Google advances speech recognition technology for people with Down syndrome

The Canadian Down Syndrome Society (CDSS) has partnered with Google’s Project Euphonia to train Google’s technology to make it more accessible for people with Down syndrome.

Under the name Project Understood, they are collecting voice samples from adults with Down syndrome to create a database that will be used to train the algorithm and improve speech recognition for unique speech patterns. The goal is to collect 500 voices with more than 400 already donated.

Voice technology has been integrated in a number of devices, but not everyone can properly use a smart assistant. People with speech impairments are having a hard time using them, because Automatic Speech Recognition (ASR) systems are trained on typical speech.

Project Euphonia had already been launched by Google to improve understanding of non-standard speech, but it had a hard time collecting a variety of voice samples to train the speech recognition models. Google assures participants that only Google and affiliates working on Project Euphonia will have access to their voice samples, unless user consent is obtained for sharing with other parties.

Article Topics

accuracy  |  biometrics  |  Google  |  speech recognition  |  training  |  voice biometrics  |  voiceprints