TalkEdit/README.md

# TalkEdit

**Edit video by editing text.** An offline, local-first desktop video editor where deleting a word from the transcript cuts it from the video.

<img width="1034" height="661" alt="TalkEdit screenshot" src="https://github.com/user-attachments/assets/b1ed9505-792e-42ca-bb73-85458d0f02a5" />

---

## Features

- **Text-based editing** — delete, reorder, or correct words in the transcript to edit the underlying video. No razor tool, no timeline slicing.
- **Word-level transcription** — Whisper.cpp with per-word timestamps and confidence scores. Low-confidence words get a visual warning.
- **Four zone types** — Cut, Mute, Sound Gain, and Speed Adjust. Create zones on the waveform timeline and drag edges to refine.
- **Waveform timeline** — zoomable, scrollable waveform with playhead scrubbing, zone visualization, markers, chapters, and thumbnail strips.
- **AI-powered editing**
  - Filler word detection and removal
  - Smart Clean: one-click filler removal + silence trim + noise reduction + loudness normalization
  - Clip suggestions for social media shorts
  - Sentence rephrase with AI alternatives
  - Supports **Ollama** (local), **OpenAI**, and **Claude** backends
- **Background music** — import a second audio track with auto-ducking via sidechain compression.
- **Export** — fast stream-copy or full re-encode to MP4, MOV, WebM, or WAV. Resolution up to 4K.
- **Captions** — generate SRT, VTT, or burn-in ASS subtitles with configurable font, color, and position.
- **Speaker diarization** — identify and label multiple speakers.
- **Audio tools** — noise reduction (DeepFilterNet), loudness normalization (LUFS targeting), background removal (MediaPipe), batch silence removal, video zoom/punch-in.
- **Project save/load** — `.aive` JSON format preserves all edits, zones, markers, and AI config.
- **Customizable hotkeys** — two presets (Standard / Left-hand) with per-key remapping and conflict detection.
- **100% offline, no account required** — everything runs on your machine. No telemetry, no cloud dependency.
- **7-day free trial** with one-time license key purchase. No subscription.

---

## Tech Stack

| Layer | Technology |
|-------|------------|
| Desktop shell | **Tauri 2.0** (Rust) |
| Frontend | **React** + **TypeScript** + **Tailwind CSS** |
| State management | **Zustand** with Zundo undo/redo |
| Transcription | **Whisper.cpp** (word-level timestamps) |
| AI / LLM | **Ollama**, **OpenAI**, **Claude** (plugable backends) |
| Media processing | **FFmpeg** |
| Python services | **FastAPI** (spawned as a child process) |

---

## Quick Start

### Prerequisites

- **Node.js** 18+
- **Python** 3.10+
- **FFmpeg** (in PATH)
- **Rust** toolchain (for Tauri)
- **Ollama** (optional, for local AI features)

### Install

```bash
# Root and frontend dependencies
npm install
cd frontend && npm install && cd ..

# Backend dependencies
cd backend && pip install -r requirements.txt && cd ..
```

### Run (Development)

```bash
# Start everything: backend + frontend + Tauri
npm run dev:tauri
```

Or run components separately:

```bash
# Terminal 1: Python backend
npm run dev:backend

# Terminal 2: Frontend + Tauri
cd frontend && cargo tauri dev
```

### Build

```bash
npm run build:tauri
```

---

## Project Structure

```
talkedit/
├── src-tauri/             # Tauri 2.0 Rust runtime
│   ├── Cargo.toml
│   └── src/
│       ├── main.rs             # App entry, backend spawner
│       ├── lib.rs              # Command handlers (IPC bridge)
│       ├── transcription.rs    # Whisper.cpp integration
│       ├── video_editor.rs     # FFmpeg-based editing
│       ├── caption_generator.rs
│       ├── diarization.rs
│       ├── ai_provider.rs      # Ollama / OpenAI / Claude
│       ├── audio_cleaner.rs
│       ├── background_removal.rs
│       ├── licensing.rs        # Trial + key activation
│       ├── models.rs           # Shared data types
│       └── paths.rs
├── frontend/              # React + Vite + Tailwind
│   └── src/
│       ├── components/         # UI components
│       │   ├── TranscriptEditor.tsx
│       │   ├── WaveformTimeline.tsx
│       │   ├── VideoPlayer.tsx
│       │   ├── AIPanel.tsx
│       │   ├── ExportDialog.tsx
│       │   ├── SettingsPanel.tsx
│       │   ├── BackgroundMusicPanel.tsx
│       │   ├── MarkersPanel.tsx
│       │   ├── ZoneEditor.tsx
│       │   ├── SilenceTrimmerPanel.tsx
│       │   ├── AppendClipPanel.tsx
│       │   ├── LicenseDialog.tsx
│       │   └── DevPanel.tsx
│       ├── store/             # Zustand state (editorStore, aiStore, settingsStore)
│       ├── hooks/             # Custom React hooks
│       ├── lib/               # Utilities and Tauri bridge
│       └── types/             # TypeScript interfaces
├── backend/               # FastAPI Python services
│   ├── main.py
│   ├── routers/               # API endpoints
│   │   ├── transcribe.py
│   │   ├── ai.py
│   │   ├── audio.py
│   │   ├── captions.py
│   │   └── export.py
│   ├── services/              # Core logic
│   ├── video_editor.py
│   ├── caption_generator.py
│   ├── ai_provider.py
│   ├── diarization.py
│   ├── audio_cleaner.py
│   ├── background_removal.py
│   └── license_server.py
├── shared/                # Schema definitions (project format)
├── models/                # Whisper model storage
└── docs/                  # Documentation
```

---

## Keyboard Shortcuts

| Key | Action |
|-----|--------|
| Space | Play / Pause |
| J / K / L | Reverse / Pause / Forward |
| I / O | Mark In / Mark Out |
| ← / → | Seek ±5 seconds |
| Delete | Delete selected words or zones |
| Ctrl+Z | Undo |
| Ctrl+Shift+Z | Redo |
| Ctrl+S | Save project |
| Ctrl+E | Export |
| Ctrl+F | Search transcript |
| Ctrl+Scroll | Zoom waveform |
| ? | Shortcut cheatsheet |

---

## License

Source code is MIT — see [LICENSE](LICENSE) for details. The distributed binary includes a 7-day free trial requiring a one-time license key purchase for continued use.