I see this also relates to #7
Anyway, the there was a combination of three issues
there was probably an issue with the encoding and you should change the command
probably bug in sequitur
the user was using english g2p model (downloaded from sf.net ). You should be explicit about the origins of the scripts and that not all the files mentioned in the scripts will be compatible with Russian (or modify it so that it is compatible)
Given the fact that the lexicon is using utf-8, this line https://github.com/freerussianasr/recipes/blob/a5368594844df9c748692415b6cf99d24f24f4bf/voxforge_ru/local/voxforge_prepare_dict.sh#L54 should specify
--encoding utf-8
explicitely The error was reported to us here https://github.com/kaldi-asr/kaldi/issues/2500 but I think adding the encoding specs and using latest sequitur should resolve this. I'd appreciate if you could assist in resolving that issue