5.4.4 Word Internal CD Models: Training State Tied Models
Now that the states are tied, significant changes have been made to the original acoustic and language model files. To increase the accuracy of the models, they must be trained again. This training process is no different than the process described in Section 5.4.2 From the directory:
Command: isip_recognize -param params_train.sof -list $ISIP_TUTORIAL/research/isip/databases/lists/identifiers_train.sof -verbose all 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/winternal_phone_models/lm_winternal_jsgf_test.sof loading statistical model pool: $ISIP_TUTORIAL/models/winternal_phone_models/smp_winternal_test.sof loading configuration file: $ISIP_TUTORIAL/sections/s05/s05_04_p04/config.sof starting iteration: 0 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 average utterance probability: -70.418975712455136, number of frames: 110 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 average utterance probability: -67.376354835879184, number of frames: 87 processing file 3 (ae_2789385a): $ISIP_TUTORIAL/research/isip/databases/sof_8k/train/ae_2789385a.sof ... |
