FAQ

Using learnings from our research, we have created an Android app designed to help those with non-standard speech communicate better. Find out more at g.co/ProjectRelate.

In order to improve the ecosystem of speech recognition tools globally, we’re also excited to provide developers with resources via our Project Euphonia GitHub page. Access our open-source repositories to develop personalized audio tools for research and train models for diverse speech patterns. We hope this can contribute to creating more localized speech applications for users with non-standard speech.

You can learn more about the insights we’ve gained from this research in our publications section.

Voice and message banking are often performed when someone is preparing to lose the ability to speak, either due to a progressive disease or pre-operatively—often before any speech changes have occurred. In contrast, Project Euphonia is an initiative that has been focused on improving speech recognition rather than recreating a person’s unique voice.

We do not currently offer a way for you to recreate your own voice. If you're interested in doing so, we'd recommend checking out other providers for voice banking, and/or storing your natural voice recordings through message banking tools.

Project Euphonia is no longer seeking additional speech samples. Since launching our initiative in 2019, we have had the opportunity to collaborate with various organizations and receive contributions from a wide array of people with non-standard speech. This data has provided a strong foundation for our work and we’re excited to be able to share some of these insights with developers through our open-source resources on GitHub. We hope these learnings can support the creation of more accessible speech recognition technology globally. 

Yes! If you have non-standard speech and are interested in contributing to data collection, please visit the Speech Accessibility Project or the Centre for Digital Language Inclusion. Google collaborates with these organizations to improve speech recognition services for everyone.

We launched Project Euphonia in 2019 with a mission to make speech recognition technology more accessible for people with non-standard speech. As part of the evolution of this research, we are excited to embark on the next stage of Project Euphonia by open-sourcing some of our audio and ASR tools. These resources, available on GitHub, come as a result of our last several years of work with community organizations and contributions from people with non-standard speech. While we will no longer be collecting additional speech samples from participants, we hope these learnings will empower the ecosystem to engage in their own data collection and speech recognition efforts, ultimately making technology more accessible for people around the world.

For any press inquiries, please reach out to press@google.com.

Yes! We have created an Android app designed to help those with non-standard speech communicate better. Find out more at g.co/ProjectRelate.

You can access and contribute to Project Euphonia via GitHub today. Project Euphonia is a research initiative that aims to provide developers in the ecosystem with insights on how to build and improve their own ASR models and data collection technologies for non-standard speech. We hope these resources will allow for more accessible speech recognition to become available in additional languages and cultural contexts.

Contact us at project-euphonia@google.com.