Fact · Accessibility and health
1 millionutterances
By August 2021, Google's Project Euphonia corpus of disordered speech held over 1 million utterances, more than 1,400 hours, from 1,330 speakers.
Checked against its source on October 16, 2026
In the source's own words
“This effort has grown Euphonia's corpus to over 1 million utterances, comprising over 1400 hours from 1330 speakers (as of August 2021).”
Personalized ASR Models from a Large and Diverse Disordered Speech Dataset, Google Research Blog, 2021, Project Euphonia data collection.
Open the original- What it measures
- Speakers with speech impairments of varying severity, across conditions such as ALS, have recorded samples since 2019. The corpus was used to build personalized recognition models, later offered to users through Google's Project Relate app.
- What it does not show
- Volunteer recordings of mostly scripted phrases in English, as of 2021. Not a random sample of people with speech disorders.