Google to Train Voice Models to Understand People With Down Syndrome

Google has announced a partnership with The Canadian Down Syndrome Society to train their technology to better understand people with Down Syndrome.

Due to their unique voice patterns, Google’s voice models don’t always understand people with Down Syndrome. By collecting voice samples and using them to train its AI models, Google and the Canadian Down Syndrome Society are aiming to bring voice-activated technology to a segment of the population that has, to this point, been mostly excluded.

The initiative is called Project Understood, and by visiting its website, volunteers can offer to ‘Donate their Voice’ by recording a phrase designated by Google, and then uploading their sample. The more samples that can be gathered, the better Google can train their models to recognize the unique speech patterns associated with individuals with Down Syndrome.

Digital assistants play a large part in the use of major operating systems, like Google Assistant for Android and Chrome OS, Siri for iOS, and Alexa for Amazon.

The sale of voice-activated devices continues to grow globally, and their use can bring with them a number of benefits to everyday life. Ensuring these devices and their potential benefits are accessible to all segments of the population is a vital part in the progression of technology.

Project Understood’s goal is to obtain 500 donated voice samples, and have so far passed the 300 mark.