PINDROP BLOG

Mozilla Wants to Build an Open Voice Engine

With manufacturers weaving voice control into a wide variety of devices, from phones to cars to home assistants, the accuracy of the voice-recognition engines in those products is becoming vital. Mozilla is hoping to help improve that accuracy with the release of its new Common Voice project.

The new initiative is a response to the proprietary way that voice engines are built right now. For the most part, voice technology is owned and operated by a handful of vendors, including Amazon, Apple, and Google. Those systems are mostly black boxes, inscrutable to outside users or developers. Mozilla is aiming to bring some transparency to the proceedings by creating a voice engine that is trained by the masses.

Project Common Voice allows anyone to participate in building the engine through several different methods. First, visitors can listen to a sentence spoken by someone else and compare it to written text to help validate it. Second, they can record their own sentences for others to validate. All of this helps the voice engine recognize specific words and phrases.

Mozilla officials said the project is meant as an answer to the problem of developers not having access to the voice technology being built by the big providers.

“We think that stifles innovation so we’re launching Project Common Voice, a project to help make voice recognition open to everyone. Now you can donate your voice to help us build an open-source voice recognition engine that anyone can use to make innovative apps for devices and the web,” Mozilla said on the project’s site.

The company is hoping to capture as many as 10,000 hours of speech as part of the project, to help develop a public database that can be used by any developer. The speech databases used by Amazon, Google and the other vendors with speech engines are proprietary and so outside developers have no access to them or any idea how they’re built.

“As voice technologies proliferate beyond niche applications, we believe they must serve all users equally well. We see a need to include more languages, accents and demographics when building and testing voice technologies. Mozilla wants to see a healthy, vibrant internet. That means giving new creators access to voice data so they can build new, extraordinary projects. Common Voice will be a public resource that will help Mozilla teams and developers around the world,” the company said in the FAQ for its project.