Subsync 〰️▶

Play your subtitles into another app. A tiny, always-on-top, dependency-free desktop tool that reads a song's .srt and drives a synced-lyric display (YouTube Music, etc.) by firing keystrokes in time with the audio — so a lagging lyric track can be nudged back into sync by hand.

It's a single Python file with zero dependencies beyond the standard library. No pip install, no build step. Windows only (uses the Win32 SendInput API).

Sibling to Subtap: Subtap edits subtitle timings by ear; Subsync plays that timing back into whatever app has keyboard focus. Same family, same colors.

Why

A synced-lyric display (like the one DistroKid publishes to YouTube Music) often lags the audio by a beat, and there's no knob to fix it on the listener's end. So you drive it yourself: hold a key while a line is showing, release it in the gaps, and the display snaps back into time. Subsync does exactly that on autopilot — it reads the .srt you already have and plays that timeline out as keystrokes, with a lead you can dial in to beat the lag.

What it does

1. Load an .srt for the song.
2. Pick a mode:
   • Hold/release space — holds the spacebar for each lyric line's duration and releases it in the gaps, following the .srt timeline.
   • Type lyrics — types each line (paced ~25 ms/char) at its start time; handy for testing against Notepad before you point it at the real player.
3. Set the lead (ms) — keys fire this many ms before the .srt time to beat display lag. Default 500 ms; type a value or nudge in 10 ms steps (−/+ or ↑/↓).
4. Hit GO — a 3 s count-in ticks down (screen red) at 0.01 s resolution. At zero the screen turns green ("CLICK PLAY NOW") — click Play on the song. That instant is t = 0, and the keystrokes follow the .srt from there.

Keys land on whatever window has keyboard focus, so after GO, click into the song/player window — the keys go there, not to Subsync.

 
The window stays on your dark theme; a rounded outline lights up only while it's running — rose during the 3 s count-in, teal while it plays (top), and nothing at rest (right).

Requirements

• Python 3.8+ (Windows — uses the Win32 SendInput API)
• No browser, no packages. It's a native Tk window on your global Python.

Usage

No Python? Download Subsync.exe from the latest release and double-click it — it's a self-contained build, nothing to install.

Otherwise, on your own Python 3.8+:

python subsync.py                       # then Load SRT from the UI
python subsync.py "path/to/song.srt"    # pre-load an .srt

Windows: double-click Subsync.cmd (it launches windowless), or drag an .srt onto it to pre-load that song.

Keys

Esc stop + release · ↑ / ↓ nudge lead ±10 ms

The lead, and why it's there

The whole point is fighting display lag, and lag isn't constant across players or machines — so the lead is front and center. A lead of 500 means every keystroke fires 500 ms early; a negative lead fires late. Nudge it while you watch a line land, and it's tuned in a few passes. The box accepts 500, -200, even 500 ms, and clamps to a sane −2 … +5 s.

How it works

Keystrokes go out through the Win32 SendInput API via ctypes — scancode-based space (most compatible with games and players), Unicode typing, and Enter. A single perf_counter clock projects the "click Play" instant, runs a red count-in down to zero, then walks a pre-built schedule of key events, firing each as it comes due. Typing mode drains through a self-throttling pump (one char every 25 ms) so the target app's input queue never floods. That's the whole thing, in one file.

Curious? Everything lives in subsync.py — the SendInput plumbing up top, then the .srt parser, then the App class with the clock and the UI. Run python subsync.py --demo to see the window posed with sample content (add run for the green playing state).

Known limitation

Some .srts pad a caption's end time out to the next line's start, so a line "shows" straight through a long instrumental. Subsync faithfully holds space that whole time — the pause isn't a gap in the file, it's baked into the timing. That's an .srt data issue, not an app bug: fix it in Subtap by trimming the line's end to where the vocal actually stops.

Layout

Subsync/
  subsync.py           the whole thing — key injection + srt parser + UI, one file, stdlib only
  Subsync.cmd          double-click to run it windowless (Windows); drag an .srt on to pre-load
  subsync.ico          window + taskbar icon
  make_icon.py         regenerates subsync.ico (pure stdlib, no PIL)
  release.ps1          one-shot: bump version, tag, cut a GitHub release with subsync.py attached
  docs/                hero images (posed --demo renders)
  README.md  LICENSE  .gitignore

Contributing

This is a personal tool, published as-is — issues and pull requests aren't accepted (PRs auto-close). Fork it and make it your own. 〰️

License

MIT — see LICENSE. © 2026 RelentlessOldMan.