what3words Releases New Voice API to Assist with Navigation

what3words Releases New Voice API to Assist with Navigation

The speech recognition specialist what3words has released a new end-to-end Voice API that was designed to help people figure out where they’re going. To that end, the API will allow users to say any three words, and what3words will provide them the address and GPS coordinates that they’re trying to get to.

The Voice API is built with machine learning and speech recognition technology from Speechmatics, and can be integrated into virtually any service or application. According to what3words, an address request is completed with only a single API call, making the system much simpler to deploy than alternatives that need to blend multiple APIs. The company claims that it takes only a few hours to get the platform up and running.

The Voice API will support eight different languages at launch, including Arabic, English, German, Hindi, Japanese, Korean, Mandarin and Spanish. However, developers will be able to introduce new languages thanks to Speechmatics’ machine learning capabilities. what3words also noted that its solution is able to account for homophones and inaccurate pronunciation.

“Speechmatics provides accurate speech recognition regardless of accent and can learn new languages at an industry-leading pace,” said Speechmatics CEO John Milliken. “Using automatic speech recognition to unlock the value in voice data is becoming more widespread and we are excited about the potential for our technology to create a global standard for addresses.”

The two companies are hoping to capture a portion of the lucrative speech and voice recognition market, which is expected to climb to nearly $32 billion by 2025. Their simple GPS solution could prove to be particularly appealing to car manufacturers, who could leverage the technology to improve the navigational capabilities of connected cars.

Of course, what3words and Speechmatics will face strong competition from companies like Facebook, which recently released its own open-source speech recognition platform.

January 24, 2020 – by Eric Weiss