Short Video to AI-Ready Text: Instagram Reels Transcript + On-Screen Text

Paste Instagram Reel links, get back everything the video says, both spoken and written on screen, as clean JSON with timestamps. Each video also gets a ready-to-paste text block for ChatGPT, Claude, Gemini, or your own pipeline.

$15 per 1,000 videos ($0.015 per video). You pay only for videos that were processed successfully. No login, no cookies, no Instagram account needed.

What you can use it for

• Content research and swipe files. Pull the exact hooks, scripts and on-screen text of top-performing Reels in your niche, then feed them to an AI model to find patterns.
• Hook analysis. Every video gets a hook field with what is said and what is shown in the first 3 seconds, the part that decides whether people keep watching.
• Competitor and brand monitoring. Track what competitors say and show in their Reels: offers, prices, claims, calls to action.
• AI agents and RAG. Turn short-form video into searchable text. The llm_text field is one block per video, ready for embeddings or a prompt.
• Repurposing. Get the full script of a Reel (yours or a public one) to rewrite as a post, thread, newsletter or blog.

Why this Actor

	This Actor	Typical transcript Actor
Spoken transcript with timestamps	Yes (Whisper large-v3-turbo)	Yes
On-screen text (titles, burned-in captions, overlays, name tags) with timestamps	Yes	No
Overlay vs. caption tagging (text that is not spoken)	Yes	No
Hook (first 3 s, spoken + on-screen)	Yes	No
Music-only videos	Empty transcript + speech_seconds (no invented "Thank you." lines)	Whisper-based tools may invent text on music
Price	$15 / 1,000 videos, any length	Often charged per minute

A lot of Reels say half their message only in text on screen (lists, prices, punchlines, explainer text over music). A speech-only transcript misses all of it. This Actor reads both.

Sample output

From a public TED Reel (https://www.instagram.com/p/DeC12b5uha9/), shortened:

{
  "url": "https://www.instagram.com/p/DeC12b5uha9/",
  "platform": "instagram",
  "creator": "TED Talks",
  "creator_handle": "ted",
  "published_at": "2026-10-03T19:45:06+00:00",
  "duration_s": 45.2,
  "likes": 27515,
  "comments": 223,
  "language": "en",
  "speech_seconds": 45.2,
  "hook": {
    "spoken": "Music is about life. In music, we learn that being in harmony with the world",
    "on_screen": "Jacob Collier Musician"
  },
  "transcript": [
    { "start": 0.0, "end": 1.2, "text": "Music is about life." },
    { "start": 2.06, "end": 4.66, "text": "In music, we learn that being in harmony with the world" }
  ],
  "transcript_text": "Music is about life. In music, we learn that being in harmony with the world doesn't mean ...",
  "on_screen_text": [
    { "start": 0.0, "end": 6.0, "text": "Jacob Collier Musician", "position": "bottom", "kind": "overlay" },
    { "start": 0.0, "end": 2.5, "text": "Music is about life.", "position": "bottom", "kind": "caption" }
  ],
  "llm_text": "VIDEO: https://www.instagram.com/p/DeC12b5uha9/\nCREATOR: TED Talks | DURATION: 45.2s | LIKES: 27515 ...\nHOOK (first 3s): spoken=\"Music is about life...\" on-screen=\"Jacob Collier Musician\"\nTRANSCRIPT:\n[0:00.0] Music is about life.\n...",
  "status": "ok"
}

Output fields

Field	Meaning
transcript	Spoken words as timestamped segments (seconds). Empty when the video has no speech.
transcript_text	The transcript as one string.
on_screen_text	Text shown in the video: start/end (seconds), position (top/middle/bottom), kind = caption (repeats the speech) or overlay (text that is not spoken).
hook	What is said and shown in the first 3 seconds.
speech_seconds	Seconds of detected voice. Lets you filter music-only videos.
caption, likes, comments, published_at, creator, creator_handle, duration_s	Post metadata. views is null: Instagram does not show view counts to logged-out visitors.
language	Detected spoken language (ISO code).
llm_text	Everything above as one plain-text block, ready to paste into an AI model.
status	ok, or error with an error message. Failed videos are not charged.

How to use it

1. Paste one or more public Reel URLs into Video URLs (instagram.com/reel/... or instagram.com/p/...).
2. Click Start.
3. Download the results as JSON, CSV or Excel, or read them through the Apify API, integrations (Make, Zapier, n8n) or the Apify MCP server for AI agents.

Input example:

{
  "urls": [
    "https://www.instagram.com/p/DeC12b5uha9/",
    "https://www.instagram.com/reel/XXXXXXXXXXX/"
  ]
}

A typical Reel takes about 10 seconds. Videos are processed one after another within a run. For large batches, split the URLs over several runs to process them in parallel.

Pricing

• $0.015 per successfully processed video ($15 per 1,000), regardless of video length (tested up to 3 minutes).
• Videos that fail (private, deleted, region-locked, unavailable) are not charged.
• Set Maximum cost per run in the run options to cap your spend. The Actor stops cleanly when the cap is reached.

Settings

Setting	Default	What it does
Video URLs	(required)	Public Instagram Reel URLs.
Read on-screen text	on	Turn off if you only need the spoken transcript.
Add llm_text field	on	One ready-to-paste text block per video.
Proxy configuration	automatic	Leave empty. The Actor uses Apify datacenter proxy and switches to residential automatically if Instagram blocks a request.

Limitations

• Public videos only. Private accounts, deleted posts and posts that require login cannot be processed. They return status: "error" and are not charged.
• On-screen text is most accurate on designed text (titles, captions, quotes, explainer text). Text that is physically in the scene (signs, jerseys, screens) is read less reliably and partly filtered out. Fast word-by-word captions are sampled, not captured word for word; the spoken transcript covers those words.
• Songs: lyrics may appear in the transcript; pure music returns an empty transcript.
• View counts are not available without logging in.
• TikTok or Facebook? Use the sister Actors TikTok Transcript + On-Screen Text and Facebook Reels Transcript + On-Screen Text, same output format.
• Podcasts, audio files or other video links? Use Audio & Video to Text (priced per minute).
• Need frames? Video Frame Extractor returns the key frames of the same links as image URLs, each with its timestamp and the text visible on it ($2 per 1,000 frames).

FAQ

Do I need an Instagram account or cookies? No. The Actor reads only what any logged-out visitor can see.

Which languages are supported? Speech in about 100 languages (Whisper large-v3-turbo). On-screen text is tuned for Latin-script text (English and European languages).

Is this legal? The Actor processes publicly available content and collects no personal data beyond the public post's creator name and handle. You are responsible for how you use the output under the platform's terms and the laws that apply to you (e.g. copyright, GDPR).

Something wrong with a result? Open an issue on the Actor's Issues tab with the video URL. Reliability fixes ship within 24 hours.