Abstract: In this paper, we describe the work of the LUPo project to integrate non-standard Luso-African varieties from Cape Verde and Mozambique into a rule system for generating accent-specific phonetic transcriptions for these and other spoken variants of Portuguese. Here, we present a description of LUPo's functions for online use, LUPo's architecture, and the manner of linguistic data collection and analysis that support this application. Implications for a subsequent text-to-speech (TTS) module are also presented in terms of yielding high-quality pronunciation lexica for regional variants of Portuguese, and towards facilitating the entry of hitherto untreated speech varieties into the digital domain.
Index Terms: pronunciation lexicon, speech synthesis, Luso-African speech varieties, digitally endangered speech varieties.