< Previous | Contents | Next >

3.7 Recognition Accuracy

If recognition accuracy is not as good as what you expected, examine the image up close by zooming into the image pane of your ENF view. You may also view Image Editor (FIle

> Open > Image file type).

Determine the following by zooming close up:

If objects lack detail, re-scan with the Auto-Resolution setting checked (On) or increase scanning resolution manually. Optimum distance between horizontal staf- flines is 20 pixels. You may use the cursor and readout in the Status Bar (pixel coordi- nates) to measure exact width of a typical staff line.

If objects such as note stems and staff lines are broken or lack thickness, re-scan with decreased brightness to fill in gaps.

If the source document is poorly printed or is faded and weak, try locating a better print copy of the music.


Following recognition, guitar chord fret symbols may overlay chord names rec- ognized as text: “Dumb” text chord names will not transpose. To remove “dumb” text fields, use Nudge mode (Shift) and drag the control handles of each text field upwards. Use the Select tool and “group delete” all selected text fields.

Recognition accuracy is directly related to the quality of the source document as well as the quality of the scanned image. Cheaper scanners, including Fax-Scan-Copy combos, are known to create poor images. Remember the old saw: “Garbage in / Garbage out”. If after examining the image and consulting the Troubleshooting chart, you are still unable to achieve good recognition results, we invite you to send us your TIFF file(s) for detailed analysis by our technicians.

You can attach your image (TIFF) files to e-mail along with a short description of the problem you experienced to the following address: tech@musitek.com. We will reply as soon as possible. Most likely, we will be able to locate the problem and offer a solution or recommendation.

NOTE: Many PDF files and music printed from laser printers have extremely narrow note stems and staff lines. If note stems appear “razor thin” and recognition was poorer than expected, open the .tif file in the Image Editor and apply darker thresholding to the image. See Section 3.9.2, "Image Editing Tools".