- Home
- AI Video Caption Generators
- SmartSub
SmartSub
What it is
SmartSub is a free, open-source desktop application for transcription, subtitle translation, proofreading, AI dubbing, and video subtitle rendering. It runs on Windows, macOS, and Linux, with builds for Windows x64, macOS arm64 and x64, and Linux x64. Local models process files on the computer, so files do not leave the device during local processing. Transcription engines include whisper.cpp, faster-whisper, FunASR, Qwen3-ASR, and FireRedASR. Translation offers 20 services, including built-in free translation and connections such as DeepSeek, Gemini, and Ollama. The proofreading workspace supports video-synced editing, undo and redo, search and replace, and AI proofreading. SmartSub can synthesize speech, align dubbed audio to a timeline, and clone voices with ZipVoice, Volcengine, or ElevenLabs. Users can burn subtitles into video or package soft subtitles, preview styles, and add dubbed audio. Video downloads use yt-dlp and lux for links from more than 1,800 sites. MCP and CLI automation provide 111 tools and commands. Cloud transcription uploads audio to the chosen provider. The project uses the MIT license.
For and against
For
- Free local transcription, translation, TTS, and rendering
- Runs on Windows, macOS, and Linux
- Multiple selectable transcription engines
- Subtitle editing includes undo, redo, and search
- Supports burn-in and soft subtitle output
Against
- Cloud transcription uploads audio to a provider
- Optional cloud services may be paid
- GPU acceleration depends on listed hardware paths
- Voice cloning uses specified external engines
The verdict
SmartSub brings transcription, translation, proofreading, dubbing, and subtitle rendering into one open-source desktop application. Its local processing and broad operating-system support are useful, while cloud workflows require attention to uploaded audio and optional service costs.
Who it suits
It suits creators who need a desktop workflow spanning transcription, translation, subtitle editing, dubbing, and rendering. Local processing is relevant to users who want files to remain on their computer during local tasks.
Plans and prices
Free
local transcription, translation, TTS and FFmpeg workflow · no API key or usage limit for local steps · optional paid cloud providers
Read 15 Oct 2026 · smartsub.linxiaodong.com
On the record
- AI assistant
- The in-app copilot provides workspace context, conversational editing, screenshot diagnostics, file chat and tool execution.github.com · read 15 Oct 2026
- AI-client integrations
- The settings workflow can connect Cursor, OpenAI Codex and Claude Code through generated MCP configuration.github.com · read 15 Oct 2026
- Automation
- The repository README describes MCP and CLI automation with 111 tools and commands for media processing workflows.github.com · read 11 Oct 2026
- Brand caption presets
- Yesgithub.com · read 19 Sep 2026
- Burn-in captions
- Yesgithub.com · read 19 Sep 2026
- Caption styling
- Yesgithub.com · read 19 Sep 2026
- Cloud transcription privacy
- For cloud transcription, audio is uploaded to the configured provider endpoint, and the README advises against using it for sensitive content.github.com · read 11 Oct 2026
- Cloud voice services
- Cloud dubbing options include Edge TTS, OpenAI-compatible endpoints, Azure Speech, Volcengine, ElevenLabs and Xiaomi MiMo.github.com · read 15 Oct 2026
- Core workflow
- Each stage can be run separately or chained into batch pipelines.smartsub.linxiaodong.com · read 15 Oct 2026
- Developer build
- Contributors clone the repository, run yarn install and yarn dev, with yarn native:fetch available when native downloads fail.github.com · read 15 Oct 2026
- Documentation and support
- The project publishes a Docusaurus documentation site with usage guides and FAQ pages and directs support to GitHub issues and community chat.smartsub.linxiaodong.com · read 15 Oct 2026
- Dubbing
- Local Kokoro and VITS engines provide 103 and 174 voices respectively, while ZipVoice supports zero-shot cloning from a reference clip.github.com · read 15 Oct 2026
- GPU acceleration
- The download guide lists CUDA for NVIDIA and Vulkan for AMD or Intel on Windows and Linux, and Core ML or Metal on Apple-silicon Macs.github.com · read 11 Oct 2026
- Hardware acceleration
- Acceleration paths include NVIDIA CUDA, AMD/Intel Vulkan and Apple Core ML/Metal, with CPU fallback when acceleration fails.github.com · read 15 Oct 2026
- Installation
- macOS users can install through Homebrew using the buxuku tap and smartsub cask.smartsub.linxiaodong.com · read 15 Oct 2026
- Intel Mac limitation
- The Intel macOS build runs CPU-only and does not support GPU acceleration.smartsub.linxiaodong.com · read 15 Oct 2026
- Licence
- The repository is released under the MIT licence, with copyright attributed to Lin Xiaodong in 2024.github.com · read 15 Oct 2026
- License and support
- The project uses the MIT license and directs users to documentation, FAQs, community groups, GitHub issues, and pull requests for help or participation.github.com · read 11 Oct 2026
