Speech recognition company Speechmatics masters a language a day - and has now clocked up 72
It offers free access to its language packs online
Speechmatics’ machine learning platform has been mastering languages at a rate of one a day and its Project Omniglot total now stands at 72.
The Cambridge company put Automatic Linguist (AL), its AI framework, to the test during phase 1 of the project to learn as many languages as possible. Having successfully built 46 new speech-to-text languages in just six weeks – exceeding the team’s expectations – the company has now taken its total number to an industry-leading 72.
Traditionally, building a new language pack takes months and is a laborious affair, meaning only the most-widely spoken languages in the world remain the focus. Due to the speed at which the languages were produced automatically, Speechmatics is offering people the chance to use them for free initially.
The initial phase of Project Omniglot has proven that the machine learning framework behind AL works extremely well. It can automatically learn the sounds (phonemes) of a language as well as the grammar and semantics in order to determine which sentences make sense.
Speech-to-text technology is much-discussed right now and, as the world is becoming increasingly more connected, broad language coverage is also becoming essential.
Benedikt von Thüngen, CEO of Speechmatics, said: “We are already seeing a shift to a speech-enabled future where voice is the primary form of communication. Transcription not only eases the lives of many people, but opens the door for new opportunities, especially in regions with lower literacy rates. As a company, we have focused on accuracy and deployability, whereas through AL and Project Omniglot we focused on increasing our language coverage. As a team we are very humbled and impressed by the results we have achieved and are excited by the potential opportunities we will now enable. We would love for people to try the new languages, give us feedback and see how we can develop them from there.”
Tom Ash, speech recognition director at Speechmatics and recent winner of the ‘Speech Luminary’ award, added: “The framework we’ve created works completely autonomously and so efficiently, that we can experiment with languages that would otherwise be uneconomic to build. Also, it now gives us the ability to iterate rapidly on any given language and improve them at an unprecedented pace. We have built ASR for languages spoken by communities in the Philippines, India or central Asia, all of which are often overlooked. There are over 7,000 languages in the world, and our ultimate goal is to make speech recognition technology available to as many as possible.”
The growth of the sector was underlined by the recent announcement of Amazon Transcribe, an automatic speech recognition (ASR) service designed to enable developers to add speech to text capabilities to their applications.
Benedikt said: “The launch of Amazon Transcribe is brilliant news for the future of speech and another step towards voice-enabled future. However, Amazon is yet another tech giant focusing on the most widely spoken languages - English and Spanish in their case.
“Transcription can not only ease the lives of many people, but open the door for new opportunities, especially in regions with lower literacy rates. At Speechmatics, we have just completed an innovation challenge called Project Omniglot. As a result, our homegrown AI learned 46 languages in under six weeks and we are now giving people the chance to upload and transcribe unlimited hours of audio for free.
“Amazon are following a well-trodden business path by focusing on developers and the same languages other tech giants target. There are over 7,000 languages in the world, and our ultimate goal is to make speech recognition technology available to as many as possible. We are looking forward to Amazon expanding their language coverage and joining us in the voice tech revolution.”
Speechmatics was joint winner of the AI scale-up category in the Cambridge Independent’s Entrepreneurial Science and Technology Awards.
Try a language for free at https://app.speechmatics.com/register.