Hello @reuben and @lissyx i have created my own language model where it include few items like
paneer
butter
cheese etc when I am doing live Inference using audio transcriber GUI with H1n mic recorder when I say “paneer” it gives Inference as “paneer” in most of the cases but for some people when they say “paneer” the result comes a “butter” in 9 cases out of 10.
Can you please explain me the reason behind it and what can I do to deal with such problem?
Sorry if I am lacking some basic understanding.
Hard to give a definitive answer with so few context … It could just be different accents from those people …