Building Li Translate: An Open-Source AI Subtitle Generator That Actually Understands Context
Subtitle generation has improved dramatically over the last few years, but most existing tools still suffer from the same problems.
They either require expensive cloud APIs, produce literal translations that ignore context, or force users into workflows that are difficult to automate.
I started Li Translate to solve those issues.
The goal is simple:
Generate high-quality subtitles with natural AI translation while remaining open, extensible, and developer-friendly.
Why another subtitle tool?
Most subtitle generators focus on only one part of the pipeline.
Some are excellent at speech recognition but weak at translation.
Others translate well but don’t produce synchronized subtitles.
Many depend entirely on proprietary cloud services, making them expensive or impossible to self-host.
Li Translate is designed around a complete workflow instead of a single feature.
The vision is to make subtitle generation feel like a modern AI pipeline rather than a collection of disconnected tools.
The idea
A typical workflow looks like this:
- Detect spoken language
- Generate accurate timestamps
- Produce subtitle segments
- Translate them naturally using AI
- Preserve timing and formatting
- Export standard subtitle formats
Instead of translating each sentence independently, the system is designed to understand surrounding context, leading to translations that sound much more natural.
Why open source?
Open source tools move faster.
Developers can inspect the pipeline, improve models, fix bugs, and integrate the project into their own products.
Whether you’re building a media platform, an education product, a localization service, or an automation workflow, you shouldn’t have to reinvent subtitle generation.
That’s exactly why Li Translate is open.
Designed for developers
The project aims to be:
- Modular
- API-friendly
- Easy to self-host
- Extensible
- Automation-ready
The long-term goal is to make every part of the subtitle pipeline replaceable.
Want a different speech recognition model?
Swap it.
Want another translation model?
Plug it in.
Need a custom export format?
Add it.
Roadmap
Some of the features planned for future releases include:
- Better multilingual support
- Faster subtitle generation
- Speaker-aware subtitle generation
- Batch processing
- REST API
- Docker deployment
- GPU and CPU optimization
- Additional export formats
- Improved AI translation quality
Who is this for?
Li Translate is intended for:
- Content creators
- Video editors
- Localization teams
- AI researchers
- Developers
- Open-source contributors
If your work involves subtitles or multilingual media, this project is built for you.
Looking for contributors
This project is still evolving, and there is plenty of room for new ideas.
Whether you’re interested in AI, speech recognition, translation, backend development, frontend development, or DevOps, contributions are welcome.
If you’d like to help shape the future of open-source subtitle generation, I’d love to hear your ideas.
⭐ If you find the project interesting, consider giving it a star.
Repository:
https://github.com/AliAkrami1375/Li-Translate
Top comments (1)
Thanks for reading!
I’d really appreciate any feedback, feature requests, or ideas for improving Li Translate. Every suggestion helps shape the roadmap.