Auris has been awarded a UK patent for our automatic speech recognition (ASR) engine, which is used in our DfE-endorsed reading app Fonetti. This is a huge and exciting step for Auris, as it protects our ground-breaking technology and allows us to control its commercial use.
Why is Auris's ASR engine unique?
Our ASR engine was developed specifically for the way children read aloud. When children read, they often skip words that they do not know, or go backwards to correct words they pronounced previously read wrongly. Keeping up with these 'jumps' is easy for humans, but very difficult for speech recognition technology.
Auris's ASR mimics the way the human brain processes children's reading and matches the words they have just said to its appropriate place in the text - whether that's before, after, or continuing on from the sentence they are currently reading. This creates a seamless reading experience for children so that they can stay focussed on the story.
How does the ASR engine compare to other speech recognition technology?
Anyone who's used any automatic captioning will know that it's not the most accurate - especially not with speech in a non-standard accent. Auris's ASR engine, which was designed using authentic data from children's read speech, is over 92% accurate for all UK accent groups, including the accents of people speaking English as an Additional Language (EAL).
To put that in perspective, Google's API is 43% accurate, and humans are about 95% accurate in understanding children's read speech. This makes the Auris ASR the only technology that understands children as they read!
"Our patents demonstrate the unique nature of our ASR engine in the voice technology space which is mostly addressing simple command and control processes rather than complex speech"
Chief Privacy Officer, Auris Tech Limited
Want to know more?
You can also download Fonetti on iOS or Google Play to see our ASR engine in action.
Or, follow us on social media to stay up to date with Auris news!