Beatflow

The open-source, in-browser beat-cut video generator — no login, no upload, 100% local render.

开源、浏览器内运行的 AI 卡点视频生成器 —— 无需注册、无需上传、本地渲染导出。

🎬 Try it live / 立即体验 →  https://future-insight.github.io/beatflow/

---

Why Beatflow?

	Beatflow	CapCut / Premiere beat-sync	Generic AI video tools
Runs in browser	✅ zero install	❌ desktop app	⚠️ cloud-only
Your images leave?	❌ never	depends	✅ uploaded
Beat-accurate cutting	✅ frame-level	✅	⚠️ loose sync
Open source	✅ MIT	❌ proprietary	❌
Login required	❌	⚠️	✅

Built for: independent musicians sharing singles, AI-art creators turning Midjourney / 可灵 / 即梦 outputs into short-form video, and indie creators who don't want to upload their WIP to a SaaS.

---

English

Beatflow is an in-browser beat-cut video generator: upload a song + pick images, the app detects BPM and beat timestamps, cuts images on every beat, and exports a ready-to-share .webm / .mp4 video — WYSIWYG, no login, no backend required for rendering.

• 🎧 Automatic beat detection — beat mode (pop / dance, downbeats) / onset mode (ambient / classical, transients)
• 🖼️ Image cutting — drag-to-reorder thumbnails, fixed or random playback, switch per beat
• 🎬 Live preview — 9:16 / 1:1 / 16:9 aspects, WYSIWYG stage
• 💾 Local export — rendered and exported in-browser, nothing uploaded
• 🔒 Privacy-first — only beat analysis hits the API; images & exports never leave your device
• 🌗 Bilingual & theming — 中文 / English, dark / light, one-click toggle

Live demo

👉 https://future-insight.github.io/beatflow/

No signup, no install — just open and go.

Project layout

• frontend/ — static frontend, deploys to GitHub Pages / Cloudflare Pages / Vercel
• api/ — Flask beat-analysis service (Docker, deployed on Hugging Face Spaces)

Run locally

Beat-analysis API:

cd api
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
PORT=8088 python app.py

Health check: curl http://localhost:8088/api/health

Analyze endpoint:

curl -F "audio=@your.mp3" -F "method=beat" -F "min_interval=0.3" \
     http://localhost:8088/api/analyze

Frontend:

cd frontend
python3 -m http.server 5173

Then open http://localhost:5173/.

---

中文简介

它解决什么问题？

• 剪映 / PR 的卡点要手动踩点，累
• 云端 AI 视频工具要上传图片 + 注册 + 跑额度
• 开源工具大多是命令行、没前端、不能所见即所得

Beatflow 把这三件事合成一件： 浏览器打开 → 丢音乐 + 图 → 自动踩点 → 所见即所得 → 一键导出。全程图片不离开你的电脑。

适合谁用

• 🎵 独立音乐人：新歌出封面 / TikTok / 小红书引流视频
• 🎨 AI 绘画创作者：把 Midjourney / 可灵 / 即梦 的出图变成卡点短视频
• 📱 短视频博主：批量做 9:16 卡点，不想把素材上传到任何 SaaS
• 🛠 独立开发者：想要个能自托管、纯前端、可魔改的卡点方案

正文介绍

Beatflow 是一款浏览器端的自动节拍卡点视频生成器：上传一首音乐 + 一组图片，系统自动检测节拍 BPM 与强拍时间点，图片按节拍切换，所见即所得，直接导出 .webm / .mp4 卡点短视频。

• 🎧 自动节拍检测 —— beat 模式（流行/电子，强拍定位） / onset 模式（氛围/古典，能量突变）
• 🖼️ 图片卡点 —— 支持拖拽排序、固定 / 随机播放，图片在每个节拍点切换
• 🎬 实时预览 —— 9:16 / 1:1 / 16:9 三种比例，所见即所得
• 💾 本地导出 —— 视频在浏览器内渲染并导出，不上传不登录
• 🔒 隐私优先 —— 仅节拍分析走 API，图片与导出永不离开设备
• 🌗 双语 & 明暗主题 —— 中文 / English，深色 / 浅色一键切换

在线体验

👉 https://future-insight.github.io/beatflow/

无需注册、无需下载，浏览器打开即用。

项目组成

• frontend/ — 纯静态前端（可直接部署到 GitHub Pages / Cloudflare Pages / Vercel）
• api/ — Flask 节拍分析服务（Docker，部署在 Hugging Face Spaces）

本地运行

启动节拍分析 API：

cd api
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
PORT=8088 python app.py

健康检查：curl http://localhost:8088/api/health

启动前端：

cd frontend
python3 -m http.server 5173

浏览器打开 http://localhost:5173/ 即可。

---

Star History

If Beatflow saved you an afternoon of manual beat-tagging (踩点), a ⭐ would mean a lot — it's the only metric indie devs have.

---

Credits & 关键词

Beat detection powered by librosa · in-browser rendering via WebCodecs + Canvas · static frontend hosted on GitHub Pages.

_{Keywords: AI video generator, beat sync, 卡点视频, music video automation, browser video editor, open source CapCut alternative, Midjourney to video, Suno visualizer, 可灵 / 即梦 / 抖音 / 小红书 视频生成}