Loora, a generative AI app that uses an audio interface to help users learn English, raises $9.25 million

Image credits: Loora.ai licensed.

The jury is still out on whether the adoption of AI will truly spell the end of humanity as we know it, but in the meantime, we continue to see a boom in AI startups and AI deals. In the latest development, a generative AI-powered language learning startup called Loora is sneaking out with $9.25 million in seed funding to improve people’s English conversational skills, regardless of topic , via on-the-fly voice-based interactions with the iOS Loora assistant of the same name.

Unlike most AI applications right now which are text based, write prompts for ChatGPT, Midjourney and the rest Loora is completely audio focused. Which is to say, you talk to him to learn, like you might a human guardian, and Loora talks back to you. The startup says its AI tutor can understand and help improve anyone’s bad English and bad English accent on any topic under the sun.

Funding comes primarily from three early-stage investment specialists: Emerge leads the round, with participation from Two Lanterns Venture Partners and Kaedan Capital (as well as various angels).

Tel Aviv-based Loora, which started in 2020, may be sneaking out today, but Roy Mor, the CEO who co-founded the company with Yonti Levin, noted that it has been quietly operating for some time and has already signed up thousands of subscribers. of paying consumer and prosumer customers who want or (for business) need to improve their English language skills and use the Looras app to do so.

And in turn, Loora uses this interactivity to keep expanding her knowledge base.

Mor, a Mobileye machine learning ally, said he and Levin (Looras CTO) decided to focus on language classes, and especially English, for a couple of reasons.

The first is that English is by far the most common language being learned right now, with over 1 billion students worldwide. (And the plan is to stay in English for now.)

The second is that while there are already dozens (if not more) popular online language learning platforms in the world, they have found a vacuum when it came to mastering conversational English in a one-on-one learning environment.

At one end are language learning apps. Platforms like Duolingo, Memrise and Babel, Mor said, do a great job but are aimed primarily at casual audiences. A typical user could be as easily on TikTok as on one of these apps and could give you basic skills, but not much more. Being able to order a coffee is great, Mor said, but if you want to work in English, it’s not relevant enough. And they’re limited on what they can do with the voice-based interactions they have.

On the other side are human tutors, either in person or online via video meetings, who can delve into specific topics you may want or need to cover, listen to you, and shape their teaching to suit your particular needs. But for many, those real tutors are prohibitively expensive, she explained.

The solution, inevitably, came through advances in technology. Leaning on the new wave of generative AI popularized by services like OpenAI’s ChatGPT and Midjourney’s illustrator that creates images based on verbal prompts, Loora has created an English-language tutor that responds to conversation prompts, provides conversation suggestions when you don’t have one in mind, and they understand and respond to what you say to them to give you guidance on how to improve your spoken English.

Mor said that initially, in 2020, Loora built their own large language models. But now that these have become widely available you can leverage a variety of them for your service, some are better suited to different tasks so different LLMs are used eg for openers (when someone starts using the app), for first conversations, subsequent conversations, different topics and a different one for each user’s app memory. Loora uses these third-party LLMs, he said, but we train on our data.

To be clear, Loora isn’t ingesting audio per se which is still a very nascent area, with companies like Meta launching Voicebox, its first effort in that space, just earlier this month; and others in academia are just starting to look into how and where voice-based inputs and interfaces might work. But he’s using a mix where audio is an essential component, as the goal is to improve spoken English.

We use a text-to-speech model as our conversational and grammar feedback models rely on text input, Mor said, but along with that, we also use speech itself to assess and provide feedback on pronunciation, prosody and fluency, as well as understand the feeling.

This also means that there is an inherent limitation to the app as it is: Loora focuses on speaking, not reading and writing, and it seems you need to have at least some knowledge of English rather than all to get started. There will be some reading introduced, Mor said, with an upcoming module providing texts to read as a mainstay of a later discussion. But writing isn’t really on the table right now.

This means that while Loora is certainly setting out to do something different from what other language learning apps on the market are now offering, if you want to master one language, as with others, you will probably still need to use it in tandem with others. learning experiences.

Still, the arrival of Looras seems like a major advance in what’s available on the market today, and in the long run, it’s likely to be just one of many efforts to advance audio interfaces for generative AI services.

We’ve been partnering with cutting-edge AI companies since 2016, so Loora was a natural fit for our portfolio. In 2020, co-founders Roy Mor and Yonti Levin already recognized the potential of generative AI for language learning and embarked on a mission to harness it for the benefit of society, long before ChatGPT was introduced, said Liad Rubin, general partner of Emergere, in a statement. General-purpose AI models are good at performing general-purpose tasks, but there is a huge benefit to building and training domain-specific AI. This is why Loora envisioned and built her own AI tutor with the sole purpose of teaching English, creating an unrivaled next-generation solution for virtual English teaching. We are thrilled to be a part of this journey.