Improving speech-to-text by training new model on domain words

[ameyagh]

We are looking through lectures on ClassTranscribe which have words that were improperly transcribed. We are using that and a list of domain words that are uncommon in everyday speech to train a new model on Microsoft Azure.