Helsinki’s Speechly raises €2M seed for its ‘natural language understanding’ API

Speechly, a startup out of Helsinki that boasts an experienced team of speech recognition and “natural language understanding” experts, has raised €2 million in seed funding to make it easier for developers to add a voice UI to their products.

The round is led by Berlin’s Cherry Ventures, with participation from Seedcamp, Quantum Angels, Joyance Partners, Social Starts, Tiny.vc, Juha Paananen (co-founder of Nonstop Games, which exited to King) and Nicolas Dessaigne (founder of Algolia). The funding will be used by Speechly to further develop and open up its API to enable non-experts to create voice-enabled applications.

“Voice has shown real promise over the past few years but a real breakthrough beyond setting kitchen timers and playing Spotify is yet to be seen,” Speechly co-founder and CEO Otto Söderlund tells TechCrunch. “The current fundamental problem of voice assistant platforms is that they tend to fail with more complex user requests and needs.”

He says Speechly’s solution is to combine natural language understanding and speech recognition in a “novel way” that enables developers to a create a “highly reactive and seamlessly multimodal user experience” that better guides the user when expressing complex intents.

“You can think of the difference a little bit like trying to explain something tricky to your friend over the phone [which can be hard] versus face-to-face [which is often a lot easier],” says Söderlund.

To achieve this, Speechly has designed its own proprietary speech recognition technology “from the ground up” to support what it claims is a significantly wider range of voice-related user experiences compared to existing products.

As well as helping voice applications better understand complex intent, the other major problem that Speechly wants to solve is the business case for voice. The startup argues current voice assistant platforms, such as Amazon’s Alexa or Apple’s Siri, force businesses “to operate in someone else’s ecosystem” and share valuable user data.

Meanwhile, Speechly says current SDKs and APIs are either too complex or do not offer developers enough control over the end user experience.

“In addition to Google, Amazon, Microsoft and Apple, there are a few smaller companies and startups developing their own spoken language understanding (SLU) technology,” adds Söderlund. “These companies, owning their own proprietary SLU technology (as we do), we consider as our main competitors. However, these competitors mainly offer products that we consider a rather straightforward continuation from the classical turn-based dialog agent (think of Siri). We want to offer an alternative vision for voice UIs where highly responsive multimodal feedback ‘guides’ the user in real-time to resolve more demanding user tasks. This vision that we are building in our product we consider unique.”