On June 28, 2022, UK-based speech-recognition company, Speechmatics, announced it had raised USD 62m in series B funding. The round was led by Susquehanna Growth Equity (SGE), with participation from existing investors AlbionVC and IQ Capital.
SGE, a US-based private equity firm, is part of Susquehanna International Group, a family of funds that includes Susquehanna Private Capital, which acquired a majority stake in language service provider, thebigword, in August 2021.
While the series B valuation remains undisclosed, Speechmatics CEO, Katy Wigdahl, told The Times that the company had been “inundated with interest from investors,” with some of the highest valuations being 2.5x of the lower ones.
When deciding on investors, Wigdahl took a pragmatic approach. “I thought no, I’m not going to accept these frothy valuations,” the CEO explained. “It’s not about the biggest valuation you can get. It’s actually about getting that right investor and the right type of terms.”
She added, “We believe the money we’re bringing in is what we need for the next three years. We are creating a plan that gets us to GBP 100m annual recurring revenues.”
On the rationale for the investment, Robert Whitby-Smith, Partner at AlbionVC, cited Speechmatics’ strong growth since AlbionVC made its series A investment in 2019. He said, “Our view is voice will become the increasingly dominant human-machine interface.”
Speechmatics also has an accuracy edge compared to its competitors, according to Jonathan Klahr, Managing Director of SGE. “We started tracking Speechmatics when our portfolio companies told us that again and again Speechmatics win on accuracy against all the other options including those coming from big tech players,” he said.
According to the Speechmatics announcement, the investment will be used for business development; the company plans to further expand across the US and Asia Pacific, adding to its existing presence in the UK, US, India, and Czech Republic.
The company will also focus on improving speech-recognition accuracy and expanding coverage beyond the 34 languages currently supported by its engine.
Use Cases From Medical to Games
Speechmatics’ cloud and on-premise offerings provide access to a speech-to-text engine that draws on deep learning and speech-recognition technology.
Use cases span communications monitoring, automotive command, e-discovery, along with transcription for various settings, such as media, medical, legal, chat applications, and games.
The company’s approach to speech recognition departs from traditional techniques, according to CEO Wigdahl. With a goal of supporting diverse voices, accents, and dialects, Speechmatics creates single, multi-use audio databases for each target language.
A TechCrunch article quoted Wigdahl as saying, “The way we look at language is global. Google will have a different pack for every version of English but our one pack will understand every one.”
In 2021, Speechmatics switched from a supervised learning approach — working with labeled audio data from hundreds of speakers — to a “self-supervised” machine learning technique, incorporating millions of hours of unlabeled data from hundreds of thousands of individual voices.
This shift has, according to Speechmatics’ quality evaluations, reduced AI bias and speech recognition errors and saw the company outperform competing offerings from Microsoft and Google.
Language Industry Partners
In addition to servicing clients from a range of industries, Speechmatics has formed partnerships with a number of language industry players.
Transcription and captioning provider, Ai-Media, for example, integrates Speechmatics’ speech-recognition engine into its captioning workflow. Speechmatics produces speech-to-text while AI-Media applies its technology to split text into captions and synchronize output with source media.