A Norwegian Letter-to-Sound Engine with Danish as a Catalyst
Peter Juel Henrichsen · CBS Research Portal (Copenhagen Business School) · 2007
Danish phonetics and Norwegian phone-tics are not all that different.This fact is exploited in an ongoing project establishing a phonetic transcription algorithm for (East-) Norwegian.Using methods known from machine learning we exploit a publicly available phonetic database for Danish (based on the Danish PAROLE corpus) arriving at a costcompetitive phonetic database for Norwegian.While the ultimate goal of this enterprise is a low-budget complete phonetic transcription of the NoTa corpus of Norwegian spontaneous speech, this paper presents the subparts related to Danish phonetics.Tree structure (dependency based) Rhetorical structure