This project is about an active controlled keyboard simulating any keyboard behaviour from harpsichord over a piano to an organ.
Usually piano players prefer to play real pianos to keyboards.In some situations, however, only digital pianos, stage pianos or master keyboards are available. But these devices are optimized for a common use, i.e. for different users at the same time which results in an unsatisfying performance for the respective situation, e.g. a piano player. Moreover, due to the mechanical construction, this behaviour cannot be changed afterwards and the keyboard always remains a compromise.
Alternatively, an active controlled key could simulate any behaviour. A possible structure is depicted here.
The principle can be described easily. A blindfolded player of an instrument (organ, harpsichrord or piano) would interfere with the keys being his means of output to the piano, apart from his ears and hearing capapilities as his input. For this player, hence, the keys generate forces according to a complex mechanism inside the instrument which makes up the feeling of the instrument, e.g. if one plays a Boesendorfer or a Steinway. Could this player be cheated by a machine providing the same input/output behaviour ?
Assuming a black box model of this keyboard, the key force can be viewed as a function of the key state and is certainly higly nonlinear. If it is possible to determine trajectories in the state space by means of experiment, a database can be built and approximated by some mathematical functions, e.g. radial basis functions. Given the state of a piano key being estimated or measured, a microcontroller can calculate the force to be produced by the actuator even compensating the dynamics of the key itself. The entire behaviour of the keyboard depends on the learned function and can be switched easily between different types of keyboard.