How Transposition Works (Plain Language)
Updated: February 20, 2026
Chord Transposer tries to detect chord symbols (for example G, F#m, Bb7) and shift them by the semitone value you choose. Lyrics and most non-chord text are kept as-is.
1. Text-first files (DOC/DOCX)
Most reliable path.
- We read structured text from the document.
- We detect chord-like tokens and transpose them.
- We preserve surrounding text as much as possible.
2. PDF files
PDFs can be either:
- Text-based PDFs: behave similarly to DOC/DOCX and are usually good quality.
- Scanned/image PDFs: require OCR and can be less accurate.
3. Image files (JPG/JPEG/PNG)
Highest variance in quality.
- We use OCR to estimate text and chord positions.
- Low resolution, blur, skew, shadows, handwriting, unusual fonts, or dense layouts can reduce accuracy.
What Can Go Wrong on Scans?
- Chord symbols may be missed or misread (for example
Bvs8). - Line alignment may shift when source spacing is irregular.
- Small superscripts/suffixes (like
maj7,sus4) may be partially detected.
How to Get Better Results
- Prefer DOC/DOCX or clean text PDFs when possible.
- For scans, use high resolution and high contrast.
- Avoid perspective tilt; crop tightly around content.
- Review output before rehearsal/performance.
Expectation Setting
Short version: text files are usually strong; scanned images are best-effort.
If a scan-based result is unsatisfactory, use the Support flow and submit a refund request tied to that transposition.