This project provides a cross-platform AI client for converting video content into text-based blog posts. It targets users who want to repurpose video content from platforms like YouTube, Bilibili, X, and Toutiao into written articles, leveraging AI for transcription, translation, and summarization.

How It Works

The core workflow involves using yt-dlp to download videos and their associated subtitles. If subtitles are missing or not in Chinese, Whisper is used to generate and translate them. Gemini then processes these subtitles to create a structured article, with placeholders for manually inserted images extracted from the video. The application is built using Electron and Vite for the frontend, with Python scripts handling the AI processing.

Quick Start & Requirements

• Install: npm i
• Run (Windows): npm run start-win
• Run (macOS): npm run start-mac
• Prerequisites: Node.js (v20.11.0+), npm (v10.2.4+), Python (v3.11.2+), pip (v24.0+). Requires pyinstaller for packaging. Specific Whisper models need to be downloaded. yt-dlp may require macOS security exceptions.
• Setup: Local setup involves installing Node.js and Python dependencies. Compiling involves GitHub Actions triggered by Git tags.

Highlighted Details

• Supports video downloading and transcription from YouTube, Bilibili, X, and Toutiao.
• Integrates Whisper for speech-to-text and translation.
• Utilizes Gemini for converting transcripts into articles.
• Cross-platform support for Windows, macOS, and Linux.
• Includes multiple prompt versions for AI article generation, from basic formatting to Quill Delta JSON and Markdown output.

Maintenance & Community

The project is actively developed by aehyok. Community interaction and updates are primarily shared via a WeChat group and official account. Links to example projects and technical documentation for Electron, Vite, FFmpeg, and Python packaging are provided.

Licensing & Compatibility

The repository does not explicitly state a license in the README. This lack of clear licensing information may pose compatibility issues for commercial use or integration into closed-source projects.

Limitations & Caveats

Toutiao downloads require a hardcoded cookie. The project mentions potential Chinese garbling issues in Windows command prompts, addressed by chcp 65001. Large model files need manual copying to specific directories. There are notes on resolving node-gyp errors and sqlite3 version compatibility issues (sqlite3@5.1.6). macOS users may need to grant explicit permissions for yt-dlp.