Self-hosted AI transcription & translation: Whisper + DeepL. Multi-arch, CPU-only, 8011.
9.5K
Maintained by: lkmeta
Where to get help:
Where to file issues: GitHub Issues
| Tag | Architectures | Notes |
|---|---|---|
latest | linux/amd64, linux/arm64 | Rolling build from main. Use this. |
v1.1 | linux/amd64, linux/arm64 | Snapshot of the current build, for pinning. |
v1 | linux/amd64 | Legacy (April 2025). Kept so existing setups keep working; no longer updated. |
Txtify is a free, open-source web application for transcribing and translating audio from YouTube videos or uploaded media files. It uses Whisper (via stable-ts), optional DeepL translation, and FastAPI, containerized with Docker.
.txt, .srt, .vtt, .sbv, .pdf (non-Latin scripts included).Pull the image:
docker pull lkmeta/txtify:latest
Run the container:
docker run -d --name txtify -p 8011:8011 lkmeta/txtify:latest
Open http://localhost:8011.
| Env var | Default | What it does |
|---|---|---|
DEEPL_API_KEY | — | Enables translation. |
MAX_UPLOAD_SIZE_MB | 1000 | Reject uploads above this size. |
MAX_VIDEO_DURATION | 900 | Reject YouTube videos longer than this (seconds). |
MAX_CONCURRENT_JOBS | 2 | Transcriptions running at once; each is a full Whisper process. |
Bigger models need more memory: medium wants roughly 5 GB free, large roughly 10 GB. If a job fails, the UI says why and the full log is in output/<job id>_logs.txt inside the container.
Security note: Txtify has no built-in authentication. Do not expose the port to the internet without a reverse proxy with auth in front.
docker logs -f txtify
Demo video, detailed instructions, and the roadmap live in the GitHub repository: lkmeta/txtify.
This project is licensed under the Apache 2.0 License.
As with all pre-built Docker images, it is the user's responsibility to ensure compliance with all relevant licenses for the software and dependencies included in this image.
Content type
Image
Digest
sha256:1acfd4117…
Size
844.6 MB
Last updated
3 months ago
docker pull lkmeta/txtify