Directory Name:
speech
Introduction:
The Speech utilities are the highest level utilities providing
basic speech processing tools.
Example:
The standard example of the utililty invocation is:
isip_recognize -param param.sof -list decode_list.sof
This reads the configuration from the parameter file "param.sof"
and execute the specified operations on the input file list
"decode_list.sof".
To see the help message available for the utility, try:
which will display this message:
isip108_[1]: isip_recognize -help
name: isip_recognize
synopsis: isip_recognize -param param.sof [-list decode_list.sof]
description: speech recognizer, transforms speech to text
example: isip_recognize -par param.sof -list decode_list.sof -output_mode FILE
-output_file hypothesis.out
flags:
-verify : run without actual decoding
-help : this message
options:
-output_mode : output mode specification (FILE, TRANSFORM, LIST)
-output_format : output file format (NIST_SCORE, SOF)
-debug_level : debugging information level (default NONE)
-verbose : log verbosity level (NONE, BRIEF, DETAILED, ALL)
arguments:
-param param.sof : parameters file
file1.raw file2.sof ... : input file sequence
-list decode_list.sof : input file list
-output_file hypotheses.out: output file name in FILE mode
-output_list out_list.sof : output files list in LIST mode
-output_dir ./output : output directory in TRANSFORM mode
-dir_pres_level N : num. of preserved directories in TRANSFORM mode
-log_file decode.log : log file used instead of stdout
The command line options allways override the parameter file values.
Multiple files can be listed on the command line using standard
Unix/shell conventions.
Example:
In this second example, we demonstrate the use of the isip_transform
utility.
isip108_[1]: isip_transform.exe file.raw -param params.sof -output_dir ./out
This command performs input file "file.raw" transformation
specified in the parameter file "params.sof"
and places the output file to the directory "out".
Utilities:
The following utilities (applications) can be found in the
speech directory: