Location name transcription error fix needed
Hi, do you have a solution for instructing the bot to understand some phrases differently than it is transcribed by the transcriber? For instance, during the call, the user is saying "Amwaj" but transcribers (I tested all of them and the result is the same) convert it into "I'm watch". I added the instruction in the prompt to handle it but it doesn't work. The instruction is: If the assistant asks for the location and the user's answer contains the phrase "watch" or "Unwatched" or "How much" then it means "Amwaj". Any idea how to fix it?