i am making a small transcription service and am thinking about my best options, any advice is very welcome.
i'm modifying the google voice aiy demo to do long form (~1hr) transcripts. for this i need to store it in a google cloud storage bucket.
i have found that transcription takes about the same time as the length of the recording. (if this is incorrect, please tell me!). so for hour long recordings waiting around for the transcript to finish is too long. so i figured i would start uploading to google cloud storage in the background.
what is the best strategy here?
should i upload in chunks, and transcribe it in the background? this would mean the chunks have to overlap, won't this cause duplications in the transcript?
should i "stream" directly to google cloud storage? and not store anything locally?
should i transcribe directly? how would i do this for hour long sessions? (synchronous transcription for google cloud speech is under 1 min)
i just want to minimize any waiting times for the transcript, and any errors caused by this.