VerbaFlow - AI Terminology & Transcript Suite

中文文档 (Chinese Documentation)

VerbaFlow is a professional-grade, privacy-focused linguistic studio designed for audio/video content creators, translators, and proofreaders.

Unlike cloud-based SaaS platforms, VerbaFlow runs entirely in your browser. It leverages powerful LLMs (Google Gemini, OpenAI, Anthropic) to ensure content integrity while providing workflow-specific tools for terminology management, automated proofreading, and subtitle synchronization.

🌟 Key Features

1. 🤖 Multi-Provider AI Support

• Gemini (Native Multimodal): Best for direct video/audio analysis and large context windows.
• OpenAI / Anthropic: Support for ChatGPT, Claude models via standard protocols for high-precision text proofreading.
• Flexible Config: Switch models and providers on the fly based on your task requirements.

2. 🎧 Intelligent Studio

• Visual Proofreading: Review AI suggestions side-by-side with the original transcript.
• Context-Aware: The AI understands the context of the entire video/audio to make accurate corrections.
• Media Sync: Click any subtitle line or term to instantly jump to that timestamp in the video/audio player.

3. 💬 AI Assistant (Copilot)

• Drag-and-Drop Interface: A floating, resizable chat window that can be moved anywhere on screen.
• Contextual Help: Ask the assistant to check consistency, explain specific technical terms, or polish selected text segments.
• Session Management: Save and manage multiple conversation histories directly in the sidebar.

4. 📚 Knowledge Base (Glossary)

• Smart Extraction: Automatically extract proper nouns, technical terms, and entities from your content.
• Consistency Loop: Extracted terms are saved to Glossary Sets, which are then fed back into the AI for future analysis, ensuring strict terminology consistency across projects.
• Import/Export: Support for JSON, CSV, and text format migration.

5. 🔒 Privacy & Local-First

• IndexedDB Storage: All projects, transcripts, and chat history are stored in your browser's local database (Dexie.js).
• No Backend: We do not have a server. Your API Keys are saved in LocalStorage and sent directly to the LLM providers.
• Data Management: A dedicated dashboard to view storage usage and clear cache/history.

🚀 Getting Started

Prerequisites

• Node.js (v18 or higher)
• An API Key from one of the supported providers:
  • Google AI Studio (Gemini)
  • OpenAI Platform
  • Anthropic Console

Installation

1. Clone the repository

   git clone https://github.com/zleo-ai/VerbaFlow.git
   cd VerbaFlow

2. Install dependencies

   npm install

3. Run the development server

   npm start

4. Configure the App

   • Open http://localhost:1234
   • Click the Settings (Gear Icon) in the sidebar.
   • Select your provider (e.g., Gemini).
   • Paste your API Key.

🛠️ Tech Stack

• Frontend: React 19, TypeScript, Tailwind CSS
• State/Storage: React Context API, Dexie.js (IndexedDB wrapper)
• AI Integration:
  • @google/genai SDK for Gemini
  • Native fetch adapters for OpenAI/Anthropic
• UI Components: Lucide React icons, Custom Modal/Toast system

📖 Documentation

• Architecture Overview - Understanding the system design.
• Contributing Guide - How to add new providers or features.

License

MIT © [Your Name]