How to get Instagram Reel transcripts in bulk - text, SRT and VTT from any spoken Reel, pay only for delivered transcripts

Turning an Instagram Reel into text is the first step for a lot of real work: repurposing content, research, accessibility, feeding clips to an AI. Instagram gives you no transcript button and no API for it, screenshots of captions are useless, and most extractors need the Reel to already have captions - most do not.

So I built an extractor that does it the direct way: it listens to the Reel.

1. Give it a list of public Reel URLs (up to 25 per run).
2. It downloads the audio and transcribes the actual speech with Whisper - no captions needed, any spoken language, auto-detected.
3. You get back, per Reel: the clean full text, timed segments, and ready-to-use SRT and WebVTT subtitle files, plus author, caption, publish date, duration and view/like counts.

Try it with zero setup - this ready-to-run task transcribes 3 cooking Reels end to end:
Cooking Reels - Instagram Transcripts with SRT & VTT

Real numbers from the first production test: 3 Reels attempted, 2 delivered in about 70 seconds (158 and 183 words of clean transcript each). The third Reel does not expose its audio track to anyone - Instagram serves that one video-only - so it was classified, skipped, and NOT charged. That is the whole pricing model: you pay only for transcripts actually delivered, $9.90 per 1,000. The established extractor on the store charges $15 per 1,000.

A note on what "delivered" means here: if a Reel is music-only there is no speech to transcribe, and if Instagram hides its audio completely there is nothing to hear - in both cases the Reel costs you $0 and the run log tells you exactly which case it was.

Honest limits: speech quality follows audio quality - clear narration transcribes beautifully, noisy street audio less so - and Instagram rate-limits bursts, so huge batches run slower than small ones.

It is also callable from code and from AI agents: full API access, and it works as a tool through Apify's MCP server, so Claude, Cursor or your own agent can pull Reel transcripts directly.

I am a solo dev and this just launched. If you try the cooking Reels task, the thing that helps most is an honest review on the Actor page - the first reviews are what get a tool like this discovered.