How do you use AbstractFifo inside a synth voice where the number of samples is constantly changing?

How do you use AbstractFifo inside a synth voice where the number of samples is constantly changing?