Tuning pianos using reinforcement learning

The tuning system of a piano has remained relatively unchanged since the instrument’s inception. A piano’s tuning system has been designed to be both inexpensive to manufacture and to preserve the tension and thus pitch of each string over long periods of time. This tuning system requires such a high degree of skill to manipulate that only trained professionals are able to tune pianos. This paper presents a novel adjustable impact tuning hammer and a reinforcement learning control system that may allow piano owners to tune their own pianos in the future.