Best Instagram transcript APIs: turn reels into text

November 8, 2026

Most of what a reel says is spoken, not written in the caption. A transcript makes it searchable, quotable and easy to summarize. There are two routes: get the video file and run speech-to-text yourself, or send the reel URL to an API that returns text.

At a glance

Whisper on your own machine

OpenAI's Whisper is open source, with code and weights under the MIT license. Models range from tiny (about 1 GB of VRAM) to large (about 10 GB), and turbo is an optimized large-v3 that needs about 6 GB. It needs ffmpeg installed source.

pip install -U openai-whisper
whisper reel.mp4 --model turbo --word_timestamps True --output_format json

Good at: zero marginal cost, privacy, and word-level timestamps when you need exact timing. Weak at: you must get the video file first, and you need a GPU for speed at volume.

OpenAI's transcription API

If you have the file but no GPU, the hosted API lists Whisper and gpt-4o-transcribe at $0.006 per minute and gpt-4o-mini-transcribe at $0.003 per minute source. A 30-second reel costs a fraction of a cent. Getting the file is still on you.

Supadata

A transcript API that takes Instagram URLs, including /reel/, /p/ and /tv/ links. Instagram transcripts are AI-generated at 2 credits per minute of video; long videos may return a job id to poll source. There is a free tier of 100 credits a month, and paid plans such as Pro at $17 a month for 3,000 credits source. Good at: longer Instagram videos, and YouTube, TikTok, Facebook and X in the same tool.

ScrapeIO

Send a reel URL to the instagram transcript API and get the text back. It costs 2 credits per reel and works on reels under 2 minutes. If a reel has no speech or is not found, the call is free.

curl "https://scrapeio.com/v2/instagram/media/transcript?url=https://www.instagram.com/reel/SHORTCODE/" \
  -H "x-api-key: YOUR_KEY"

Good at: short reels at volume, next to the profile, reel and comment data in the same schema. Weak at: videos over 2 minutes, and it returns text without word timestamps. For timing, use Whisper.

A pipeline from account to searchable text

1. List an account's reels with the instagram user reels API.
2. Keep reels under 120 seconds using duration_seconds.
3. Request a transcript for each.
4. Store text with the reel id, plays and created_at.
5. Search or summarize across the set.

import os, requests
API = "https://scrapeio.com"
H = {"x-api-key": os.environ["SCRAPEIO_API_KEY"]}

reels = requests.get(f"{API}/v1/instagram/user/reels", params={"handle": "natgeo"}, headers=H).json()["data"]["items"]
for r in reels[:10]:
    if (r.get("duration_seconds") or 0) >= 120:
        continue
    t = requests.get(f"{API}/v2/instagram/media/transcript", params={"url": r["url"]}, headers=H).json()
    if t.get("success"):
        text = " ".join(x["text"] or "" for x in t["data"]["transcripts"])
        print(r["plays"], text[:120])

Ten reels cost about 21 credits: one for the list and two per transcript.

Quality checks

• Music-only reels produce empty or junk text. Drop transcripts under five words.
• Check the language. Mixed-language reels are common; any engine can misread them.
• Proper nouns and product names are where errors cluster. Spot-check ten before you trust a batch.
• Store the reel id with every transcript, so you can return to the source when a quote matters.

Doing it with ScrapeIO

Pair the instagram transcript API with the instagram reels search API to transcribe the top reels for any keyword. Our transcripts guide covers YouTube and TikTok with the same response shape.

Sources

Checked on 8 November 2026 from each vendor's own pages. Prices and limits change; follow the links before you decide.

• https://github.com/openai/whisper
• https://developers.openai.com/api/docs/pricing
• https://supadata.ai/instagram-transcript-api
• https://supadata.ai/pricing