Analysts expect the speech recognition market to reach close to €23 billion by 2026, growing 16.8% on average between 2021 and 2026. One of the drivers behind this growth is the increasing demand for biometrics in the banking, financial services and insurance industry to help tackle fraudulent transactions. Another driver is voice-based searches, increasingly adopted by consumers in the detriment of traditional search. Also, by adding the option to search for text within audio and video files, content managers can better and faster understand, review, share and reference this type of content.
It is this use case that propelled the Romanian startup Vatis Tech to launch an infrastructure solution in the speech recognition market:
“We started with the idea to search in audio. I listen to more podcasts and audio content, and often I need to search inside their content, but I don’t have the necessary tools,” Adrian Ispas tells The Recursive.
In a nutshell, the company offers search and transcription infrastructure for audio and video content, offering users a fast way to accurately identify terms in this type of content. Just how accurately? The tests performed so far indicated a 94% accuracy of transcription, which is 10% higher than that of competitors. Part of the reason is that the company’s speech recognition and machine learning algorithms are specialized for various industries and verticals, such as media, finance, healthcare, and technology.
Vatis Tech to further develop the team and technology with new funds
Adrian Ispas further explains the company’s current business model: “We changed our initial plans of building a B2C platform for podcasters, journalists, and general users because we quickly realized that the necessary technology to implement our vision doesn’t exist. So, we’ve started to develop the technology, and step by step, we pivoted more to a B2B business and a provider of speech recognition infrastructure.”
The model includes a subscription option for users with a smaller data volume, starting at 3 euros per hour. The enterprise version offers dedicated support and infrastructure for corporate customers.
Developers can use the on-cloud and on-premise APIs to integrate Vatis technology into their products and services, benefiting from dedicated support and taking advantage of advantageous prices. Currently, Vatis Tech aims to have recognition algorithms for European languages such as French, Italian, German, Dutch, and Polish, after having started with Romanian.
Recently, Vatis Tech raised €200K from Early Game Ventures, Sparking Capital, and Malin Stefanescu, president of TechAngels. With the new funds, they plan to extend the development, sales, and data validators teams, as well as accelerate the development process.
“We already have a stable version for static transcription and a beta version for our real-time transcription API and on-premise version, and we’re planning to release a stable version of it in the next few months. Also, we’re planning to develop new algorithms for transcription post-processing to improve the current accuracy and develop new verticals for Romanian language,” Adrian Ispas adds.
Investors are confident about the startup’s winning formula. In the press release, Vlad Sarca, Partner at Sparking Capital, points out the solution’s relevance for business leaders who want to improve the customer experience or gain a competitive advantage through voice recognition integrations. Malin Stefanescu, President of TechAngels, believes in the company’s opportunity to tap into a growing market that still has room for innovation.