A parser for continuous speech has to deal with lattices where the word hypotheses of the correct sentence are not usually perfectly aligned and short function words can be missing. To cope with these problems, a two-way interaction between the recognition module and the parser, called feedback verification procedure (FVP), has been investigated. The parser generates many solutions, that are fed back to the front-end processor (FEP) which realigns them against the acoustical data, finds the missing function words among the given candidates and attributes them a new score. The best scored solution is finally selected by the parser.
Bibliographic reference. Baggia, Paolo / Fissore, Lorenzo / Gerbino, E. / Giachin, Egidio P. / Rullent, C. (1991): "Improving speech understanding performance through feedback verification", In EUROSPEECH-1991, 211-214.