Instagram Reels Transcript & On-Screen Text

Turn public Instagram reels into text you can search, quote, and reuse. Each reel comes back with its exact transcript, timestamped sentences, the text shown on screen, and the opening hook, from one pass over the video.

What you get

• Transcript: every spoken word in the reel's original language, never translated.
• Timestamped segments: the transcript split into sentences with start and end times.
• On-screen text: titles, lists, and labels shown in the video, with the second each one appears. Captions that only repeat the speech are left out.
• Music-only reels work too: when nobody speaks, you still get the on-screen text, and hasSpeech is false.
• Hook and summary: the opening line of the first 3 seconds and a one-sentence description.
• Reel data: account, date, duration, views, likes, comments, caption, hashtags, and music.

Who it is for

Content creators, social media managers, agencies, and researchers who study what reels say, plus AI agents that need reels as text.

How to use it

1. Add reel links, usernames, or both. Each username gives you its newest reels.
2. Set Maximum reels to transcribe to control spending.
3. Run it and export the results to JSON, CSV, or Excel, or use them through the API.

Already scraped reels? Pass the dataset ID from apify/instagram-reel-scraper or apify/instagram-scraper in Dataset ID and nothing is scraped again. Instagram video links expire after a few days, so use recent datasets.

Input example

{
  "reelUrls": ["https://www.instagram.com/reel/DfP5dwQgFh7/"],
  "usernames": ["natgeo"],
  "reelsPerProfile": 5,
  "maxReels": 20
}

Output example

Shortened result from a real run:

{
  "status": "transcribed",
  "reelUrl": "https://www.instagram.com/reel/DfP5dwQgFh7/",
  "ownerUsername": "iturrilabs",
  "postedAt": "2026-11-02T17:59:06.000Z",
  "durationSeconds": 33.1,
  "music": "Original audio",
  "language": "en",
  "hasSpeech": true,
  "hook": "This 49-second ad just became a creative brief.",
  "transcript": "This 49-second ad just became a creative brief. Usually you'd pause, rewind, and type it out. So I ran it through Meta Ad Creative Analyzer...",
  "onScreenText": ["THIS 49-SECOND AD", "META AD CREATIVE ANALYZER", "HOOK", "Cutting with a SWEET TOOTH?"],
  "summary": "This video demonstrates how the Meta Ad Creative Analyzer tool extracts hooks, angles, offers, and CTAs from competitor ads.",
  "segments": [
    {"start_seconds": 0, "end_seconds": 3, "text": "This 49-second ad just became a creative brief."},
    {"start_seconds": 3, "end_seconds": 6, "text": "Usually you'd pause, rewind, and type it out."}
  ],
  "onScreenTimeline": [
    {"start_seconds": 0, "text": "THIS 49-SECOND AD"},
    {"start_seconds": 7, "text": "META AD CREATIVE ANALYZER"}
  ],
  "error": null
}

Each record has a status: transcribed, not_found, video_unavailable, too_long, too_large, invalid_input, or failed. Only transcribed reels are charged. A SUMMARY record in the run's key-value store counts the results.

Pricing

You pay $0.01 per transcribed reel. Reels that are missing, private, too long, or fail are not charged.

When you add reel links or usernames, the reels are fetched with apify/instagram-reel-scraper, which is billed separately at its own price (on the free plan, $0.0026 per reel plus $0.001 per run when this page was written). Passing a dataset ID skips that step and its cost.

Set a maximum cost per run to control spending.

Limits

• Only public reels are processed. Private accounts, stories, and photos are skipped.
• Reels longer than 3 minutes or larger than 18 MB are skipped and not charged.
• Transcripts are AI-generated and can contain mistakes, especially with strong background music, overlapping voices, or very small text.
• Timestamps are approximate, to about a second.