I keep seeing the same grammatical mistake and so I started looking into writing a grammar file (.jsgf file) hoping this would solve the issue. However, in this case it could actually be a lot easier to define a list of grammatical structures that are NOT possible. For example, let's say I keep getting "you was" in my output. Is it somehow possible to define that this is NOT possible and thus the next most probable choice should be picked instead?
Thanks,
Soren
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Overall, language model must only help to distinguish the homonyms, acoustic model should be the factor for the accuracy. If you get bad results you need to improve your acoustic model first, it's a bad idea to enforce accurate answer with a language model.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi,
I keep seeing the same grammatical mistake and so I started looking into writing a grammar file (.jsgf file) hoping this would solve the issue. However, in this case it could actually be a lot easier to define a list of grammatical structures that are NOT possible. For example, let's say I keep getting "you was" in my output. Is it somehow possible to define that this is NOT possible and thus the next most probable choice should be picked instead?
Thanks,
Soren
No, there is no such way.
Overall, language model must only help to distinguish the homonyms, acoustic model should be the factor for the accuracy. If you get bad results you need to improve your acoustic model first, it's a bad idea to enforce accurate answer with a language model.