Video On-Screen Text Extractor (OCR) for Reels & TikTok
Pricing
from $4.50 / 1,000 video transcripts
Video On-Screen Text Extractor (OCR) for Reels & TikTok
Read the text shown in TikToks, Instagram Reels, X and Facebook videos: captions, titles, list items and burned-in subtitles, in order of appearance and without watermarks. Ideal for music-only videos. Includes the spoken transcript. About $0.008 per video.
Pricing
from $4.50 / 1,000 video transcripts
Rating
0.0
(0)
Developer
Aemon
Maintained by CommunityActor stats
0
Bookmarked
2
Total users
1
Monthly active users
a day ago
Last modified
Categories
Features
Share

What does Video On-Screen Text Extractor do?
Video On-Screen Text Extractor reads the text shown inside short videos: captions, titles, list items, prices and burned-in subtitles from TikToks, Instagram Reels, X and Facebook videos. It returns each distinct text in order of appearance, without usernames and platform watermarks, together with the spoken transcript. Many Reels and TikToks carry their whole message as text over music, and this is the way to get it out.
- 📝 AI vision OCR: frames are taken across the video and a vision model reads the overlaid text.
- 🧹 Clean output: duplicates, usernames and "TikTok"/"Instagram" watermarks are removed.
- 🎙️ Spoken transcript included, in 90+ languages.
- 🔓 No login: only public videos, no account needed.
- 💸 About $0.008 per video (transcript + on-screen text). If reading the text fails, that part is not charged.
What you get
{"url": "https://www.tiktok.com/@apifylife/video/7338085038258457889","platform": "tiktok","onScreenText": ["of course it took us half a year to get what we actually do","of course we have the best office in the center of Prague","of course you can work anytime","we're Apify and we're hiring!"],"text": "Hi, we're Appify. Of course it took us half a year to get what we actually do...","hasSpeech": true,"durationSec": 32}
Real output of the example run. The on-screen text even spells the brand correctly where the speech recognition heard "Appify".
What can you use it for?
- Music-only and text-only videos: get the message that a plain transcript misses
- Content research: collect the on-screen hooks, lists and tips of top videos
- Accessibility: turn burned-in captions into real text
- AI pipelines: feed both the spoken words and the on-screen text to an LLM
How to extract text from videos
- Click Try for free.
- Paste TikTok, Instagram Reel, X or Facebook video links.
- Click Start. Each video becomes one row with
onScreenTextand the transcript.
Ready-to-run examples
Each example opens a pre-filled setup that you can run in one click and save as your own task:
- Extract Text from TikTok Videos: On-Screen Captions: Read the text shown inside TikTok videos: captions, titles, list items and burned-in subtitles, in order of appearance and without the TikTok watermark, plus the spoken transcript.
- Instagram Reel OCR: Extract Text Shown in Reels: Extract the text shown in Instagram Reels and video posts: captions, titles, lists and burned-in subtitles, in order of appearance and without watermarks, plus the spoken transcript.
- Music-Only Video Text: Read Reels & TikToks Without Speech: Many Reels and TikToks have no speech, only music and text on screen, so a normal transcript comes back empty.
- X (Twitter) Video Text Extractor: On-Screen Text: Extract the text shown in videos posted on X (Twitter): captions, titles, chart labels and burned-in subtitles, in order of appearance, plus the spoken transcript and the post's details.
- Facebook Reel Text Extractor: On-Screen Text & Transcript: Extract the text shown in public Facebook Reels and videos: captions, titles, lists and burned-in subtitles, in order of appearance and without watermarks, plus the spoken transcript.
Pricing
| Event | Price | Per 1,000 |
|---|---|---|
| Video transcript (up to 2 minutes) | $0.006 | $6 |
| On-screen text | $0.002 | $2 |
| AI analysis (optional) | $0.004 | $4 |
| Each additional minute of AI transcription | $0.003 | $3 |
A typical short video costs $0.008. Discounts apply on Apify Starter, Scale and Business plans. Failed, private and removed videos are never charged.
Errors are free and explained
Every link you paste gets a row. If a video cannot be processed, the row has "status": "error", an errorCode and a readable error message, and it is not charged:
| Error code | What it means |
|---|---|
PRIVATE | The video or account is private. |
UNAVAILABLE | The video was removed or does not exist. |
PROFILE_NOT_SUPPORTED | An Instagram profile link: Instagram only lists profiles to logged-in users, so paste Reel links. |
PHOTO_POST_NOT_SUPPORTED | A TikTok photo slideshow, which has no video. |
PLATFORM_NOT_SUPPORTED | A link from a platform this Actor does not handle; the message names the Actor that does. |
NO_VIDEO / NO_AUDIO | The post has no video, or the video has no audio track. |
TOO_LONG | The video is longer than Skip videos longer than. |
LOGIN_REQUIRED | The platform asked for a login (age restriction or bot check). |
RATE_LIMITED / BLOCKED / NETWORK / DOWNLOAD_FAILED | A temporary problem. These are retried automatically up to 3 times first. |
Use it from code, AI agents and automations
Input example
{"urls": ["https://www.tiktok.com/@apifylife/video/7338085038258457889"],"extractOnScreenText": true}
JavaScript
import { ApifyClient } from 'apify-client';const client = new ApifyClient({ token: 'YOUR_APIFY_TOKEN' });const run = await client.actor('aemon/video-on-screen-text-extractor').call({"urls":["https://www.tiktok.com/@apifylife/video/7338085038258457889"],"extractOnScreenText":true});const { items } = await client.dataset(run.defaultDatasetId).listItems();console.log(items);
Python
from apify_client import ApifyClientclient = ApifyClient("YOUR_APIFY_TOKEN")run = client.actor("aemon/video-on-screen-text-extractor").call(run_input={"urls":["https://www.tiktok.com/@apifylife/video/7338085038258457889"],"extractOnScreenText":True})for item in client.dataset(run["defaultDatasetId"]).iterate_items():print(item)
AI agents (MCP): Claude, ChatGPT, Cursor
This Actor works as a tool in any MCP client through the Apify MCP server. Add it to the MCP configuration of Claude Desktop, Claude Code, Cursor or VS Code, with your Apify API token:
{"mcpServers": {"apify": {"url": "https://mcp.apify.com?tools=aemon/video-on-screen-text-extractor","headers": { "Authorization": "Bearer YOUR_APIFY_TOKEN" }}}}
Then ask your agent in plain words, for example: "Read the text shown on screen in this TikTok and summarise it: https://www.tiktok.com/@apifylife/video/7338085038258457889"
Agents without an Apify account can still use it: the Actor accepts agentic payments, so an agent can run it and pay per run with x402 (USDC) or a Skyfire token instead of an API token.
n8n, Make and Zapier
One HTTP request runs the Actor and returns the results, so it fits into any automation tool:
POST https://api.apify.com/v2/acts/aemon~video-on-screen-text-extractor/run-sync-get-dataset-items?timeout=300Authorization: Bearer YOUR_APIFY_TOKENContent-Type: application/json{"urls":["https://www.tiktok.com/@apifylife/video/7338085038258457889"],"extractOnScreenText":true}
In n8n use an HTTP Request node, in Make HTTP → Make a request and in Zapier Webhooks by Zapier → Custom Request. Every video comes back as one item, ready for Google Sheets, Notion, Slack or email. You can also connect integrations and webhooks directly in the Integrations tab.
Ask an AI assistant about this Actor
Open a ready-made question about this Actor in ChatGPT, Claude or Perplexity.
FAQ
How many frames are read? About one frame every 4 seconds, up to 5 frames per video, at a resolution where captions stay legible.
Does it work on YouTube? No, YouTube is not supported for on-screen text. For YouTube transcripts use Video to Text.
Can I also get the hook and summary? Yes, turn on AI analysis, or use TikTok & Reels Hook Analyzer, which is set up for that.
Support
Found a video where text was missed? Open an issue in the Issues tab with the link; issues are answered within a day. If the Actor saves you time, a rating on the Reviews tab helps others find it.
Built and maintained by Aemon.