Decision Tree-Based Context Dependent Sublexical Units for Continuous Speech Recognition of Basque

Abstract

Abstract. This paper presents a new methodology, based on the classical decision trees, to get a suitable set of context dependent sublexical units for Basque Continuous Speech Recognition (CSR). The original method proposed by Bahl [1] was applied as the benchmark. Then two new features were added: a data massaging to emphasise the data and a fast and efficient Growing and Pruning algorithm for DT construction. In addition, the use of the new context dependent units to build word models was addressed. The benchmark Bahl approach gave recognition rates clearly outperforming those of context independent phone-like units. Finally the new methodology improves over the benchmark DT approach.

RWTH LVCSR systems for quaero and EU-bridge: German, Polish, Spanish and Portuguese

“…The decomposed words are post-processed to produce a cleaner set of sublexical units and boundary markers are added to regenerate fullwords later after recognition. Very short units are avoided as they are usually difficult to recognize and also could harm the overall WER with more insertion errors [30,31]. To generate N -gram backoff sub-lexical LMs, different hybrid vocabularies are selected, where top-most 5k full-word forms are preserved.…”