SubLens - 视频字幕提取工具，从视频中提取硬编码字幕，支持多种格式输出。基于 Tauri + Vue 3 + Rust 构建。
Every subtitle maps to exact video frames. Timeline thumbnail preview shows actual video frames on hover.
- j/k keys for quick subtitle jumping with toast preview
- Confidence filter (All / High / Medium / Low)
- Search within subtitle text
- Virtual scrolling for smooth performance with 1000+ subtitles
- OKLCH design system — perceptually uniform colors, consistent visual experience
- Dark/light theme — professional video editing aesthetics
- Tab-based interface — Files / Progress / ROI / OCR / Export / Settings
- Micro-interactions — hover effects, transitions, toast notifications
| Engine | Technology | Accuracy | Speed | Languages |
|---|---|---|---|---|
| PaddleOCR | PP-OCRv5 Deep Learning | Excellent | Fast | 80+ |
| EasyOCR | PyTorch | Good | Medium | 80+ |
| Tesseract.js | LSTM + WASM | Good | Fastest | 100+ |
- Multi-pass OCR — recognize multiple times, take the best result
- Text normalization — full-width to half-width punctuation, Chinese typo correction
- Confidence calibration — mixed language / short text / repeated chars auto-degraded
- Subtitle merge — Levenshtein similarity auto-deduplication
- Scene detection — histogram + chi-square test for intelligent frame sampling
| Format | Frame-Mapped | Best For |
|---|---|---|
| SRT | No | Universal subtitle players |
| WebVTT | No | Web video |
| ASS | No | Anime fansub, advanced styling |
| SSA | No | Legacy subtitle format |
| JSON | Yes | Frame-accurate editing |
| CSV | Yes | Spreadsheet analysis |
| TXT | No | Plain text |
| LRC | No | Lyrics sync |
| SBV | No | YouTube subtitles |
| MD | No | Markdown documentation |
| STL | No | Spruce subtitle format |
| TTML | No | Timed Text ML |
Bottom · Top · Left · Right · Center · Custom — one-click selection
MP4 · MKV · AVI · MOV · WebM
# Clone the repo
git clone https://github.com/Agions/SubLens.git
cd SubLens
# Install frontend dependencies
pnpm install
# Run in development mode
pnpm tauri dev
# Build production package
pnpm tauri build# Basic extraction
hardsubx-cli extract video.mp4 --output ./subs
# Multi-format output
hardsubx-cli extract video.mp4 --format srt,vtt,json --output ./subs
# Specify ROI region + OCR engine
hardsubx-cli extract video.mp4 --roi bottom --ocr paddle --lang ch,en
# Custom confidence threshold
hardsubx-cli extract video.mp4 --confidence 80
# Preview a specific frame
hardsubx-cli preview video.mp4 --frame 1500
# Show video metadata
hardsubx-cli info video.mp4
# Display help
hardsubx-cli --help| Layer | Technology |
|---|---|
| Desktop Framework | Tauri 2.x |
| Frontend | Vue 3 + TypeScript |
| Backend | Rust |
| OCR Engines | Tesseract.js (WASM), PaddleOCR (Native), EasyOCR |
| State Management | Pinia |
| Build Tool | Vite |
SubLens/
├── src/ # Vue 3 frontend
│ ├── components/ # Vue components
│ │ ├── common/ # Button, Modal, Tooltip, SubtitleToast
│ │ ├── layout/ # ToolBar, SidePanel, VideoPreview
│ │ │ ├── tabs/ # Tab components (Files/Progress/ROI/OCR/Export/Settings)
│ │ │ ├── BatchProcessView.vue
│ │ │ └── StatusBar.vue
│ │ ├── video/ # ROISelector, Timeline (with thumbnail preview)
│ │ └── subtitle/ # SubtitleList, ExportDialog
│ ├── composables/ # 17 composables (logic/UI separation)
│ │ ├── useSubtitleList.ts # Subtitle filtering, search, pagination
│ │ ├── useVideoPlayer.ts # Video playback, captureFrame
│ │ ├── useOCREngine.ts # OCR engine abstraction + post-processing
│ │ ├── useSubtitleExtractor.ts
│ │ ├── useBatchProcessor.ts
│ │ └── use*.ts # Tab-specific composables
│ ├── stores/ # Pinia stores
│ │ ├── subtitle.ts # Subtitle list + export formats
│ │ ├── project.ts # Project file state + video metadata
│ │ └── settings.ts # Theme, language, OCR preferences
│ ├── core/ # Business logic
│ │ ├── SubtitlePipeline.ts # 4-stage OCR post-processing
│ │ ├── SubtitleExporter.ts # 12 format writers
│ │ ├── SceneDetector.ts # Histogram + chi-square scene detection
│ │ └── ConfidenceCalibrator.ts
│ └── types/ # TypeScript type definitions
│
├── src-tauri/ # Rust backend
│ └── src/
│ ├── commands/ # Tauri IPC commands
│ │ ├── video.rs # Frame extraction, metadata, ffmpeg
│ │ ├── ocr_engine.rs # PaddleOCR Python bridge
│ │ ├── ocr.rs # EasyOCR / Tesseract.js
│ │ ├── scene.rs # Scene detection
│ │ ├── export.rs # Format writers
│ │ ├── file.rs # File dialogs
│ │ └── system.rs # System diagnostics
│ └── main.rs # Tauri app entry
│
├── docs/ # Documentation
│ ├── index.md # Documentation hub
│ ├── getting-started.md # Installation and first extraction
│ ├── architecture.md # Project structure and design
│ └── cli.md # CLI reference
│
└── cli/ # Node.js CLI tool
└── src/
├── extract.ts
├── formats.ts
└── index.ts