class DecoderSingleTest

This class is used to recognise a single input data file, post-process the recognition result and output the result.

Public Methods

void removeSentMarksFromActual( Vocabulary* vocabulary )

Internal functions

Public Members

Configures the test

'test_id_' is an ID assigned to this particular test

'test_filename_' is the absolute pathname of the input data file

'n_expected_words_' and 'expected_words_' contain information about

the expected result of the test (ie.

'data_format_' indicates the type of data contained in the input

data file.

'phone_models_' is a pointer to the PhoneModels instance that contains

the phone HMM's.

data (eg.

'remove_sent_marks_' indicates whether the sentence start and end

words in the recognition result are to be removed

'output_result_' indicates whether the recognition result is to be

output after the decoding has occurred

'out_fd' is the file where output will be written (if NULL then

stdout assumed)

'output_ctm_' indicates if the output is to be in CTM format

'frame_msec_step_size' is the frame step size in milliseconds and is

Configures the test.

in a big archive file

Parameters are same as other 'configure' method except 'test_filename_'

has been replaced by 'archive_offset_'

'archive_offset' is the offset into an archive input file where the

Runs the tests using the recogniser pointed to by 'decoder'

'archive_fd' is the big file containing the input data.

this is not NULL, the archive_offset member variable is used

Outputs the recognition result.

recognition result (ie.

strings and displayed.

Documentation

This class is used to recognise a single input data file, post-process the recognition result and output the result. The DecoderBatchTest class contains of an array of these objects.

Configures the test

Configures the test

'test_id_' is an ID assigned to this particular test

'test_id_' is an ID assigned to this particular test

'test_filename_' is the absolute pathname of the input data file

'test_filename_' is the absolute pathname of the input data file

'n_expected_words_' and 'expected_words_' contain information about

'n_expected_words_' and 'expected_words_' contain information about

the expected result of the test (ie.

the expected result of the test (ie. the correct transcription).

'data_format_' indicates the type of data contained in the input

'data_format_' indicates the type of data contained in the input

data file.

data file. Only 1 input data format has been implemented so far.

'phone_models_' is a pointer to the PhoneModels instance that contains

'phone_models_' is a pointer to the PhoneModels instance that contains

the phone HMM's.

the phone HMM's. This used to verify that datafiles contain expected

data (eg.

data (eg. emission probs or features, correct vector sizes)

'remove_sent_marks_' indicates whether the sentence start and end

'remove_sent_marks_' indicates whether the sentence start and end

words in the recognition result are to be removed

words in the recognition result are to be removed

'output_result_' indicates whether the recognition result is to be

'output_result_' indicates whether the recognition result is to be

output after the decoding has occurred

output after the decoding has occurred

'out_fd' is the file where output will be written (if NULL then

'out_fd' is the file where output will be written (if NULL then

stdout assumed)

stdout assumed)

'output_ctm_' indicates if the output is to be in CTM format

'output_ctm_' indicates if the output is to be in CTM format

'frame_msec_step_size' is the frame step size in milliseconds and is

'frame_msec_step_size' is the frame step size in milliseconds and is

Configures the test.

Configures the test. This version is used when we have input data

in a big archive file

in a big archive file

Parameters are same as other 'configure' method except 'test_filename_'

Parameters are same as other 'configure' method except 'test_filename_'

has been replaced by 'archive_offset_'

has been replaced by 'archive_offset_'

'archive_offset' is the offset into an archive input file where the

'archive_offset' is the offset into an archive input file where the

Runs the tests using the recogniser pointed to by 'decoder'

Runs the tests using the recogniser pointed to by 'decoder'

'archive_fd' is the big file containing the input data.

'archive_fd' is the big file containing the input data. When

this is not NULL, the archive_offset member variable is used

this is not NULL, the archive_offset member variable is used

Outputs the recognition result.

Outputs the recognition result. If 'vocab' is defined, then the

recognition result (ie.

recognition result (ie. vocab entry indices) are converted to

strings and displayed.

strings and displayed. If 'vocab' is NULL then the indices are

void removeSentMarksFromActual( Vocabulary* vocabulary )

Internal functions

This class has no child classes.

Author:

Darren Moore (moore@idiap.ch)

Alphabetic index HTML hierarchy of classes or Java