As we say goodbye to another year, we're taking a moment to reflect on our accomplishments, month-by-month. There's been more than a few. In fact, despite some challenging aspects, 2022 was an incredible year for productivity at Speechmatics, with huge improvements in accuracy, incredible new features, the introduction of a host of new languages... and plenty more besides.
Join us as we reflect on the highlights of a year delivering best-in-class speech-to-text.
January
Traditionally a quieter month, we kicked off the year with a bit of noise, being named in the top spot for AI Magazine's Top 10 AI/ML companies. Described as a company that "outperforms Amazon, Apple, Google and Microsoft and delivers similar improvements in accuracy across accents, dialects, age, and other sociodemographic characteristics.”
February
February saw the release of our annual Voice Report. In it we took an in-depth look at AI Bias, the importance of accuracy and where the industry is heading next. If you want to contribute to this year's Report, there's still time for you to do so. Simply follow this link and have your say.
March
Events finally began to return to normal in March – after a long COVID-related disruption – and we attended a triple header of the MWC in Barcelona, the Ai4 Finance Summit and Enterprise Connect in Orlando, Florida. Meanwhile, our CEO was named one of Europe’s Most Influential Women, the first of many accolades in 2022.
April
In April, we introduced our Entity Formatting feature, with the ability to consistently and accurately interpret how entities such as numbers, currencies, percentages, addresses, dates and times should appear in written form. A huge help for the world of finance and beyond, this addition makes transcripts more readable and consistent across 14 languages and counting.
May
Another month, another award, this time for Boldest AI at the Bold Awards in Venice. Open to any AI standalone project or as part of a larger project, Speechmatics was judged on “how the development of AI-led to improved performance and its ability to solve real problems.” May also saw us take a giant step towards environmental sustainability when we teamed up with Kao Data.
June
There was huge news in June, as Speechmatics raised $62m in Series B funding. Led by Susquehanna Growth Equity (with participation from existing investors AlbionVC and IQ Capital), the funding was a statement of belief in all the hard work we've accomplished and great encouragement for our goal to understand every voice with human-level accuracy.
July
A super-busy July saw us launch our self-service portal and our Financial Language Pack, able to identify finance usage in conversation, helping to avoid confusion with abbreviations, acronyms (such as FX, SEC or GAAP) and finance-specific terms. July saw some great uplifts in our language accuracy too.
August
Despite the crazy heat of the summer, we released another game-changing feature in the form of Language Identification. By allowing customers to identify the predominant language spoken in any media file automatically, we're able to reduce time and costs for a number of different use cases.
September
An astonishing 14 new languages were added to our offering in September, meaning we could give a voice to an extra 340 million people. Our technology now supports over half of the world's population, enabling more people to access speech-to-text that outperforms all our competitors.
October
We held our first LinkedIn Live in October, interviewing Mariano Tan, the founder of Prosodica. You can watch the video here, which involves a wide-ranging discussion about Speechmatics’ breakthrough in self-supervised learning, what it means for the Contact Center industry, and how AI-led speech-to-text helps Prosodica optimize its business.
November
In November, we introduced the world to our Real-Time SaaS API. This new offering allows anyone to use our speech-to-text, whatever size their business, wherever they are in the world. If you'd like to test it out yourself for free, here's a link to our portal.
December
At the start of December, we announced our partnership with NVIDIA. Their market-leading AI systems and software stack will train larger models and take our technology ecosystem to the next level. Elsewhere, we've already started work on some incredible projects for 2023, so, watch this space.
Awards
FT 1000: Fastest Growing Companies in Europe
Speech Technology Industry Award Winner
Bold Awards – Boldest AI
Computing’s AI and ML Award
UK Business Tech Awards – FinTech Innovator of the Year
Growing Business Award – Technology Business of the Year
What We’ve Been Writing About
As experts in deep learning and AI, we’re always keen to share our thoughts on all aspects of speech technology. Here are a few of our favourite topics of conversation from the past twelve months.
Continuous Content and the Need for Innovation
Speechmatics’ Data Science Engineer, Benedetta Cevoli, looks at why YouTube’s automated captioning service is notoriously unreliable and asks what can be done to improve things within the speech-to-text industry.
The Problem With WER
In part one of a two-part piece, Accuracy Team Lead, John Hughes, explains how and why word error rate (WER) as an accuracy measure is outdated and often misaligned with human judgment. Part two looks at where we go next.
How to Successfully Achieve Multinode Training in PyTorch
Join Speechmatics’ Machine Learning Engineer, Ellena Reid, as she looks at the best way to turn a single node training setup into a robust, platform-agnostic, multinodal one.
How Speech-to-Text Can Be Improved
George Lodge, Computational Linguist at Speechmatics, looks at some of the problems and issues facing speech-to-text today and discusses where we can make necessary improvements.
2022 has been a great year in many ways, and we can’t wait to get our teeth stuck into 2023 and what it has to offer. We’ll be kicking off with our annual Voice Report, which we’d love your input on. Have your say about the future of speech technology here.
The Speechmatics Team