isip_extract_hypo [options] trans_db.sof

isip_extract_hypo -help;

isip_extract_hypo -level word -format NIST_TRN -output output.text
trans_db.sof -exclude_symbols ex_sym_sdb.sof

The isip_extract_hypo utility extracts the hypotheses/alignments from the transcription database given the level(s) and a list of identifiers. Note that the output of the recognizer is a transcription database that can be potentially used as an input to the trainer. The list of identifier is optional and by default, all the transcriptions in the database are extracted.

• -debug: debug level (default=NONE, BRIEF, DETAILED, ALL)
• -exclude_symbols: list of symbols that will be excluded in the output hypotheses. It is specified as a Sdb object in sof
• -format: hypotheses/alignments format (default=NIST_TRN, HYP_ALIGN)
• -help: display a help message
• -level: transcription level (default=topmost level)
• -list: list of identifiers as Sdb object (optional)
• -log_file: log messages
• -output: output hypotheses file

• trans_db.sof: input transcription database in sof

• -alignment: output alignments

• -format "word : state" (outputs word, phone & state level alignments)
• -format "word, state" (outputs word & state level alignments)

None.

This utility is written entirely in C++.

None.

• none