I am creating a lexicon for ASR (automatic speech recognition) purposes and facing a problem: I grab phonetization of words with espeak which gives me something like this: