isip_segment_concat [options]

isip_segment_concat -help;

isip_segment_concat -min_max -min_max_file min_max.sof -output_type text \
 -list ident_sdb.sof -audiodb audiodb.sof -input_format raw -input_type binary \
-dimension 39 -verbosity brief;

isip_segment_concat -duration -normalize -min_max_file min_max.sof -list ident_sdb.sof \
-audiodb audiodb.sof -input_format raw -input_type binary  \
-dimension 39 -ratio 3:4:3 -transdb transdb.sof -level phone -output_type text \
-output_format sof -suffix text -extension sof -verbosity brief;

isip_segment_concat -duration -normalize -min_max_file min_max.sof -list ident_sdb.sof \
-audiodb audiodb.sof -input_format raw -input_type binary -symbol_list symbols_sdb.sof \
-dimension 39 -ratio 3:4:3 -transdb transdb.sof -level phone -output_type text \
-output_format sof -suffix text -extension sof -verbosity brief;

The isip_segment_concat utility has two purpose:


• Generate the minimum and maximum value for each dimension given the input feature-vectors. These minimum and maximum vectors can then be used for normalizing the input feature-vectors between [1, -1].
  
  
• Generate the segmental features given the ratio. Let say the ratio is 3:4:3. In this case, the feature-vectors corresponding to a typical alignment is divided into three parts following the ratio 3:4:3. The feature-vectors in each of these portions are averaged to form a single feature vector. All the three averaged feature vector corresponding to the three portions are then concatenated. A log of length of this aligment is appended to the output concatenated feature-vector if the duration flag is specified from the commandline. The segmental features can be generated on an utterance basis or symbol basis. In the former case, an entire input feature file is converted into segmental feature file based on the symbol alignments. However, in the latter case, only the features corresponding to the input symbol are converted into segmental features.
  
  

• -ratio: ratio used to concat the segmental features
• -min_max_file: output min_max file
• -list input identifier list as sdb
• -audiodb: audio database specifying the input feature files
• -input_format: input feature file format(default=sof, raw)
• -symbol_list: input symbols list as sdb
• -input_type: input feature file type(default=binary,text)
• -output_format: output feature file format(default=sof, raw)
• -output_type: output feature/min-max file type(default=binary,text)
• -transdb: transcription database specifying the aligments
• -level transcription level for the alignments
• -debug: debug level (default=NONE, BRIEF, DETAILED, ALL)
• -help: this message
• -log_file: log messages
• -verbosity: verbosity level (default=NONE, BRIEF, DETAILED, ALL)

• none

• -min_max: generate min_max file
• -normalize: normalize input features before computing segmental features
• -duration: log-duration of the alignment length added to the output feature-vector

• none

This utility is written entirely in C++.

• none

• none