Data-Oriented Methods for Grapheme-to-Phoneme Conversion

Data-Oriented Methods for Grapheme-to-Phoneme Conversion Antal van den Bosch and Walter Daelemans ITK (Institute for Language Technology and AI) Tilburg University P.O. Box 90153, NL-5000 LE Tilburg Tel: +31 13 663070 Email: antalb~kub.nl, walter@kub.nl Abstract It is traditionally assumed that various sources of linguistic knowledge and their in- teraction should be formalised in order to be able to convert words into their phone- mic representations with reasonable accu- racy. We show that using supervised learn- ing techniques, based on a corpus of tran- scribed words, the same and even better performance can be achieved, without ex- plicit modeling of linguistic knowledge. In this paper we present two instances of this approach. A first model implements a variant of instance-based learning, in which a weighed similarity metric and a database of prototypical exemplars are used to pre- dict new mappings. In the second model, grapheme-to-phoneme mappings are looked up in a compressed text-to-speech lexicon (table lookup) enriched with default map- pings. We compare performance and accu- racy of these approaches to a connectionist (backpropagation) approach and to the lin- guistic knowledge-based approach. 1 Introduction Grapheme-to-phoneme conversion is a central task in any text-to-speech (reading aloud) system. Given an alphabet of spelling symbols (graphemes) and an alphabet of phonetic symbols, a mapping should be achieved transliterating strings of graphemes into strings of phonetic symbols. It is well known that this mapping is difficult because in general, not all graphemes are realised in the phonetic transcription, and the same grapheme may correspond to different phonetic symbols, depending on context. It is traditionally assumed that various sources of linguistic knowledge and their interaction should be formalised in order to be able to convert words into their phonemic representations with reasonable accu- racy. Although different researchers propose differ- ent knowledge structures, consensus seems to be that at least morphological and phonotactic knowledge should be incorporated in order to be able to find morphological and syllable structure. These struc- tures are deemed necessary to define the proper do- mains for phonological and phonetic rules. As a typi- cal architecture for grapheme-to-phoneme conversion in Dutch, consider the modules in [Daelemans, 1988] shown in Figure 1. It contains most of the traditional datastructures and processing components proposed by computational linguists. A problem with this approach is that the knowl- edge needed is highly language-dependent and re- quires a significant amount of linguistic engineer- ing. We argue that using data-oriented learning tech- niques on a corpus of transcribed words (information. which is readily available in many machine-readable dictionaries), the same and even better performance can be achieved, without explicit modeling of linguis- tic knowledge. The advantages of such an approach are that the technique is reusable for different sets of data (e.g. different languages or sublanguages), and that it is automatic (no explicit linguistic engi- neering is needed to handcraft the rules and knowl- edge structures necessary for implementing the tar- get mapping). In this paper we present two instances of this approach in the domain of Grapheme-to-Phoneme conversion. …

Content truncated to 500 words