Upload a 1-hour livestream, then just type: "Merge all his speaking clips", "Cut out Mr. Wang's scenes", "Clip the product intro", or "Make a 60s highlight" — ClipTalk understands your footage, plans the edit, calls the right tools, and delivers the finished clips.
- [2026-XX-XX] 🎉 ClipTalk is now open-source!
- [2026-XX-XX] 🚀 Released Voice ID — target any speaker by name across hours of footage.
- [2026-XX-XX] ✨ Added one-command highlight generation for livestreams.
Real editing tasks completed by ClipTalk with a single instruction.
📺 Livestream Highlights |
🎤 Interview Cuts |
📦 Product Demos |
🏢 Meeting Recordings |
🎓 Course Lectures |
➕ More coming |
ClipTalk is an AI video-editing agent. You don't drag clips on a timeline or scrub through hours of footage — you just describe what you want in natural language, and the agent handles the entire process: understanding the footage → locating the target content → planning the edit → executing cuts → delivering the final clips.
It's not a keyword search over subtitles. ClipTalk actually understands who is on screen, who is speaking, and what they are talking about — combining speech recognition, voiceprint identification, face detection, and topic segmentation — so instructions like "cut out Mr. Wang's scenes" just work, even in a multi-speaker 1-hour recording.
Describe the task, get the result. No timeline skills required — "merge all his speaking clips" is a complete editing workflow in one sentence. Follow up with refinements: "make it shorter", "start from the part about pricing".
ClipTalk builds a structured understanding of your video before editing:
- Speech transcription — full subtitles with timestamps
- Voiceprint ID — knows who is speaking, not just that someone is
- Face detection & tracking — locates every person's on-screen segments
- Topic segmentation — finds where the product intro starts and ends
Name a person, get their clips. "Cut out Mr. Wang's scenes" combines face and voice identification to extract every segment where the target person appears or speaks — across the entire footage.
Automatically detect and merge all segments where a specific speaker is talking, removing silence, filler, and other speakers' turns — turning scattered moments into one continuous clip.
"Make a 60s highlight" — the agent scores the footage for energy, key statements, and audience reactions, then assembles a tight highlight reel with subtitles, ready to publish.
Behind every instruction, the agent plans and executes a tool pipeline — reading subtitles, running voice ID, locating segments, cutting clips — and shows you its working log in real time, step by step, with nothing hidden.
Results don't disappear into a black box. Every edit lands on a multi-track timeline (video / audio / subtitles) where you can inspect the cuts, nudge boundaries, and re-render — AI does the heavy lifting, you keep the final say.
Not happy with a result? Just keep talking. Each instruction builds on the previous state — "remove the second clip", "add subtitles", "export vertical for mobile" — no need to start over.
Designed for the hard cases: 1-hour livestreams, multi-speaker panels, full-day event recordings. The longer and messier the footage, the more time ClipTalk saves.
Transcription, voiceprint, face detection, and rendering are pluggable tools behind a unified agent interface — swap in your preferred models or extend the toolbox with your own.
| Tool | Capability |
|---|---|
| 📝 Transcriber | Speech-to-text with word-level timestamps |
| 🔊 Voice ID | Identifies and tracks individual speakers by voiceprint |
| 👤 Face Tracker | Detects and tracks every person's on-screen presence |
| 🧭 Topic Segmenter | Splits footage into semantic sections and topics |
| ⭐ Highlight Scorer | Ranks moments by energy, key statements, and reactions |
| ✂️ Cutter | Executes precise, frame-accurate cuts and merges |
| 🎞️ Compositor | Assembles clips, subtitles, and audio into the final render |
💬 "Cut out Mr. Wang's scenes"
↓
🤖 Agent plans the edit
↓
📝 Read subtitles → 🔊 Voice ID → 👤 Face tracking → 🧭 Locate segments
↓
✂️ Cut clips → 🎞️ Compose → ✅ Done
One instruction in, finished clips out — every step visible in the agent's working log.
git clone https://github.com/yourorg/ClipTalk.git
cd ClipTalk
# 安装与启动步骤，待补充See the Documentation for configuration and API keys.
🤝 Contributing We welcome contributions of all kinds! Please see our Contributing Guide to get started.
📄 License This project is licensed under the GNU General Public License v3.0.
You are free to run, study, share, and modify this software. Any distributed derivative work must also be released under the GPL v3, keeping the software free for all users. See the LICENSE file for the full text.
💬 Community Discord · Twitter/X · WeChat Group
⭐ If you find ClipTalk useful, please give us a star!
Made with ❤️ by the ClipTalk Team