Convert Presentation Graphic Stream (SUP File) to text-based SRT format with Tesseract.
Download a package from the releases page. Each one bundles Tesseract, Leptonica,
the English OCR model and the pgs2srt command-line tool, so nothing else has to be installed. Other OCR languages can be
downloaded from inside the desktop app.
| Platform | File | Notes |
|---|---|---|
| Windows 10/11 x64 | pgs2srt-<version>-windows-x64-setup.exe or …-portable.zip |
The installer needs no administrator rights. The packages are not code-signed: on the SmartScreen warning choose More info → Run anyway. |
| macOS 15+ (Apple Silicon) | pgs2srt-<version>-macos-arm64.dmg |
Drag pgs2srt.app to Applications. The app is not notarized: on first launch open System Settings → Privacy & Security and choose Open Anyway, or run xattr -cr /Applications/pgs2srt.app. The CLI is /Applications/pgs2srt.app/Contents/MacOS/pgs2srt. |
| Linux x86_64 (glibc 2.35+: Ubuntu 22.04, Debian 12 or newer) | pgs2srt-<version>-linux-x86_64.AppImage |
chmod +x the file and run it. The CLI is in pgs2srt-cli-<version>-linux-x86_64.tar.gz (bin/pgs2srt). |
License notices of the bundled software are in the Windows install folder, in pgs2srt.app/Contents/Resources on macOS,
and under usr/share/doc in the AppImage (share/doc in the CLI archive).
The desktop app also runs in the browser at mbolaric.github.io/pgs2srt, with OCR by
tesseract.js. Subtitle files never leave your computer: they are read and converted in
the browser. Finished SRTs wait in the queue until you download them, one at a time or all together with Download All as one
.zip file. The first visit loads about 10 MB, and each OCR language about 4 MB more. The browser caches both. Conversion is slower
than in the desktop app, and folders, output folders and overwrite options are not available.
To host it yourself, unzip pgs2srt-<version>-web.zip from the releases page
into a folder of any static web server and open it over HTTP(S) (not as a local file). It works under any path. OCR languages
are still downloaded from GitHub on first use.
sudo apt install libleptonica-dev libtesseract-devsudo apt install tesseract-ocr-eng
sudo apt install tesseract-ocr-hrv
# or other languages: tesseract-ocr-<lang>cargo build --releaseThe compiled binary will be at target/release/pgs2srt.
./target/release/pgs2srt -p <input.sup> [options]-p, --pgs-file-name <FILE>: Path to input PGS/SUP file (required).-l, --language <LANG>: Tesseract language code (default:eng), orautoto detect it from the file name tag and, failing that, from the subtitle images.-s, --srt-file-name <FILE>: Output.srtfile path (optional, defaults to<name>.<lang>.srtnext to the input file).-f, --force: Overwrite the output file if it already exists (by default an existing file is never replaced).
Convert using default English OCR (creates subtitles.eng.srt next to subtitles.sup):
./target/release/pgs2srt -p subtitles.supConvert with specific language and output path:
./target/release/pgs2srt -p subtitles.sup -l hrv -s subtitles.srtDetect the language automatically and replace an existing output:
./target/release/pgs2srt -p subtitles.sup -l auto -fErrors are written to a log file (rolled at 5 MB, 3 old files kept):
- Linux:
~/.local/state/pgs2srt/logs/(or$XDG_STATE_HOME/pgs2srt/logs/) - macOS:
~/Library/Logs/pgs2srt/ - Windows:
%LOCALAPPDATA%\pgs2srt\logs\
The desktop app renders with wgpu (Metal / DX12 / Vulkan) and falls back to OpenGL (glow) if wgpu cannot start.
Set PGS2SRT_RENDERER=glow to force OpenGL.
On Linux Wayland desktops the app runs through XWayland, because winit (the windowing library) has no file drag and drop on native Wayland yet.
Set PGS2SRT_BACKEND=wayland to run as a native Wayland client instead (dragging files onto the window will not work; use Add files / Add folder).