Welcome to OpenSTT
a Mycroft Project
An Open Source Speech To Text Project
The OpenSTT project is aimed at creating an open source speech-to-text model that can be used by individuals and company to allow for high accuracy, low-latency conversion of speech into text.
Currently there are no open source speech-to-text models available, instead this technology is locked deep within large companies either tied to only their own proprietary products and services or behind expensive APIs that, in many cases, don’t respect user privacy.
OpenSTT is being led by members of the Mycroft A.I. team as they strive to create a powerful voice interface and artificial intelligence platform. Our goal is to make this technology available to as many people as possible. We do this by leading development of open source projects, OpenSTT is one of these initiatives.
The journey to an open source speech-to-text model has begun with the birth of the OpenSTT project. We have set up resources in order to engage our fantastic community to help us move forward with this initiative. If you are interested in contributing, we recommend checking out our Community website, the OpenSTT blog, and the Github Repo.
This quest starts with the evaluation of algorithms and frameworks to determine which approach is best suited to solve this difficult problem. Following the identification of strong possible ways to tackle creating the model, we will move on to conduct tests and analyze the outcomes. If it looks like one approach is the best, we will move forward to pursuing further development of that approach.
Finally, once we have done sufficient development we can continue to work on improving our approach and the model until it is in good enough shape for a 1.0 release. We hope that during this process developers, students, researchers, and other organizations will join us in creating an open source speech-to-text model for everyone.