Speechmatics takes language accuracy to the next level
Today, Speechmatics is announcing its Next Generation languages update, marking a significant improvement in accuracy across many of its core languages for use in speech-to-text transcription, providing a reliable, scalable and cost-effective solution for partners and customers.
Machine learning innovation leads to Next Generation languages, with improved accuracy of up to 16%
The new development is based on several factors - refinements to existing technology, deployment of entirely new machine learning algorithms, and extending and enhancing Speechmatics' use of data resources to ensure close alignment with its customers’ needs.
Speechmatics’ internal testing showed an increase in accuracy of up to 16% for Global English and many of its core languages. The significant accuracy improvement for Global English means Speechmatics will now have one English language model supporting all major accents and dialects. Removing the need to use multiple languages packs for English dialects means customers will benefit from simplified deployments as well as a reduction in the overall footprint. In turn this reduces the overhead costs for customers regardless of application or use case.
The Next Generation languages update is applicable across all use cases, from subtitling news reports, to transcribing meetings, to flagging potential customer issues within a call centre. Speechmatics is committed to providing regular improvements to its core language offering.
David Pye, Head of Speech at Speechmatics, says: “We’ve analysed the market-leading quantity of data we have on this and, while we may be the first speech technology company to do away with English dialects completely, our expertise has proven that Global English is the right way to drive a shift change in the market and our Next Generation languages update supports this. By continuing to innovate and stay ahead of our competition with regards to accuracy, we will remain a leader in this field.”
The testing done with customers to date shows the team has made double-digit improvements in accuracy using Global English and the Next Generation language models, meaning many of the core languages offered by Speechmatics are better than competitors in the field.
Hewson Maxwell, Head of Technology, Access Services at Red Bee Media, commented: “At Red Bee Media, we systematically assess ASR solutions in order to know we are using the best ASR for the job. We’ve been really thrilled to have seen an average of 22% improvement in accuracy using the new Speechmatics’ Next Generation models across all our core languages, on top of the already excellent base. These improvements will allow us to produce more subtitles for less and we are fast-tracking the new models in to production as fast as we can.”
Pye continued: “Our innovation in machine learning means we can make big jumps in advancing speech recognition technology, including dialect-agnostic speech recognition. We’re doing away with specific dialect language models for English as our modelling is now so advanced we no longer need them.”
Speechmatics’ Next Generation language models are available now at www.speechmatics.com.
Speechmatics is a world-beating authority in Machine Learning and AI, applying the latest research to the problem of understanding speech.