I am using Speech to Text API to transcribe audio files. I see that the output contains a lot of characters which might be possibly occurring due to inverse text normalization somewhere. Symbols like $ for dollars and other currency symbols and also numbers written in numeric format rather than words. Is there some option in RecognitionConfig which gives me verbatim output in words instead of numbers and symbols?
I see there's a "transcriptNormalization" option in the config but then I have to provide my own rules.
LinkedIn
Twitter
