AI Viral Clip Cutter — Long Video to Shorts
Pricing
from $200.00 / 1,000 clip generateds
AI Viral Clip Cutter — Long Video to Shorts
Turn a long video (podcast, interview, stream, talk) into ranked short clips automatically. Transcribes, finds the most viral moments with AI scoring, cuts them to 9:16 with a blurred background, and burns word-synced captions. For TikTok, Reels, and Shorts repurposing.
Pricing
from $200.00 / 1,000 clip generateds
Rating
5.0
(1)
Developer
Dami's Studio
Maintained by CommunityActor stats
0
Bookmarked
1
Total users
0
Monthly active users
2 days ago
Last modified
Categories
Share
AI Viral Clip Cutter
Takes one long video (a podcast, interview, stream, or talk) and returns a handful of short clips ranked by how likely they are to do well on social. It transcribes the audio, has an LLM score every candidate moment for virality, then cuts the top ones to vertical 9:16 with a blurred background and burns word-synced captions on top. Made for people who already publish shorts and don't want to scrub through two hours of footage to find the three usable minutes.
How it works
Download, extract audio, transcribe with word-level timestamps, ask a chat model to find and score self-contained moments, then for each kept moment ffmpeg cuts the segment, reframes to the target aspect ratio (blurred fill, no hard crop), and libass burns the captions. Runs CPU-only, no GPU.
Input
| Field | Required | Notes |
|---|---|---|
videoUrl | yes | Public direct URL to the source video (.mp4/.mov/.webm). Pair it with a YouTube downloader actor to feed YouTube links. |
maxClips | no | How many clips to produce, highest-scoring first. Default 5. |
minScore | no | Virality threshold, 0-100. Moments below this are dropped. Default 60. Lower it if you get back fewer clips than you wanted. |
aspectRatio | no | 9:16, 1:1, 16:9, or original. The two portrait options use a blurred background instead of cropping. Default 9:16. |
burnCaptions | no | Burn word-synced captions into each clip. Default true. |
captionPreset | no | Caption look: karaoke, hormozi, beast, tiktok, or clean. |
language | no | Spoken-language ISO code (en, es, ...) or auto. |
openaiApiKey | no | Your OpenAI key, used for both transcription and moment scoring. Bring your own and you only pay the per-clip result fee. |
Output
One dataset record per clip, ordered by rank, plus the rendered MP4 in the run's key-value store. Each record carries the score (0-100), a short title and reason for why it was picked, suggested hashtags, the source start/end and durationSeconds, and clipKey / clipUrl for downloading the finished video. Failed clips come back with ok: false and an error so you can see what dropped.
Example
Pricing
$0.35 per clip, pay per result, no subscription. You bring your own OpenAI key for the steps that use a model.
Notes
Scoring quality depends on the transcript, so clean speech does noticeably better than noisy or heavily overlapping audio. Without an OpenAI key the transcription and scoring steps can't run, so the key is effectively required even though the field is optional. You're responsible for having the rights to whatever video URL you submit.
