So, what’s the magic behind this scientific oasis?

Led by Professor Tanel Alumäe, the lab currently consists of 11 members: the team’s focus is on speech-centric natural language processing including everything from speech recognition and speaker identification to speaker anonymization. While Estonian is at the heart of their research, a lot of what they do is multilingual.

What sets them apart? They’re firm believers in open source, which means they’re not just building fancy models that never leave the lab — but also developing real-world speech applications and making their tools usable for the wider public.

So, who’s standing behind the scenes?

We’ve met two students working in the lab with very different stories — but one common passion!

Meet Artem — a third-year bachelor’s student at TalTech. Coming from a Physics- Mathematics lyceum, he was actively involved in various Olympiads, including (but not limited to!) Maths, Physics, Informatics, and English. With that kind of background, it’s no surprise he initially leaned toward Applied Mathematics and Mathematical Physics.

But plot twist — fast forward a few years, and here he is, nearly graduating from TalTech’s Cybersecurity program, working in one of the most unique university labs in Europe.

“It happened historically that apart from Ukraine, my family members were most related to Estonia,” Artem explains. When he applied in 2022, his Estonian wasn’t quite what it is today, which actually helped him narrow down his options. The Cybersecurity Engineering programme, one of the few offered in English, stood out.

Yet it wasn’t just about language. Artem made a conscious choice, knowing the field’s broad scope and rich connections to other CS subfields. As he put it, he wanted a solid foundation to explore many future directions after graduation.

But how does a bachelor’s student student join a research team? It started with curiosity about the newly launched ChatGPT-3 and wondering if he could explore it in academia. After discovering Professor Alumäe’s “Natural Language and Speech Processing” master course and hearing about lab openings, Artem didn’t waste a minute. One interview later, he was in. (He did admit the interview was pretty short and added: “Studying math well really helped.”)

Meanwhile, Joonas — another talented lab member, brings an equally impressive — and inspiring — story. After earning bronze medals in the International Mathematical and Physics Olympiads during high school in Estonia, he set his sights on studying abroad — particularly in the UK or US. His academic achievements helped him secure admission to Cambridge University, where he completed both a Bachelor’s and Master’s degree in Mathematics.

However, after four years of theoretical study at one of the world’s most prestigious universities, the intense focus on pure math eventually pushed him toward more applied areas. During COVID, nostalgia for Estonia’s endless forests brought him back home, where he began his PhD. Considering new directions in applied machine learning, TalTech presented a great opportunity.

The Language Technology Lab wasn’t a random choice for Joonas. Back in 2010, he discovered a public Estonian speech transcription service — tekstiks.ee. Curious, he looked up who was behind it and found Professor Tanel Alumäe. With strong recommendations and great enthusiasm, Joonas joined the team — and is now a PhD student and Junior Researcher in the lab.

So what do they actually do?

Artem holds the official title of Research Software Developer, focusing on modeling and training methods for speech recognition. His recent thesis project, “Linguistic Content-based Speech Anonymization,” aims to anonymize spoken content without modifying speaker- specific attributes. As Artem explains, it also ties into cybersecurity — voice carries a lot of personal information: accent, pitch, language, even age or ethnicity. All this makes voice data a potential privacy risk — and that’s exactly where cybersecurity meets language tech.

Joonas works on methods for extracting speaker-related content from audio. More specifically, he works on speaker segmentation, speaker diarization (answering “who spoke when?” in recorded conversations), and separating overlapping voices, which is especially useful for TV shows or meeting recordings. The work has been applied to both high-resource languages, such as English and Mandarin, and low-resource languages like Estonian, where training data is scarce. For example, it forms part of the lab’s closed captioning system for the Estonian parliament.

The Lab culture: Working solo but never alone

Collaboration isn’t just a buzzword here — it’s how things get done. As Artem puts it about working with Joonas: “Having him next to me pretty much every day gives me a good perspective on how to become a better researcher in the future, what I need to understand, and what I should do.”

Joonas also noted that the lab is always open to supervising bachelor students. Typically, they start with technical tasks, but if you’re proactive and bring in ideas — especially ones related to existing projects — your contribution will be genuinely appreciated.

And Artem confirms it: even as a bachelor student, he feels his work truly matters.

“You usually work alone on your project alongside your supervisor. So the success of the project is very much dependent on you.”

The lab structure is pretty flexible — everyone works independently, but there’s also a supportive community. Panels meet regularly to discuss progress, exchange ideas, and keep things moving. It’s a space not just for science, but for friendships too.

One of the coolest perks? The lab opens doors to travel while you’re developing your skills and sharing your research at international conferences. From recent experiences shared with us, Joonas has collaborated with labs in Toulouse and, in September last year, presented his findings at conferences like Interspeech in Greece.

For Artem, the lab has also played an active role in his professional development. Beyond his daily work, he has attended two summer schools in fascinating locations — one in Antibes (France) and another in Athens — both focused on generative modeling.

And here’s the thing — when you successfully deliver a scientific paper, you’re encouraged to present your findings at various conferences. This turns out to be incredibly self- rewarding. As Artem shared, his presentation at the NoDaLiDa-25 conference became one of the moments he was most proud of during his time at the lab.

Ready to join? Here’s what you need to know

Let’s be real — research can sound intimidating. But as Joonas puts it, “It’s easier to get involved than you might think.” And he’s right! Artem adds that there’s no room for hesitation: there’s always something students can contribute to.

Joining a lab isn’t just about professional growth — it’s a chance to level up personally, too. As Artem discovered, lab work often involves sharing your findings through talks at conferences or seminars. That’s how he uncovered his unexpected passion for explaining things to people — live and in person.

What’s really exciting is that the work you’ll do in a lab will seriously improve your problem- solving skills. You’ll spend hours diving deep into challenges that haven’t been solved before — that’s what research is all about. It’s not always easy to measure, but it’s how real scientific contributions are made.

From a practical perspective, the main tools you’ll use are Python and PyTorch. But more important than coding is your foundations in mathematical analysis, probabilistic modeling, and statistics — which are extremely helpful in this field.

The Laboratory of Language Technology isn’t just about code and algorithms. They’re tackling something bigger: making sure that in our AI-dominated future, every language — no matter how “small” — has a digital presence. Whether you’re a cybersecurity student fascinated by privacy challenges or a mathematician looking for practical applications, this place has room for diverse minds.

Massive thanks to our incredibly talented guests Artem and Joonas for sharing their journeys with us. Hopefully, it’s now clear that joining the lab isn’t just about growing in language technology — it’s about connecting with inspiring people who are genuinely passionate about what they do.

Interested? Admission is still ongoing at TalTech for the candidates from EU/EEA and Switzerland. If you’re ready to work on technology that actually matters — where your research could help preserve linguistic diversity or protect digital privacy — this is your chance. More info on admissions: taltech.ee/en/admissions