The importance of languages for speech recognition

Roughly half the world speaks just one language – and very few people are fluent in more than two or three. Yet, Speechmatics’ any-context speech recognition engine can understand more than 30 different languages. From Spanish, Hungarian and Polish to Japanese, Russian and Korean, the list of our speech recognition languages spans the globe. To build and iterate languages faster than ever before, we developed an innovative machine learning framework – called The Automatic Linguist (AL).

A machine learning framework to rapidly build new languages

Most languages have inherent similarities in their fundamental sounds and grammatical structures. AL can recognize patterns within and across languages and apply these to a new language build – significantly reducing the time and data required to build new speech recognition languages. For example, AL enabled us to build Hindi in just a week.

AL won Speechmatics a Queen's Award for Enterprise in 2019 in the innovation category. But don't worry, Speechmatics’ technology isn't limited to the Queen's English – we operate in Global English, so you don't have to adjust the way you speak to be understood by our speech-to-text engine.

Whether you speak Australian English, American English, Jamaican English or African English, all you have to do is select our Global English language pack. We were the first company to do away with multiple language packs for different accents and dialects.

A global approach to understanding accents

In the UK alone there are about 56 main 'accent types'. The concept of having one language pack per accent or region is outdated in our increasingly connected and mobile world. We’ve all heard stories about people being misunderstood by their personal voice assistants – or closed captioning getting something awkwardly wrong.

Although very entertaining, these stories highlight a big issue. That's why Speechmatics’ Global English language pack encompasses all major English accents and dialects. Trained on thousands of hours of spoken data from more than 40 countries – and tens of billions of words drawn from global sources – our any-context speech recognition engine can cope with even the strongest accent. It also overcomes the industry-wide issue of handling multiple English accents in one recording.

But it's not just the variety of speech recognition languages and accents that's important for unlocking global value for our customers. Accuracy is also crucial – in all the languages we offer and in real-world situations such as noisy environments.

Accurately converting speech-to-text in multiple languages

Our world-leading machine learning algorithms can cope with anything from news subtitling or transcribing meeting notes to flagging up potential customer issues within a contact center. It's why our speech-to-text technology has been adopted by some of the largest blue-chip companies in the world.

We are already seeing a shift to a speech-enabled future where voice is the primary form of communication. The practical applications of our speech-to-text technology are now changing the way companies work – automating laborious tasks and unlocking the value of both live and recorded media.

So, what are you waiting for? Simply select Speechmatics and let the technology do the talking.