British English male speaker ("RAB") for Festival.
This voice provides a British English male voice using a residual excited LPC diphone synthesis
method. It uses the CMU Lexicon for pronunciations. Prosodic phrasing is provided by a
statistically trained model using part of speech and local distribution of breaks. Intonation is
provided by a CART tree predicting ToBI accents and an F0 contour generated from a model trained
from natural speech. The duration model is also trained from data using a CART tree.
