Turn one long video
into great shorts.
Two ways to work. Mark clips yourself in the browser and export a clean edit decision list, or upload the video and let our server transcribe it, score its moments, and cut captioned vertical clips for you. Free.
Plan your shorts.
Two ways to work. Mark clips yourself in the browser, or have the server transcribe your video and score its moments for you. Pick a tab.
Choose a video file to get started
MP4, MOV, WebM, AVI — any format your browser supports
Your clip list autosaves to browser storage. Your video file itself cannot be saved — when you return, choose the same file again and your marks will still be there.
Four steps to a clip list.
Upload your video
Choose any video file from your device. It loads directly in your browser — nothing is sent to a server.
Mark your moments
Play or scrub the video. Hit "Set Clip Start" and "Set Clip End" at the moments worth turning into shorts.
Caption and note
Add a title, the on-screen caption text, and any b-roll or production notes for each clip before adding it to your list.
Export and produce
Download your edit decision list as a .txt or .json file. Produce the clips yourself, or hand the list to our Studio.
What exists. What is coming.
No vaporware. Here is exactly what this tool does today, and what it deliberately does not.
The clip planner
- Upload any video — stays on your device
- Native video player with controls and scrubbing
- Set in/out points by clicking at the current playback position
- Per-clip title, caption, and b-roll notes
- Visual warning if a clip exceeds 60 seconds
- Jump to a clip's start time in the player
- Edit and delete individual clips
- Autosave to browser storage; survives page refreshes
- Export as plaintext EDL (.txt) or structured (.json)
- Copy the full clip list to clipboard
- Export a JPEG summary card
Automatic clipping
- Upload a video; the server transcribes it with whisper.cpp on the CPU
- Moments ranked by a rule-based transcript scorer, best first
- Every score shown with its full breakdown — each feature, its weight, and the points it contributed
- Cut to vertical 9:16 with three reframe modes
- Four caption styles, burned in from the transcript timings
- Render several clips from one upload and download each as .mp4
- Send any suggested moment into the manual planner to edit by hand
- Delete your upload and every clip from our server on demand
What we have not done
- No "viral potential" prediction. Nothing here has been validated against real post performance, so we do not claim it.
- No speaker labels — it cannot tell you who said what
- No languages other than English
- No automatic posting to any platform
- No written-for-you captions or titles — captions are your own words, transcribed
- No music, b-roll, or transitions added for you
What it actually does.
No magic and no language model. Four ordinary programs run in order on our server.
Transcribe, score, cut
- ffmpeg extracts the audio track from your upload.
- whisper.cpp transcribes that audio on the CPU, producing word-level timings. This is the slow step.
- A rule-based scorer ranks windows of the transcript on six measurable features: how the segment opens, words per second, loudness relative to the file average, variation in delivery, whether it starts on an unresolved reference, and whether it ends on a complete sentence.
- ffmpeg cuts the segments you pick, reframes them to 9:16, and burns in captions from the transcript timings.
Where it falls short
- It does not understand your video. It scores transcript patterns. A brilliant silent visual moment scores zero.
- Every score is shown with its full arithmetic breakdown so you can judge whether the ranking deserves your trust on your footage.
- Accuracy drops on noisy audio, music beds, crosstalk and strong accents. Our speed figures come from clean single-speaker speech.
- English only — the transcription model in use is English-only.
- No speaker labels. It cannot tell you who said what.
- One job runs at a time on shared hardware, so you may wait in a queue.
- You can always fall back to the manual planner and mark clips yourself.