Six months and 2,000 lessons later: what MuseCool has learned about music education
Tech.eu Cate Lawrence
MuseCool has run about 2,000 music lessons through its AI since April. It’s trying to understand the playing, not just the chat, and that could change music teaching.
Based on reporting by Tech.eu, Cate Lawrence — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Music lessons are usually built on habit. Teachers often teach the way they were taught, and that can mean a lot of tradition, some of it useful and some of it stale. MuseCool, a UK startup, is trying to break that loop by putting audio AI into the middle of the lesson itself.
Since going public in April, the company says it has processed around 2,000 lessons. That matters because MuseCool is not just transcribing speech. It is trying to understand what’s being played, where the student is in a piece, and what’s happening around the music. CEO and co-founder Petru Cotarcea compares it to a kind of Shazam 2.0, but for lessons, and says that understanding the music itself is the core difference.
The adoption story has been messy, which is usually the honest part of any education startup. Of roughly 100 tutors approached, about 40 didn’t even try the product. Of the 60 who did, 28 became users, and they brought their studios with them. Cotarcea says that once teachers get used to a system, they tend to stick with it. That stickiness is useful, but getting there is the hard bit.
MuseCool is also building the rest of the lesson around that audio layer. It has around 4,000 piano pieces in the app, with the catalogue growing quickly after a summer hiring push, mostly in Ukraine, to expand public-domain music into XML. By the end of the year, it hopes to add licensed music too. The platform also now offers more detailed summaries and heat maps that let users click individual bars to see what was discussed and what was happening musically.
The bigger play is data. MuseCool says its analysis suggests only around 40 per cent of an average lesson is spent actually playing. The rest goes to explanation, discussion, and general conversation, especially with children. That could help tutors, but it could also help institutions spot problems earlier, give researchers real evidence instead of anecdotes, and maybe even tell manufacturers how instruments are actually used. Cotarcea’s long-term vision is broad enough to include a talking piano, which sounds absurd until you remember how often education software starts by solving one practical problem and ends by quietly trying to rewrite the whole category.
My take — AI-written commentary, not fact-checked reporting
This is the kind of AI people should be funding more: dull, specific, and attached to a real workflow instead of a demo reel. The music education world is conservative for a reason, but that also leaves room for tools that do something concrete instead of waving at “transformation.” The industry keeps pretending data is only for the flashy sectors; MuseCool is betting it belongs in a piano lesson too.
Read more about this at: Tech.eu
