Transpose PDF chord charts and lead sheets to a new key — right in your browser, in seconds.
SheetShift reads the chords in a PDF, lets you pick a new key (or shift by an interval), and hands back a transposed PDF that looks exactly like the original — same layout, lyrics, spacing, and fonts, just with the chords changed. Everything runs on your device; your files are never uploaded.
- Keeps your chart intact. It doesn't rebuild the page. The original PDF is preserved and only the chord symbols are covered and redrawn in place, so lyrics, staves, and spacing stay exactly where they were.
- Works with the tools you already use. Charts exported from MuseScore, Finale, Sibelius, Dorico, and LilyPond are read exactly from their text layer. Scanned or image-only PDFs are handled by OCR.
- Private by design. Rendering, chord detection, and PDF generation all happen locally in the browser. Nothing is sent to a server, and after the first visit it works offline.
- Free, no account, no limits.
- Circle-of-fifths wheel — tap a "from" key and a "to" key to transpose; the interval is shown as you go.
- By interval — shift up or down by a set number of semitones.
- Major / minor aware, with a relative-minor view on the wheel.
- Nashville numbers — output chords as a number system relative to the key instead of letter names.
- Enharmonic control — choose automatic sharp/flat spelling, or force sharps or flats.
- Simplify — reduce extended chords to their triad (e.g.
Cmaj9→C). - Capo suggestions to play in an easier shape.
- Exact text extraction for digitally-exported PDFs — chords are read directly from the embedded text, with no OCR and no guessing. They're identified by chord spelling and by the font/size the chart uses for chords, so it's completely color-independent.
- In-browser PP-OCR for scanned and image-only PDFs — a modern text detector + recognizer (PaddleOCR PP-OCR) running via ONNX Runtime Web with WebGPU acceleration, and an automatic Tesseract fallback if needed.
- Music-aware filtering — chords are separated from staff lines, note-name letters, fingerings, page/measure numbers, and other musical marks using chord grammar, text-size clustering, and chord-line structure. Numbers like
8or67won't be turned into chords. - Detection strength — choose Precise, Balanced, or Aggressive to trade false positives against catching every last chord.
- Two review modes — a compact before → after chip list, or an on-page overlay where detected chords are highlighted directly on your chart.
- Edit, add, or delete any chord — corrections re-transpose instantly.
- Confidence flags — low-confidence OCR reads are highlighted and summarised so you know what to double-check.
- Find & replace across all detected chords, plus undo/redo.
- Transposed chords are drawn in a matched font, size, and color over the original chords, preserving the chart's look.
- Font choices — bundled fonts (Arial-compatible and more), any font installed on your computer, or upload your own
.ttf/.otf. - Pick the ink color for the new chords, with an eyedropper to match your chart.
- Setlist mode — queue several PDFs and work through them one after another.
- Session resume — pick up where you left off.
- Installable PWA — add it to your device and use it offline; OCR models and assets are cached after first use.
- Accessible — keyboard-operable controls (including the key wheel) and clear focus states.
- Light & dark themes.
- Drop in a PDF chord chart or lead sheet (one, or several for a setlist).
- Pick the new key on the circle-of-fifths wheel, or shift by an interval.
- Review the detected chords in the list or on the page, and fix anything that needs it.
- Download a transposed PDF that matches your original.
All four steps happen entirely in your browser.
SheetShift is a client-only web app — a static site with no backend or API.
- React + Vite front end.
- pdf.js for rendering and text extraction.
- onnxruntime-web (WebGPU/WASM) running PP-OCRv3 detection + recognition, with tesseract.js as a fallback.
- pdf-lib + fontkit for layout-preserving PDF output.
Because it's fully client-side, your charts never leave your device.