Improving STT Language Model
Opened this issue · 0 comments
jkelly80 commented
Our current STT system is working, but the accuracy is rather poor. According to the documentation, we can improve the accuracy of our system by tweaking the language model that the library uses to transcribe the audio. One possible approach to doing this is reducing the number of words the model is looking for, and instructing users to use purposefully simple language. This may be a necessary tradeoff due to the limited computation resources available.