Were you able to solve your problem and can you share a code solution? It seems that Apple's Speech Recognition framework is limited to 1000 requests per hour and device, and can process at most 1 minute of transcription, so I am not sure how you manage to transcribe 1+ hour of audio.

Would love to know the answer to that as well. It would be amazing to be able to automate the funneling of App Store data into Salesforce.