Case Study: CloudBounce achieves automated professional music and podcast mastering with Valohai

Mastering music tracks and podcasts with machine learning

CloudBounce is a SaaS platform that uses machine learning to automate mastering for music tracks and podcasts, letting users get a finished master within minutes. With a small data science team facing a vast number of experiments (model architectures, training parameters and preprocessing choices) and the need for reproducible results, CloudBounce turned to Valohai to avoid spending their limited time building and maintaining training and inference infrastructure.

Valohai provided a flexible deep learning platform that automatically stores reproducible experiments (code, parameters and data selection) and offers pipeline functionality to close the loop on user feedback. Using Valohai, CloudBounce’s team focused on improving genre-recognition and mastering models more quickly, integrated customer edit signals to trigger retraining, and shortened iteration cycles—resulting in steadily improving model accuracy and a production mastering service that users often accept without further adjustments.

CloudBounce

Juho Pennanen

ML Lead