Now that we have run a few participants we can get an idea of the language model (the sentences and vocabulary) used in the test. Instead of using the Wall Street Journal language model from Sphinx we should retrain the model for the simpler speech of the bilingual aphasia test.
Alternatively, we are looking into useing other transcription API see the Teamates nodejs knockout project which we forked recently.
Now that we have run a few participants we can get an idea of the language model (the sentences and vocabulary) used in the test. Instead of using the Wall Street Journal language model from Sphinx we should retrain the model for the simpler speech of the bilingual aphasia test.
Alternatively, we are looking into useing other transcription API see the Teamates nodejs knockout project which we forked recently.