A Chat with Katy Wigdahl, CEO at Speech-to-Text Software AI: Speechmatics

Speechmatics is the world’s leading expert in deep learning and speech recognition, which delivers the most accurate speech recognition through the first Autonomous Speech Recognition (ASR) engine.

Our speech recognition technology is used across a range of verticals, including media & entertainment, contact centres, CRM, financial services, security, and software. Every month, we process millions of hours of transcription worldwide  in 33 languages.

Having pioneered machine learning voice engineering, We are enabling companies to build applications that detect and transcribe voice in any context and in real-time, using neural network which encompass acoustics, languages, dialects, multiple speakers, punctuation, capitalisation, context and implicit meanings.

Where did the idea for Speechmatics come from?

Speechmatics has deep academic roots – our founder Tony Robinson was one of the first to discover the practical capabilities of deep neural networks and how they can be used to benefit speech recognition, at Cambridge University in the 1980s. This commitment to academic rigour remains core to what we do today: we are global experts in deep learning and speech recognition, tackling inclusion and bias in AI.

We are unlocking the potential of speech technology, extracting business value with human-centred artificial intelligence. Our long term goal remains true to the foundation on which Speechmatics was built: providing enterprise-grade speech recognition technology that understands every voice in any context, any time, anywhere.

The use of speech recognition technology is so wide and varied and fundamental to the provision of services in major industries, including media & entertainment through captioning and subtitles, contact centres and customer relations call transcriptions, financial services through analytics and biometrics, and accessibility and e-learning in the edtech space.

How has Speechmatics evolved during the pandemic?

As a business we’ve grown 250% since Series A investment in October 2019 – right through the Covid-19 crisis. The pandemic was a period of major innovation at Speechmatics. After a year of research and development, we launched our Autonomous Speech Recognition (ASR) technology in October 2021.

Our vision is to spearhead the evolution of human-to-machine interfaces, increase inclusivity in AI technology and lower AI bias. With the rollout of ASR, we broke through a significant number of boundaries standing in the way of making this vision a reality.

This autonomous learning method – generally using artificial neural networks – does not necessarily require sample data to be labelled in advance by humans. In real terms, this means engines can be trained on data straight from the gigantic pool of audio files taken directly from the internet, thus delivering a far more comprehensive representation of all voices and dramatically reducing AI bias and errors in speech recognition.

We conducted a study in partnership with Stanford in 2021 which showed our technology out-performing the likes of Google, Amazon, Microsoft and Apple in this area, which was a fantastic achievement for a UK-based AI company with just over 100 people competing directly with Big Tech. Seeing the CEO of Softbank commenting on Twitter that “Speechmatics is setting the standard” and being a catalyst for “other major tech companies to do better at reducing racial bias” validated so much of what our team is striving to achieve.

What can we hope to see from Speechmatics in the future?

The reason we have been able to drive these major advancements is entirely due to the calibre of the Speechmatics team and creating a workplace for innovators. We have grown significantly in the past two years, and will be continuing to expand our team throughout 2022. We hope to continue to be the home for the ambitious next generation of AI talent looking to push the boundaries of this technology in the years to come.

We are aiming to make speech recognition capable of understanding every voice, in any context, any time. We’ve got some exciting projects ahead of us – including a partnership with the prestigious Mila Labs in Canada – tackling bias and making speech recognition truly effective for everyone. With our ASR engine, the possibilities are endless and I cannot wait to see what the future holds.