Hi. I'm looking for help constructing the query parameters to instruct DG to specify that the input is numeric and specify the digits we expect the input to be. If the input does matches the audio, I expect the confidence for each search item to be high. For example if the audio sample is "12345" I want to also tell DG that we expect 1, 2, 3, 4, 5 as the only input. If I pass 99999 for the expected values I want DG to return something that indicates the audio recording does not contain 99999 (perhaps confidence is low?) Here is what I'm passing now: https://api.deepgram.com/v1/listen?search=12345 if I pass search=99999 I get the same confidence for each word. Can DG do this or am I misunderstanding it's function?
#API Questions
lunar fox
tawny rose
The search function will return additional data in the response, including the time in the audio the word was mentioned. It will also match similar terms, or words that might have been missed/transcribed wrong. It will also show you where in a sentence it was said.
I'm not sure we support your use case. We've had requests before, for the submission of truth text. I'll raise this with the product team and see if they have any suggestions