I’ve integrated a neural network for gesture path tracking. Now, the system can recognize and classify users’ conducting paths in real time, and save them to a database. This makes it possible to identify and match various conducting gestures to musical beat patterns—even when the user’s gestures aren’t identical each time. Since music often involves tempo changes, this feature will help train the system to recognize a wide variety of conducting shapes through repeated practice. The path data is updated and processed every 3 seconds, allowing the model to continually learn and adapt to the user’s real conducting style.
