Downloading raw audio from Dialogflow ES

Question

Having an issue where google is not recognizing the speech correctly in a Dialogflow ES deployment for alphanumeric character by character strings.

Example: audio has "b q 5 2 2 1 2", but, the initial b is sometimes missing or being recognized as a d, etc. If it's said more slowly and carefully then it works much better, but... Since normal google speech to text recognizes at a "normal" pace, I'd hope this would be fine as well.

I want to make sure there aren't any issues with the audio being provided to Google. Is there a way to download the .wav file (or base64 for the .wav and I can convert) that Google uses for recognition from the Google side (limited options on the system integrating with Google)?

and/or, does anyone have any suggestions on troubleshooting this situation from the Dialog flow perspective? Again, history shows the text not being recognized properly.

Thanks!