5.5.1 Cross-Word CD Phones: Triphone Generation
To begin, we must generate the triphones. For cross-word context dependent
models, only triphones are used to model speech since the models cross
word boundaries. Once again, a list of all
units seen in the transcription database will be generated. For an
explanation of the purpose of this step, see
Section 5.4.1.
Go to the directory:
$ISIP_TUTORIAL/sections/s05/s05_05_p01/
From this directory, run the following command:
isip_recognize -param params_context.sof -list $ISIP_TUTORIAL/research/isip/databases/lists/identifiers_train.sof -verbose all
Expected Output:
Command: isip_recognize -param params_context.sof -list $ISIP_TUTORIAL/research/isip/databases/lists/identifiers_train.sof -verbose all
Version: 1.23 (not released) 2003/05/21 23:10:45
loading audio database: $ISIP_TUTORIAL/research/isip/databases/db/tidigits_audio_db.sof
*** no symbol graph database file was specified ***
loading transcription database: $ISIP_TUTORIAL/research/isip/databases/db/tidigits_trans_word_db.sof
loading front-end: $ISIP_TUTORIAL/recipes/frontend.sof
loading language model: $ISIP_TUTORIAL/models/xword_phone_models/lm_xword_jsgf.sof
loading statistical model pool: $ISIP_TUTORIAL/models/xword_phone_models/smp_xword.sof
loading configuration file: $ISIP_TUTORIAL/sections/s05/s05_04_p01/config.sof
processing file 1 (ae_12a): $ISIP_TUTORIAL/research/isip/databases/sof_8k/train/ae_12a.sof
retrieving annotation graph for identifier: ae_12a, level: word
transcription: ONE TWO
processing file 2 (ae_1a): $ISIP_TUTORIAL/research/isip/databases/sof_8k/train/ae_1a.sof
retrieving annotation graph for identifier: ae_1a, level: word
transcription: ONE
processing file 3 (ae_2789385a): $ISIP_TUTORIAL/research/isip/databases/sof_8k/train/ae_2789385a.sof
....
The
context list
generated by the step above can now be used to generate
the triphone model file. From the same directory, run the command:
isip_recognize -param params_generate.sof -verbose all
Expected Output:
Command: isip_recognize -param params_generate.sof -verbose all
Version: 1.23 (not released) 2003/05/21 23:10:45
loading audio database: $ISIP_TUTORIAL/research/isip/databases/db/tidigits_audio_db.sof
*** no symbol graph database file was specified ***
loading transcription database: $ISIP_TUTORIAL/research/isip/databases/db/tidigits_trans_word_db.sof
loading front-end: $ISIP_TUTORIAL/recipes/frontend.sof
loading language model: $ISIP_TUTORIAL/models/xword_phone_models/lm_xword_jsgf.sof
loading statistical model pool: $ISIP_TUTORIAL/models/xword_phone_models/smp_xword.sof
loading configuration file: $ISIP_TUTORIAL/sections/s05/s05_04_p01/config.sof
The context independent phone models generated and trained in
Section 5.3
are used to build the triphone acoustic model. For a more in-depth
explanation of triphone generation, see
Section 5.4.1.
|