Two passes, not twenty
The mistake is treating twenty interviews as twenty separate jobs done end to end. Do it in two passes: get all twenty transcribed, then work across the set. The first pass is mechanical and you should not be sitting through it.
Processing usually runs about two minutes of wall clock per hour of audio, so twenty hour-long interviews is a short wait rather than a day.
Getting twenty files in without twenty uploads
If the recordings are in Google Drive, import the whole folder at once. That is the fastest route for a set this size and it is one action, not twenty.
If they are somewhere else, paste links to them, or use the API with keys to submit them from your own code. One file can be up to 5 GB and up to 12 hours, so length is rarely the constraint.
Per-interview summaries are the unit you compare
Each transcript gets an overview, the takeaways, chapters, and key quotes carrying the timestamp each sits at. Twenty of those is a set you can actually read in an afternoon.
Then use the Library search across all twenty: search what was said, not what you named the file, and pull every interview where a topic came up.
Going back into any single one
When the cross-reading turns up something worth chasing, open that interview and chat over its transcript — ask what was said about it, and follow the timestamped quote back into the audio.
Export whatever you are going to write from as TXT, DOCX or JSON. Transcripts are never used to train AI.
What twenty interviews cost
The first transcript is free at any length, no card and no account needed to start, and the second is $0.99, once per email address.
For the rest, minute packs are one-time with no subscription: $8 / 1,000 minutes, $12 / 2,000 minutes, $24 / 5,000 minutes, and minutes do not expire. If the set is large or ongoing, Workspace is $19/month for unlimited transcription on the web app.
How to do it
- Put all twenty recordings in one Google Drive folder, or gather their links.
- Import the whole folder at once, or submit the files through the API.
- Let the batch process — usually about two minutes per hour of audio.
- Read the twenty overviews and takeaway lists as your first pass.
- Search the Library by what was said to find every interview on a topic.
- Open the interesting ones and chat over the transcript to go deeper.