Command Line Interface - Get errors

[Blaxzter]

Hey everyone,

i'm using the tool to transcribe a bunch of sheet music and had to realize that this:

was sadly transcribed to this

which is then - in an opened audiveris GUI - marked as error, which I can correct.
But when using the CLI it happily exports the music (mxl) and removes the second note.

Here are my questions:
What would we have to change in the original score to minimize these detection errors, as I don't really understand why it was unable to detect the beam.
And when using the CLI, how do I mute all output and get wrong tabs. Or parse the output so I can flag it for error and go over it later.

Best wishes
Frederic

[hbitteur]

@Blaxzter
Generally, beams recognition is pretty good... provided the typical beam height has been correctly determined upfront during the SCALE step (or manually specified by the user).

Here, the image is of very good quality, therefore we can suspect a wrong height guess (due to the lack of enough beams in the sheet at hand).
Could you post the input file for this sheet?

[Blaxzter]

Thank you for your response.

Do you have any information regarding the second question about errors?

[Blaxzter]

Output from GUI:

Audiveris version 5.4-alpha

LogUtil. Property logback.configurationFile not defined, skipped.
LogUtil. No ...\AudiverisLtd\audiveris\config\logback.xml, skipped.
LogUtil. Configuration found jar:file:/C:/Program%20Files/Audiveris/lib/audiveris.jar!/res/logback.xml
LogUtil. Logging to ...\AppData\Roaming\AudiverisLtd\audiveris\log\20240328T181300.log

Classifier loaded XML norms.
Classifier data loaded from default uri jar:file:/C:/Program%20Files/Audiveris/lib/audiveris.jar!/res/basic-classifier.zip
No ...\AudiverisLtd\audiveris\config\plugins.xml file found
Versions. Poll frequency: Weekly, next poll on: 29-Mar-2024
MainGui submitting Book "...\audivers\test.omr"
Loading book ...\audivers\test.omr
[test] Loaded /sheet#1/sheet#1.xml
[test] Closing {Score 1}
[test] 1 sheet in ...\test.png
[test] Loaded image 1 1890x587 from. ...t\test.png
[test] Discarding alpha band ...
[test] Converting max RGB to gray
[test] Book reaching PAGE on sheets:[#1]
[test] Beam  guessed height: 15 -- 0.50 of 30 interline
[test] Beam measured height: 29 -- 0.95 of [8..30] range at 110% of needed quorum
[test] Scale{ interline(30,30,30) line(3,3,3) beam(29)}
[test] Global slope: 0.00000
[test] Retrieved raw line clusters: 2 of sizes [5] with interline(30,30,30)
[test] Systems: #1[1] #2[2]
[test] Staff#1 no clear end on LEFT
[test] Staff#2 no clear end on LEFT
[test] 1 part along 2 systems
[test] Created scores: [{Score 1}]
[test] Staff#2 no recognized header clef.
[test] Discarding alpha band ...
[test] Converting max RGB to gray
[test] stem(3 max:4)
[test] Scale information: HeadSeeds{NOTEHEAD_BLACK[R:-0.5]}
[test] Stems free length median value: 105 pixels, 3.5 interlines
[test] Step CUE_BEAMS is skipped because small heads switch is off
[test] 9 raw measures: [4 in system#1, 5 in system#2]
[test] Slurs: 1
[test] Segments: 0
[test] Measure{#8P1} No timeOffset for HeadChordInter#400{(0.947/0.947) slot#5 dur:1/4}

And the cli command:
audiveris -batch -print -transcribe -export -output "audivers" "test.png"

INFO  []                       CLI 282  | CLI args: [-batch, -print, -transcribe, -export, -output, audivers, test.png]
INFO  []              TesseractOCR 116  | TESSDATA_PREFIX value: null
INFO  []              TesseractOCR 245  | OCR folder: ...\AudiverisLtd\audiveris\config\tessdata
INFO  []                      Main 402  | Environment:
- Audiveris:    5.4-alpha:698b2bb69
- OS:           Windows 11 10.0
- Architecture: amd64
- Java VM:      OpenJDK 64-Bit Server VM (build 17.0.10+7-LTS, mixed mode, sharing)
- OCR Engine:   Tesseract OCR, version 5.3.1
INFO  []             AliasPatterns 134  | Alias patterns: [(IMSLP[0-9]*)-.*]
INFO  []                      Main 242  | Running in batch mode
INFO  []                      Main 190  | Submitting 1 task(s) in sequence:
    Input "test.png"
INFO  [test]                      Book 552  | 1 sheet in test.png
INFO  []                      Book 2551 | Stored /book.xml
INFO  []                      Book 2510 | Book stored as audivers\test.omr
INFO  []                      Book 1951 | Book reaching PAGE on sheets:[#1]
INFO  [test]            StepMonitoring 98   | LOAD
INFO  [test]                      Book 1794 | Loaded image 1 1890x587 from test.png
INFO  [test]                 ImageUtil 204  | Discarding alpha band ...
INFO  [test]                 ImageUtil 122  | Converting max RGB to gray
INFO  [test]            StepMonitoring 98   | BINARY
INFO  [test]            StepMonitoring 98   | SCALE
INFO  [test]              ScaleBuilder 238  | Beam  guessed height: 15 -- 0.50 of 30 interline
INFO  [test]              ScaleBuilder 261  | Beam measured height: 29 -- 0.95 of [8..30] range at 110% of needed quorum
INFO  [test]                 ScaleStep 65   | Scale{ interline(30,30,30) line(3,3,3) beam(29)}
INFO  [test]            StepMonitoring 98   | GRID
INFO  [test]            LinesRetriever 1518 | Global slope: 0.00000
INFO  [test]         ClustersRetriever 328  | Retrieved raw line clusters: 2 of sizes [5] with interline(30,30,30)
INFO  [test]                 PeakGraph 310  | Systems: #1[1] #2[2]
WARN  [test]            StaffProjector 341  | Staff#1 no clear end on LEFT
WARN  [test]            StaffProjector 341  | Staff#2 no clear end on LEFT
INFO  [test]             SystemManager 736  | 1 part along 2 systems
INFO  [test]                      Book 535  | Created scores: [{Score 1}]
INFO  [test]            StepMonitoring 98   | HEADERS
INFO  [test]           BasicClassifier 277  | Classifier loaded XML norms.
INFO  [test]        AbstractClassifier 396  | Classifier data loaded from default uri jar:file:/C:/Program%20Files/Audiveris/lib/audiveris.jar!/res/basic-classifier.zip
WARN  [test]               ClefBuilder 664  | Staff#2 no recognized header clef.
INFO  [test]            StepMonitoring 98   | STEM_SEEDS
INFO  [test]                 ImageUtil 204  | Discarding alpha band ...
INFO  [test]                 ImageUtil 122  | Converting max RGB to gray
INFO  [test]             StemSeedsStep 86   | stem(3 max:4)
INFO  [test]            StepMonitoring 98   | BEAMS
INFO  [test]            StepMonitoring 98   | LEDGERS
INFO  [test]            StepMonitoring 98   | HEADS
INFO  [test]             HeadSeedTally 235  | Scale information: HeadSeeds{NOTEHEAD_BLACK[R:-0.5]}
INFO  [test]            StepMonitoring 98   | STEMS
INFO  [test]            StepMonitoring 98   | REDUCTION
INFO  [test]             ReductionStep 93   | Stems free length median value: 105 pixels, 3.5 interlines
INFO  [test]            StepMonitoring 98   | CUE_BEAMS
INFO  [test]              CueBeamsStep 81   | Step CUE_BEAMS is skipped because small heads switch is off
INFO  [test]            StepMonitoring 98   | TEXTS
INFO  [test]            StepMonitoring 98   | MEASURES
INFO  [test]                      Page 307  | 9 raw measures: [4 in system#1, 5 in system#2]
INFO  [test]            StepMonitoring 98   | CHORDS
INFO  [test]            StepMonitoring 98   | CURVES
INFO  [test]              SlursBuilder 240  | Slurs: 1
INFO  [test]           SegmentsBuilder 155  | Segments: 0
INFO  [test]            StepMonitoring 98   | SYMBOLS
INFO  [test]            StepMonitoring 98   | LINKS
INFO  [test]            StepMonitoring 98   | RHYTHMS
INFO  [test]             MeasureRhythm 985  | Measure{#8P1} No timeOffset for HeadChordInter#400{(0.947/0.947) slot#5 dur:1/4}
INFO  [test]            StepMonitoring 98   | PAGE
INFO  [test]                      Book 2571 | Book{test} storing
INFO  [test]                      Book 2551 | Stored /book.xml
INFO  [test]                DataHolder 348  | Stored /sheet#1/BINARY.png
INFO  [test]                     Sheet 1544 | Stored /sheet#1/sheet#1.xml
INFO  [test]                      Book 2510 | Book stored as audivers\test.omr
INFO  [test]                 SheetStub 1578 | Disposed sheet
INFO  [test]                      Book 2038 | End of Stub#1 memory: 29,782,832
INFO  [test]                 SheetStub 967  | Loaded /sheet#1/sheet#1.xml
INFO  [test]           PartwiseBuilder 2651 | Exporting sheet(s): [#1]
INFO  [test]             ScoreExporter 164  | Score test exported to audivers\test.mxl
INFO  [test]             BookPdfOutput 142  | Printed sheet(s): [1]
INFO  [test]                      Book 1902 | Book printed to audivers\test-print.pdf
INFO  [test]                     Score 187  | Closing {Score 1}

My question:

Is the log indicating a warning / error this one: Measure{#8P1} No timeOffset for HeadChordInter#400{(0.947/0.947) slot#5 dur:1/4} ?
And can i filter for it in the logs to find faulty lines?

[hbitteur]

@Blaxzter
I'm getting back to you after several days away from home. Sorry for that.

I just ran Audiveris 5.4-alpha on your input file, and I got this:

[316659608-300a47de-0354-4941-ab46-9270f1060814] Beam guessed height: 15 -- 0.50 of 30 interline
[316659608-300a47de-0354-4941-ab46-9270f1060814] Beam measured height: 29 -- 0.95 of [8..30] range at 77% of needed quorum
[316659608-300a47de-0354-4941-ab46-9270f1060814] No reliable beam height found, guessed value: 15
[316659608-300a47de-0354-4941-ab46-9270f1060814] Scale{ interline(30,30,30) line(3,3,3) beam(15 extra)}

to be compared with "your" results:

INFO [test] ScaleBuilder 238 | Beam guessed height: 15 -- 0.50 of 30 interline
INFO [test] ScaleBuilder 261 | Beam measured height: 29 -- 0.95 of [8..30] range at 110% of needed quorum
INFO [test] ScaleStep 65 | Scale{ interline(30,30,30) line(3,3,3) beam(29)}

In "my" results, the measured beam height (29 pixels) has been rejected, and the guessed value (15 pixels) was finally chosen.
If we manually check the beam height by looking in the Binary tab, we can see that 15 is indeed the correct beam height value.

Not surprisingly, the later BEAMS step is able to recognize the beam element. While in your case, 29 pixels is way too big, resulting in no beam being recognized.

The question is: Why do we get different behaviors of the SCALE step?
Could you check the constant beamMinCountRatio using the dialog Tools | Options.
Its value should be 0.01 (and not 0.007 as it was before)

Please make sure you are using the latest development version. For this use:

git pull --all

Final transcription result is:

Note that Audiveris complains:

• There is no blank margin on the left side of the image (fortunately, the starting barlines could be detected)
• The second staff lacks a mandatory starting clef