On the road to V5
Next Audiveris version (V5) is meant to support even poor-quality scores.
To reach this goal, a deep software redesign was launched in early 2013 and
should get completed within a few months from now (April 2014).
During this very busy development phase, support on current V4 will be limited
essentially to Audiveris Community forum.
is an open-source Optical Music Recognition
software which processes the image of a music sheet to
automatically provide symbolic music information in
This opens the door to many tools (score editor, MIDI
sequencer, ...) which can use this symbolic data for such
tasks as edit, play, print, re-publish, transpose, query,
- Printed music as input (no handwritten music)
- Standard music notation (no tablatures yet)
- Input formats: PDF, JPG, PNG, TIFF, BMP, ...
- Output format:
MusicXML version 3
- Any number of pages per score, of parts per system,
of staves per part, of voices per measure
- Internal neural network trainable by end user
- Available on Windows and Linux
- GNU GPL V2 license
Audiveris is available through Java Web Start:
Simply click on this Launch button:
Or click on this
(assuming JNLP MIME type is properly defined as application/x-java-jnlp-file)
Or type this command from a terminal
(mind the 's' in https):
site provides representative examples of PDF scores and
Wiki details how Audiveris can be used to
process each of these examples.
Audiveris is developed in Java, and invokes Google Tesseract
(C++) for text OCR.
Project components (source code, binaries, issues tracking, forum)
are available on
Help wanted for
- Coupling with MuseScore
to edit, print, play, etc
- Cloud computing to provide "OMR as a service"
- Automated evaluation of transcription results
- Microedition features to enable crowd-sourcing
See details on dedicated
and share your expectations on Audiveris
- Test campaign on large data set?
- Missing features: cue notes, endings, etc?
- Ability to move & resize virtual glyphs?
- Map/reduce for scores with lots of pages?
- Text OCR extended to musical symbols?
- Support for low-quality scans?