Why this answer has a date on it
ChatGPT's audio and file handling changes often and differs between plans and between the web, desktop and phone apps. Check your own plan before you build a workflow on it.
Voice conversation in ChatGPT is a way to talk to it, not a way to get a transcript of an hour-long meeting or interview you already recorded.
What to use instead
A transcription service built for recordings. WhipScribe takes any audio or video file up to 10 hours and 5 GB, or a YouTube, podcast or Google Drive link, detects the language out of about 99, and returns timestamped text, usually in about two minutes per hour of audio.
Export the transcript as TXT and give it to ChatGPT, or as SRT or VTT if you need subtitles.
Or skip the round trip
Next to each transcript WhipScribe writes a summary, takeaways, chapters and the names, numbers and dates that were said, and you can chat over the whole transcript. That is often what people were going to ask ChatGPT for.
What it costs
The summary of any recording is free to read. The full transcript opens with a minute pack: $4 / 500 minutes, $8 / 1,000 minutes, $12 / 2,000 minutes or $24 / 5,000 minutes, one-time. Credits never expire. Workspace is $19/month on the web app.
How to do it
- Open WhipScribe and drop the audio or video file onto the upload box, or paste a public link.
- Read the summary, then open the full transcript with a pack.
- Export as TXT.
- Paste or upload the text into ChatGPT and ask your questions there.