Machine Learning Systems can
be made to predict F0, but…
•Opaque
results
•Require
large amounts of training data
•Internal
representation is typically very complex
•Models
of similar speakers aren’t internally similar
It’s hard to
take the learning out of a machine learning system and make it understandable
to a human.It’s also often hard to
port the system from one problem to another, even to a closely related
problem.