AI tool for music production trained on VSC

Vienna-based startup Beat Shaper has developed an AI tool that generates music from text prompts — producing not just complete tracks but also editable project files compatible with professional music production software.

While many generative AI tools in music focus on producing finished audio, Beat Shaper set out to build something fundamentally different: an AI that understands how music is constructed — not just how it sounds. The team adopted a symbolic music approach, training their model to interpret digital “sheet music” with detailed instructions for tempo, rhythm, and instrumentation — similar to the way a synthesizer uses dozens of adjustable parameters to shape sound.

Since no suitable dataset existed for this kind of modeling, the team had to build one from scratch — a process that demanded extensive testing and continuous refinement. One of their key challenges was ensuring the model could create original, musically coherent output without merely recycling its training data.

The role of VSC-5

Training even small AI models from scratch requires significant computational resources — especially when experimenting with hundreds of versions in parallel. That’s where VSC-5, Austria’s flagship supercomputer, came in.

With support from the competence center EuroCC Austria, Beat Shaper gained access to VSC-5 and trained more than 500 individual models. The cluster enabled them to cut training time by a factor of 50 compared to their local infrastructure — reducing model training from several days to just 1 – 2 hours. This capacity for parallel experimentation proved essential to the quality of the final model.

“The biggest advantage of using the cluster was being able to train so many models simultaneously. Without that, we never could have experimented at this scale — and that would have negatively impacted the final model’s quality,”
— Taylor Peer, co-founder of Beat Shaper.

Scheduled for launch in summer 2025, Beat Shaper’s tool will be available in free, premium, and professional versions. Unlike many competitors, the startup trained the model entirely on licensed or fair-use data — ensuring legal certainty for its users.

Support through public funding

With early-stage investment from Vienna’s high-tech incubator INiTS, funding from the Austrian Research Promotion Agency (FFG), and technical support from EuroCC Austria, the Beat Shaper team succeeded in bringing a cutting-edge AI music tool to life — despite facing competition from well-funded companies in the U.S.

"Given the lack of substantial early-stage venture capital in Europe, programs like these are crucial for helping European companies stay globally competitive,” says Taylor Peer. Beat Shaper’s story highlights the important role of public support programs in helping people turn bold ideas into working products. By providing access to funding, infrastructure, and expertise, these initiatives enable young companies to gain traction, build real solutions, and prepare to stand on their own in the global market.