How to Build an Automated TikTok Ad Spy & Hook Analyzer with Python

A developer has released an open-source Python tool, tiktok-subtitle-ripper, that extracts on-screen text overlays and subtitles from TikTok and Reels ads to identify the first three-second hook. The tool uses GPU OCR and can be integrated with AI models to categorize ad angles, offering a low-cost alternative to commercial ad spy platforms. The developer also provides a free web demo on Hugging Face Spaces and a RapidAPI marketplace listing.

In performance marketing and e-commerce, the first 3 seconds of a video ad the "Hook" determines 80% of its return on ad spend. If you analyze winning TikTok and Instagram Reels ads, you'll find that most top-performing ads don't rely on voiceover alone: they burn bold, dynamic text overlays directly onto the video screen. Commercial ad spy platforms like Foreplay or PiPiADS charge anywhere from $99 to $299/month for access to their creative databases. In this tutorial, we will build an automated TikTok Ad Hook Extractor in Python in under 20 lines of code. TikTok / Reels Ad URL │ ▼ Video OCR Ripper Engine ──► Skips static frames, runs GPU OCR │ ▼ Extracted Hook Text First 3s + Discount Codes + Full .SRT Transcript │ ▼ OpenAI / Claude ──► Categorizes the Angle Problem/Solution, Curiosity, Social Proof │ ▼ Notion Database / Airtable Ready for Creative Team We'll use tiktok-subtitle-ripper , a lightweight open-source library that extracts on-screen text overlays and subtitle files directly from video links: pip install tiktok-subtitle-ripper Here is the complete script to extract the visual hook and full transcript from any competitor ad: python import httpx RAPID API KEY = "YOUR API KEY" API URL = "https://tiktok-reels-subtitle-video-ocr-ripper.p.rapidapi.com" headers = { "X-RapidAPI-Key": RAPID API KEY, "X-RapidAPI-Host": "tiktok-reels-subtitle-video-ocr-ripper.p.rapidapi.com" } def analyze ad creative video url: str : 1. Ingest Video URL directly res = httpx.post f"{API URL}/jobs/url", json={"url": video url, "preset": "tiktok reels"}, headers=headers job id = res.json "job id" 2. Fetch OCR Results & Timecodes data = httpx.get f"{API URL}/jobs/{job id}", headers=headers .json 3. Isolate the First 3-Second Hook hook texts = for segment in data.get "segments", : if segment "start" <= 3.0: hook texts.extend d "text" for d in segment.get "detections", print f"🎯 Detected Visual Hook 0-3s : {' | '.join hook texts }" return hook texts Example usage on any TikTok / Reels link: analyze ad creative "https://www.tiktok.com/@competitor/video/123456789" When run against a top dropshipping or brand video ad, the script immediately outputs: 🎯 Detected Visual Hook 0-3s : THE VIRAL 3-STEP ROUTINE | 50% OFF TODAY WITH CODE GLOW50 From here, you can pipe this hook text directly to ChatGPT with a prompt like: "Analyze this hook angle and write 3 variations for my brand." If you want to test the video OCR extraction on your own ads without writing code, you can use the free web demo on Hugging Face Spaces https://huggingface.co/spaces/Jals-builds/tiktok-reels-subtitle-ripper . For developers processing bulk ad libraries, check out the RapidAPI Marketplace Listing https://rapidapi.com/jals/api/tiktok-reels-subtitle-video-ocr-ripper free tier included, with scaling plans starting at $2.99/mo .