From guesswork to prototype — Bravori explores AI powered music learning with FAIR
Helsinki-based EdTech start-up Bravori is building a gamified practice companion app for young music students. Faced with an ambitious technical question, could software automatically evaluate a student’s piano playing and deliver structured feedback? The team turned to FAIR to find out. The answer, it turned out, was a qualified yes.
Text by Solja Sulkunen & Martti Asikainen, 1.6.2026 |Photo by Bravori
Anyone who has learned a musical instrument knows the gap between lessons. A student practises alone, makes the same mistakes, and receives no feedback until the following week’s session. For teachers, that silence is frustrating: progress stalls, bad habits form, and there is no visibility into what happens in the practice room.
Many educators working in digital learning have asked whether technology could close this gap. Until recently, the honest answer was uncertain.
Bravori is a Helsinki-based EdTech start-up that emerged from the University of the Arts Helsinki Hub Pre-incubator programme. The company is developing a mobile application that encourages daily music practice through gamification — points, rewards, and interactive tasks — while giving teachers learning analytics to track student progress.
After growing through the Helsinki Education Hub incubator, Kiuas Accelerator, and the University College of London EdTech Accelerator, the team arrived at a question it could not answer from within its own expertise: was AI-powered performance evaluation technically feasible for the kind of beginner-level piano material its users practise every day?
A realistic question before an ambitious investment
As a company founded by musicians, Bravori understood the pedagogical need well. What the team lacked was clarity on the engineering side.
Before committing development resources to an AI feature, they needed to know whether automated note-error detection was achievable at a level of accuracy meaningful enough to be useful, what kind of audio data the app should be collecting, and how any AI component might integrate with an existing mobile application.
According to the Bravori team, the fear was not that AI could not help, it was that the company might invest significantly and discover only later that the technical constraints made the vision unworkable.
They sought support from Finnish AI Region’s (FAIR) Test-before-invest services, beginning with a pop-up event aimed at regional EdTech solutions, followed by AI advisory sessions, and ultimately proceeding to a structured Proof of Concept engagement.
Building and testing a functional prototype
FAIR carried out a Proof of Concept project focused specifically on piano performance evaluation, an area where several distinct AI technologies must operate together.
FAIR carried out a Proof of Concept project focused specifically on piano performance evaluation, an area where several distinct AI technologies must operate together.
The work involved building a functional prototype capable of analysing student recordings, implementing detection of note, rhythm, and tempo errors, and generating colour-coded scores alongside written feedback for teachers.
The process was not without complication. Consumer-grade microphones introduce what the team describes as “ghost notes” — audio artefacts that can mislead analysis — and expressive playing that deviates from strict metronomic timing presents particular challenges for automated systems.
These limitations were documented transparently as part of the PoC output. According to the FAIR team, mapping the boundaries of what the technology can currently do was as important as demonstrating what it can achieve.
Validated feasibility, and a clearer path forward
The Proof of Concept demonstrated that note accuracy evaluation is achievable for beginner-level repertoire. In simple pieces, the prototype reached approximately 95% accuracy in detecting wrong notes. Teachers can already experiment with a working prototype that accepts uploaded sheet music and recordings and returns automated analysis and visualisations.
The collaboration produced outcomes that went beyond the prototype itself. Bravori now has an actionable development roadmap identifying three parallel paths: daily teaching use with the existing prototype, integration of the AI feature into the main app with a developer partner, and collaboration with European music informatics research groups.
FAIR also connected the company with leading research groups at Queen Mary University of London and JKU Linz, and identified how the wider European EDIH network could support future local pilots and market entry.
Ecosystem anchors and the road ahead
Bravori’s development has been shaped by a layered network of support from Business Helsinki, Helsinki Education Hub, and now FAIR. This positioning within overlapping innovation ecosystems has allowed a small team to access capabilities well beyond what a start-up of its size could typically sustain internally.
The company’s ambitions extend beyond piano. Bravori is already planning how AI-powered feedback might expand to other instruments and learning contexts, with the longer-term goal of giving both students and teachers meaningful insight into what happens between lessons — wherever those lessons take place.
“Working with FAIR EDIH helped us understand what AI can actually do in music learning right now,” said a member of the Bravori team. “Instead of guessing, we now have real results, a working prototype, and a clearer picture of what’s possible. The collaboration gave us better confidence and helped us move steadily from ideas to concrete development.”
