voice revolution.

To increase access to Google’s voice technology, we turned the Down syndrome community into Google’s teachers. We worked with Google– a technology that usually teaches us– and empowered people with Down syndrome to use their voice, and voice data, to be the ones teaching Google.

Introducing Project Understood, a campaign that turns people with Down syndrome into Google’s teachers, by recruiting them to donate their voice data to train Google’s speech recognition model to understand them. Making voice technology more inclusive, by including people with Down syndrome in creating the solution. And empowering the community by turning a speech limitation into a valuable data asset to future-proof themselves.

Voice technology requires millions of data points (human voices) to perform optimally. Unfortunately, for those with Down syndrome, the small size of their community means these AI systems lack the data to reliably understand them. CDSS directly recruited the Down syndrome community to collect enough data that Google’s voice algorithm could start to learn the unique patterns in their speech. Data collection was critical to unlocking voice technology for people with Down syndrome – giving them access to life-changing independence.

Google’s voice assistant has the most market share in Canada, yet currently misunderstands about one in every three words of a person with Down syndrome. This made Google the ideal partner for CDSS as they would have the greatest impact on the Down syndrome community in Canada.

Project Understood was a rallying cry for the Down syndrome community to donate their voice to Google and improve voice technology for everyone. The campaign launched during Canadian Down Syndrome Week, with social videos acting as a recruitment tool, mobilizing and empowering the community the community to be part of the solution.

With only $1000 in media, we targeted this niche Down syndrome community organically, knowing the more they engaged, the more we’d reach them. Through email and organic social, we engineered direct outreach to Down syndrome groups across North America to collect voice data, who in turn engaged another 735 international Down syndrome groups to participate.

Participants were driven to Projectunderstood.ca where they received a login to enter Chit Chat– an engine for raw data collection and machine learning, where participants were served pre-determined phrases guided by speech pathologists and Google AI scientists. Chit Chat was designed to capture the speech patterns/characteristics required to train Google’s AI and voice assistant technology.

Earned media further amplified our message, changing perceptions of the community.