Predicting phrase breaks with memory-based learning
G.J. Busser, Walter M. P. Daelemans, Antal van den Bosch · Research portal (Tilburg University) · 2001
We investigate whether Memory-Based Learning (MBL) can be used to predict Phrase Breaks (PBs) in speech production reliably. The MBL approach is compared to the HMM approach described in [Taylor and Black, 1998] using the same corpus and information sources. We show that a simple memory-based learning algorithm that uses only minimal context and information outperforms the HMM approach, in terms of precision and recall. An exhaustive search of variants of algorithms, metrics and information sources does not bring any significant further improvement.