AI Cut API
AI Cut API: remove silences, filler words and bad takes automatically
Send a talking-head recording and get back one clean MP4, with a list of every part that was cut and why.
Cloud API on the Starter and Premium plans, or the Desktop API on your own computer.
# 1. Get an upload link, then PUT the file to upload_urlcurl -X POST https://api.vcp.name/v1/uploads \ -H "Authorization: Bearer $VCP_API_KEY" \ -H "Content-Type: application/json" \ -d '{"content_type": "video/mp4", "size": 912384000}' curl -X PUT "UPLOAD_URL" -H "Content-Type: video/mp4" --data-binary @raw-tutorial.mp4 # 2. Cut filler words, retakes and long pausescurl -X POST https://api.vcp.name/v1/cuts \ -H "Authorization: Bearer $VCP_API_KEY" \ -H "Content-Type: application/json" \ -d '{"video_file_id": "FILE_ID", "pace": "tight"}'What developers build with it
Each cut is labelled as a filler word, a retake or a long pause. In the Desktop API, your code can keep or drop each cut before rendering.
- First-pass edits for every raw upload in a video team's pipeline
- Tutorial recordings tightened automatically before review
- Long recordings of up to 3 hours cut to one clean video
{ "id": "04bd0571-e2c8-4d5b-a3cc-e56f85773abf", "type": "cut", "status": "succeeded", "result": { "video_url": "https://files.example.com/cut.mp4", "duration_s": 412.6, "source_duration_s": 655.1, "removed": [ { "start": 12.4, "end": 14.9, "reasons": ["filler"] }, { "start": 31.0, "end": 58.2, "reasons": ["retake"] } ] }, "tokens_charged": 42, "finished_at": "2026-10-30T00:18:01Z"}Clean up your users' recordings with one API call.
Endpoints
AI Cut endpoints
| Method | Path | What it does |
|---|---|---|
| POST | /v1/uploads | Get an upload link for a file |
| POST | /v1/cuts | Cut a video of up to 3 hours |
| GET | /v1/jobs/{id} | Get the cut video and the list of cuts |
Desktop API
Run the API on your own computer
Every desktop plan includes a local API on Windows and Mac. It needs no key and answers requests from your computer while the app is open.
# 1. Create a project from a raw recordingcurl -X POST "http://127.0.0.1:8100/api/v1/aicut/projects" \ -H "Content-Type: application/json" \ -d '{ "name": "Tutorial take 3", "source_path": "/Users/you/Videos/raw-tutorial.mp4" }' # 2. Find silences, filler words and bad takescurl -X POST "http://127.0.0.1:8100/api/v1/aicut/projects/PROJECT_ID/analyze"curl "http://127.0.0.1:8100/api/v1/aicut/projects/PROJECT_ID/status" # 3. Read the labelled cut list, then POST it to /render for the MP4curl "http://127.0.0.1:8100/api/v1/aicut/projects/PROJECT_ID"Desktop API endpoints
| Method | Path | What it does |
|---|---|---|
| POST | /api/v1/aicut/projects | Create a project from a recording |
| POST | /api/v1/aicut/projects/{project_id}/analyze | Find the cuts |
| GET | /api/v1/aicut/projects/{project_id} | Read the project and its labelled cuts |
| POST | /api/v1/aicut/projects/{project_id}/render | Render the cut list to MP4 |
| POST | /api/v1/aicut/projects/{project_id}/overdub | Re-voice selected words |
Create a key and send your first request in a few minutes.
FAQ
Common Questions
Long silences, filler words such as um and uh, stutters, false starts and bad takes. When a line is repeated, it keeps the last good take.
In the Desktop API, yes. Analysis returns the cut list first, and you send the final list to /api/v1/aicut/projects/{project_id}/render. The Cloud API renders in one job and lists every removed part with its reason.
Only on desktop. The Desktop API uses your own LLM key: Gemini, OpenAI, OpenRouter, DeepSeek or any OpenAI-compatible endpoint, set in the app or with /api/v1/aicut/llm-config. The Cloud API needs no LLM key.
In the Desktop API, yes. POST /api/v1/aicut/projects/{project_id}/overdub re-voices the words you select in the speaker's own voice, so the take stays usable.
Explore
Related APIs
Text to speech
Turn text into speech with preset or cloned voices.
Voice cloning
Clone a voice from a short clip and speak new text with it.
Voice design
Describe a voice in words and generate speech in it.
Speech to text
Transcribe audio with word timestamps and speaker labels.
Dubbing
Dub videos into 21 languages, keeping each speaker's voice.
Subtitles
Generate SRT caption files with word-level timing.
Clips
Cut long videos into short, captioned clips.
Avatars
Make a lip-synced talking-head video from one photo.
Voice changer
Re-voice a recording while keeping its timing and emotion.
Audiobooks
Narrate a manuscript with a voice for every character.