[ { "path": "AUTHORS", "content": "# This is the list of Moonlight OMR authors for copyright purposes.\n#\n# Copyright for contributions made under Google's Corporate CLA belong to the\n# contributor's organization. Contributions made under the Individual CLA belong\n# to the author. Individual contributors are recognized separately in the\n# \"CONTRIBUTORS\" file.\nGoogle Inc.\nNuno Jesus\n" }, { "path": "CONTRIBUTING.md", "content": "# How to Contribute\n\nWe'd love to accept your patches and contributions to this project. There are\njust a few small guidelines you need to follow.\n\n## Contributor License Agreement\n\nContributions to this project must be accompanied by a Contributor License\nAgreement. You (or your employer) retain the copyright to your contribution;\nthis simply gives us permission to use and redistribute your contributions as\npart of the project. Head over to to see\nyour current agreements on file or to sign a new one.\n\nYou generally only need to submit a CLA once, so if you've already submitted one\n(even if it was for a different project), you probably don't need to do it\nagain.\n\n## Code reviews\n\nAll submissions, including submissions by project members, require review. We\nuse GitHub pull requests for this purpose. Consult\n[GitHub Help](https://help.github.com/articles/about-pull-requests/) for more\ninformation on using pull requests.\n\n## Community Guidelines\n\nThis project follows [Google's Open Source Community\nGuidelines](https://opensource.google.com/conduct/).\n" }, { "path": "CONTRIBUTORS", "content": "# This is the list of individual contributors to the Moonlight OMR project.\n#\n# Some contributions belong to the organization of the contributor. Copyright is\n# tracked separately, in the \"AUTHORS\" file.\n#\n# We will make an effort to recognize all contributors here. However, the source\n# of truth for contributors is the commit author in source control history.\nDan Ringwalt\nLarry Li\nNuno Jesus\n" }, { "path": "LICENSE", "content": "\n Apache License\n Version 2.0, January 2004\n http://www.apache.org/licenses/\n\n TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION\n\n 1. Definitions.\n\n \"License\" shall mean the terms and conditions for use, reproduction,\n and distribution as defined by Sections 1 through 9 of this document.\n\n \"Licensor\" shall mean the copyright owner or entity authorized by\n the copyright owner that is granting the License.\n\n \"Legal Entity\" shall mean the union of the acting entity and all\n other entities that control, are controlled by, or are under common\n control with that entity. For the purposes of this definition,\n \"control\" means (i) the power, direct or indirect, to cause the\n direction or management of such entity, whether by contract or\n otherwise, or (ii) ownership of fifty percent (50%) or more of the\n outstanding shares, or (iii) beneficial ownership of such entity.\n\n \"You\" (or \"Your\") shall mean an individual or Legal Entity\n exercising permissions granted by this License.\n\n \"Source\" form shall mean the preferred form for making modifications,\n including but not limited to software source code, documentation\n source, and configuration files.\n\n \"Object\" form shall mean any form resulting from mechanical\n transformation or translation of a Source form, including but\n not limited to compiled object code, generated documentation,\n and conversions to other media types.\n\n \"Work\" shall mean the work of authorship, whether in Source or\n Object form, made available under the License, as indicated by a\n copyright notice that is included in or attached to the work\n (an example is provided in the Appendix below).\n\n \"Derivative Works\" shall mean any work, whether in Source or Object\n form, that is based on (or derived from) the Work and for which the\n editorial revisions, annotations, elaborations, or other modifications\n represent, as a whole, an original work of authorship. For the purposes\n of this License, Derivative Works shall not include works that remain\n separable from, or merely link (or bind by name) to the interfaces of,\n the Work and Derivative Works thereof.\n\n \"Contribution\" shall mean any work of authorship, including\n the original version of the Work and any modifications or additions\n to that Work or Derivative Works thereof, that is intentionally\n submitted to Licensor for inclusion in the Work by the copyright owner\n or by an individual or Legal Entity authorized to submit on behalf of\n the copyright owner. For the purposes of this definition, \"submitted\"\n means any form of electronic, verbal, or written communication sent\n to the Licensor or its representatives, including but not limited to\n communication on electronic mailing lists, source code control systems,\n and issue tracking systems that are managed by, or on behalf of, the\n Licensor for the purpose of discussing and improving the Work, but\n excluding communication that is conspicuously marked or otherwise\n designated in writing by the copyright owner as \"Not a Contribution.\"\n\n \"Contributor\" shall mean Licensor and any individual or Legal Entity\n on behalf of whom a Contribution has been received by Licensor and\n subsequently incorporated within the Work.\n\n 2. Grant of Copyright License. Subject to the terms and conditions of\n this License, each Contributor hereby grants to You a perpetual,\n worldwide, non-exclusive, no-charge, royalty-free, irrevocable\n copyright license to reproduce, prepare Derivative Works of,\n publicly display, publicly perform, sublicense, and distribute the\n Work and such Derivative Works in Source or Object form.\n\n 3. Grant of Patent License. Subject to the terms and conditions of\n this License, each Contributor hereby grants to You a perpetual,\n worldwide, non-exclusive, no-charge, royalty-free, irrevocable\n (except as stated in this section) patent license to make, have made,\n use, offer to sell, sell, import, and otherwise transfer the Work,\n where such license applies only to those patent claims licensable\n by such Contributor that are necessarily infringed by their\n Contribution(s) alone or by combination of their Contribution(s)\n with the Work to which such Contribution(s) was submitted. If You\n institute patent litigation against any entity (including a\n cross-claim or counterclaim in a lawsuit) alleging that the Work\n or a Contribution incorporated within the Work constitutes direct\n or contributory patent infringement, then any patent licenses\n granted to You under this License for that Work shall terminate\n as of the date such litigation is filed.\n\n 4. Redistribution. You may reproduce and distribute copies of the\n Work or Derivative Works thereof in any medium, with or without\n modifications, and in Source or Object form, provided that You\n meet the following conditions:\n\n (a) You must give any other recipients of the Work or\n Derivative Works a copy of this License; and\n\n (b) You must cause any modified files to carry prominent notices\n stating that You changed the files; and\n\n (c) You must retain, in the Source form of any Derivative Works\n that You distribute, all copyright, patent, trademark, and\n attribution notices from the Source form of the Work,\n excluding those notices that do not pertain to any part of\n the Derivative Works; and\n\n (d) If the Work includes a \"NOTICE\" text file as part of its\n distribution, then any Derivative Works that You distribute must\n include a readable copy of the attribution notices contained\n within such NOTICE file, excluding those notices that do not\n pertain to any part of the Derivative Works, in at least one\n of the following places: within a NOTICE text file distributed\n as part of the Derivative Works; within the Source form or\n documentation, if provided along with the Derivative Works; or,\n within a display generated by the Derivative Works, if and\n wherever such third-party notices normally appear. The contents\n of the NOTICE file are for informational purposes only and\n do not modify the License. You may add Your own attribution\n notices within Derivative Works that You distribute, alongside\n or as an addendum to the NOTICE text from the Work, provided\n that such additional attribution notices cannot be construed\n as modifying the License.\n\n You may add Your own copyright statement to Your modifications and\n may provide additional or different license terms and conditions\n for use, reproduction, or distribution of Your modifications, or\n for any such Derivative Works as a whole, provided Your use,\n reproduction, and distribution of the Work otherwise complies with\n the conditions stated in this License.\n\n 5. Submission of Contributions. Unless You explicitly state otherwise,\n any Contribution intentionally submitted for inclusion in the Work\n by You to the Licensor shall be under the terms and conditions of\n this License, without any additional terms or conditions.\n Notwithstanding the above, nothing herein shall supersede or modify\n the terms of any separate license agreement you may have executed\n with Licensor regarding such Contributions.\n\n 6. Trademarks. This License does not grant permission to use the trade\n names, trademarks, service marks, or product names of the Licensor,\n except as required for reasonable and customary use in describing the\n origin of the Work and reproducing the content of the NOTICE file.\n\n 7. Disclaimer of Warranty. Unless required by applicable law or\n agreed to in writing, Licensor provides the Work (and each\n Contributor provides its Contributions) on an \"AS IS\" BASIS,\n WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or\n implied, including, without limitation, any warranties or conditions\n of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A\n PARTICULAR PURPOSE. You are solely responsible for determining the\n appropriateness of using or redistributing the Work and assume any\n risks associated with Your exercise of permissions under this License.\n\n 8. Limitation of Liability. In no event and under no legal theory,\n whether in tort (including negligence), contract, or otherwise,\n unless required by applicable law (such as deliberate and grossly\n negligent acts) or agreed to in writing, shall any Contributor be\n liable to You for damages, including any direct, indirect, special,\n incidental, or consequential damages of any character arising as a\n result of this License or out of the use or inability to use the\n Work (including but not limited to damages for loss of goodwill,\n work stoppage, computer failure or malfunction, or any and all\n other commercial damages or losses), even if such Contributor\n has been advised of the possibility of such damages.\n\n 9. Accepting Warranty or Additional Liability. While redistributing\n the Work or Derivative Works thereof, You may choose to offer,\n and charge a fee for, acceptance of support, warranty, indemnity,\n or other liability obligations and/or rights consistent with this\n License. However, in accepting such obligations, You may act only\n on Your own behalf and on Your sole responsibility, not on behalf\n of any other Contributor, and only if You agree to indemnify,\n defend, and hold each Contributor harmless for any liability\n incurred by, or claims asserted against, such Contributor by reason\n of your accepting any such warranty or additional liability.\n\n END OF TERMS AND CONDITIONS\n\n APPENDIX: How to apply the Apache License to your work.\n\n To apply the Apache License to your work, attach the following\n boilerplate notice, with the fields enclosed by brackets \"[]\"\n replaced with your own identifying information. (Don't include\n the brackets!) The text should be enclosed in the appropriate\n comment syntax for the file format. We also recommend that a\n file or class name and description of purpose be included on the\n same \"printed page\" as the copyright notice for easier\n identification within third-party archives.\n\n Copyright [yyyy] [name of copyright owner]\n\n Licensed under the Apache License, Version 2.0 (the \"License\");\n you may not use this file except in compliance with the License.\n You may obtain a copy of the License at\n\n http://www.apache.org/licenses/LICENSE-2.0\n\n Unless required by applicable law or agreed to in writing, software\n distributed under the License is distributed on an \"AS IS\" BASIS,\n WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n See the License for the specific language governing permissions and\n limitations under the License.\n" }, { "path": "README.md", "content": "\n\n# Moonlight Optical Music Recognition (OMR) [![Build Status](https://travis-ci.org/tensorflow/moonlight.svg?branch=master)](https://travis-ci.org/tensorflow/moonlight)\n\nAn experimental [optical music\nrecognition](https://en.wikipedia.org/wiki/Optical_music_recognition) engine.\n\nMoonlight reads PNG image(s) containing sheet music and outputs\n[MusicXML](https://www.musicxml.com/) or a\n[NoteSequence message](https://github.com/tensorflow/magenta/blob/master/magenta/protobuf/music.proto).\nMusicXML is a standard sheet music interchange format, and `NoteSequence` is\nused by [Magenta](http://magenta.tensorflow.org) for training generative music\nmodels.\n\nMoonlight is not an officially supported Google product.\n\n### Command-Line Usage\n\n git clone https://github.com/tensorflow/moonlight\n cd moonlight\n # You may want to run this inside a virtualenv.\n pip install -r requirements.txt\n # Build the OMR command-line tool.\n bazel build moonlight:omr\n # Prints a Score message.\n bazel-bin/moonlight/omr moonlight/testdata/IMSLP00747-000.png\n # Scans several pages and prints a NoteSequence message.\n bazel-bin/moonlight/omr --output_type=NoteSequence IMSLP00001-*.png\n # Writes MusicXML to ~/mozart.xml.\n bazel-bin/moonlight/omr --output_type=MusicXML --output=$HOME/mozart.xml \\\n corpus/56/IMSLP56442-*.png\n\nThe `omr` CLI will print a [`Score`](moonlight/protobuf/musicscore.proto)\nmessage by default, or [MusicXML](https://www.musicxml.com/) or a\n`NoteSequence` message if specified.\n\nMoonlight is intended to be run in bulk, and will not offer a full UI for\ncorrecting the score. The main entry point will be an Apache Beam pipeline that\nprocesses an entire corpus of images.\n\nThere is no release yet, and Moonlight is not ready for end users. To run\ninteractively or import the module, you can use the [sandbox\ndirectory](sandbox/README.md). Moonlight will be used offline for digitizing\na scanned corpus (it can be installed on all Cloud Compute platforms, and OS\ncompatibility is not a priority).\n\n### Dependencies\n\n* Linux\n - Note: Our Google dep versions are fragile, and updating them or updating other OS may break directory structure in fragile ways.\n* [Protobuf 3.6.1](https://pypi.org/project/protobuf/3.6.1/)\n* [Bazel 0.20.0](https://github.com/bazelbuild/bazel/releases/tag/0.20.0). We\n encountered some errors using Bazel 0.21.0 to build Protobuf 3.6.1, which is\n the latest Protobuf release at the time of writing.\n* Python version supported by TensorFlow (Python 3.5-3.7)\n* Python dependencies specified in the [requirements](requirements.txt).\n\n### Resources\n\n[Forum](https://groups.google.com/forum/#!forum/moonlight-omr)\n" }, { "path": "WORKSPACE", "content": "http_archive(\n name = \"com_google_protobuf\",\n sha256 = \"40f009cb0c190816a52fc21d45c26558ee7d63c3bd511b326bd85739b2fd99a6\",\n strip_prefix = \"protobuf-3.6.1\",\n url = \"https://github.com/google/protobuf/releases/download/v3.6.1/protobuf-python-3.6.1.tar.gz\",\n)\n\nnew_http_archive(\n name = \"six_archive\",\n build_file = \"six.BUILD\",\n sha256 = \"105f8d68616f8248e24bf0e9372ef04d3cc10104f1980f54d57b2ce73a5ad56a\",\n strip_prefix = \"six-1.10.0\",\n url = \"https://pypi.python.org/packages/source/s/six/six-1.10.0.tar.gz#md5=34eed507548117b2ab523ab14b2f8b55\",\n)\n\nbind(\n name = \"six\",\n actual = \"@six_archive//:six\",\n)\n\nnew_http_archive(\n name = \"magenta\",\n strip_prefix = \"magenta-48a199085e303eeae7c36068f050696209b856bb/magenta\",\n url = \"https://github.com/tensorflow/magenta/archive/48a199085e303eeae7c36068f050696209b856bb.tar.gz\",\n sha256 = \"931fb7b57714d667db618b0c31db1444e44baab17865ad66b76fd24d9e20ad6d\",\n build_file_content = \"\",\n)\n\nnew_http_archive(\n name = \"omr_regression_test_data\",\n strip_prefix = \"omr_regression_test_data_20180516\",\n url = \"https://github.com/tensorflow/moonlight/releases/download/v2018.05.16-data/omr_regression_test_data_20180516.tar.gz\",\n sha256 = \"b47577ee6b359c2cbbcdb8064c6bd463692c728c55ec5c0ab78139165ba8f35a\",\n build_file_content = \"\"\"\npackage(default_visibility = [\"//visibility:public\"])\nfilegroup(\n name = \"omr_regression_test_data\",\n srcs = glob([\"**/*.png\"]),\n)\n\"\"\",\n)\n" }, { "path": "docs/concepts.md", "content": "# Moonlight OMR Concepts\n\n## Diagram\n\n\n\n## Glossary\n\n| Term | Definition |\n| ------------ | ------------------------------------------------------------- |\n| barline | A vertical line spanning one or more staves, which is not a |\n: : stem. :\n| beam | A thick line connecting multiple stems horizontally. Each |\n: : beam halves the duration of a filled notehead connected to :\n: : the stem, which would otherwise be a quarter note. :\n| notehead | An ellipse representing a single note. May be filled (quarter |\n: : or lesser value) or empty (half or whole note). :\n| ledger line | An extra line above or below the 5 staff lines. |\n| staff | The object which notes and other glyphs are placed on. |\n| staff system | One or more staves joined by at least one barline. |\n| staves | Plural of staff. |\n| stem | A vertical line attached to a notehead. All noteheads except |\n: : for whole notes should have a stem. :\n\n## Staves\n\nStaves have 5 parallel, horizontal lines, and are parameterized by the center\n(3rd) line, and the staffline distance (vertical distance between consecutive\nstaff lines). The staffline distance is constant for reasonable quality printed\nmusic scores, and this representation avoids redundancy and makes it possible to\nfind the coordinates of each staff line.\n\n## Staff Positions\n\nGlyphs are vertically centered either on staff lines (or ledger lines), or on\nthe space halfway between lines. We refer to each of these y coordinates as a\n*position*.\n\n**Note**: We still refer to positions as *stafflines* in many places which is\ncounter-intuitive, since positions are also on the space between lines, and can\nbe a potential ledger line which is empty in a particular image. We are in the\nprocess of renaming stafflines in this context to staff positions.\n\n* Staff position 0 is the third staff line (staff center line), which is also\n the y-coordinate that is outputted by staff detection.\n* Staff positions are half the staffline distance apart. Always calculate the\n relative y position by tf.floordiv(staffline_distance * y_position,\n 2) instead of dividing by 2 first.\n* In treble clef, staff position 0 is B4, -6 is C4, -1 is A4, and +1 is C5.\n* In bass clef, staff position 0 is D3, +3 is G3, and +6 is C4.\n\n## Glyphs\n\nGlyphs are defined in\n[musicscore.proto](../moonlight/protobuf/musicscore.proto). Each\nglyph has an x coordinate on the original image, and a y position (staff\nposition). The staff position determines the pitch of the glyph, if applicable.\nIf the glyph is especially large (e.g. clefs) or is not centered on a particular\nvertical position, we choose a *canonical* staff position (e.g. the G line for\ntreble clef, aka G clef).\n\n### Glyph centers\n\n* Every glyph needs a canonical center coordinate. The classifier will detect\n the glyph if the window (e.g. convolutional filter) is centered on this\n point.\n* Noteheads are centered on the middle of the ellipsis, which should be\n exactly on a staffline or halfway between stafflines.\n* Accidentals should be centered on their center of mass. For flats, sharps,\n and naturals, this is the center of the empty space inside them. For double\n sharps, this is the middle of the crosshairs.\n* Treble clef ('G clef\") is centered on the intersection of the thin vertical\n line with the G staffline (staff position -2).\n* Bass clef (\"F clef\") is centered on the F staffline (staff position +2). The\n x coordinate is halfway between the filled circle on the left and the\n vertical segment on the right.\n* All rests should normally be centered at staff position 0, unless shifted\n from their usual position.\n" }, { "path": "docs/engine.md", "content": "## OMR Engine Detailed Reference\n\nThe OMR engine converts a PNG image to a Magenta\n[NoteSequence message](https://github.com/magenta/note-seq/blob/master/note_seq/protobuf/music.proto),\nwhich is interoperable with MIDI and MusicXML.\n\nOMR uses TensorFlow for glyph (symbol) classification, as well as all other\ncompute-intensive steps. Final processing is done in pure Python. The entry\npoint is [OMREngine.run](../moonlight/engine.py). Glyph classification is\nconfigurable by the`glyph_classifier_fn` argument, and other subtasks of image\nrecognition are part of the [Structure](../moonlight/structure/__init__.py).\n\n### Diagram\n\n\n\n### API\n\nThe entry point for running OMR is [`OMREngine.run()`](../moonlight/engine.py).\nIt takes in a list of PNG filenames and outputs a `Score` or `NoteSequence`\nmessage. The `Score` can further be converted to\n[MusicXML](../moonlight/conversions/musicxml.py).\n\n### TensorFlow Graph\n\nFor maximum parallelism, all processing is run in the same TensorFlow graph. The\ngraph is run by `OMREngine._get_page()`. This also evaluates the `Structure` in\nthe same graph. `Structure` is a wrapper for detectors which extract information\nfrom the image, including [staves](../moonlight/staves/hough.py), [vertical\nlines](../moonlight/structure/verticals.py), and [note\nbeams](../moonlight/structure/beams.py).\n\n#### Structure\n\n[Structure](../moonlight/structure/__init__.py) holds the structural elements\nthat need to be evaluated for OMR, but does not do any symbol recognition. The\nstructure encompasses staves, beams, and vertical lines, and may contain more\nelements in the future (e.g. full connected component analysis) which can be\nused to detect more elements (e.g. note dots). Structure detection is currently\nsimple computer vision rather than ML, but it can easily be swapped out with a\ndifferent TensorFlow model.\n\n##### Staffline Distance Estimation\n\nWe estimate the [staffline\ndistance(s)](../moonlight/staves/staffline_distance.py) of the entire image.\nThere may be staves with multiple different sizes for different parts on a\nsingle page, but there should be just a few possible staffline distance values.\n\n##### Staff Detection\n\nConcrete subclasses of [BaseStaffDetector](../moonlight/staves/base.py) take in\nthe image and produce:\n\n* `staves`: Tensor of shape `(num_staves, num_points, 2)`. Coordinates of the\n staff center line (third line on the staff).\n* `staffline_distance`: Vector of the estimated staffline distance (distance\n between consecutive staff lines) for each staff.\n* `staffline_thickness`: Scalar thickness of staff lines. Assumed to be the\n same for all staves.\n* `staves_interpolated_y`: Tensor of shape `(num_staves, width)`. For each\n staff and column of the image, outputs the interpolated y position of the\n staff center line.\n\n##### Staff Removal\n\n[StaffRemover](../moonlight/staves/removal.py) takes in the image and staves,\nand outputs `remove_staves` which is the image with the staff lines erased. This\nis useful so that [glyphs](concepts.md) look the same whether they are centered\non a staff line or the space between lines. It is also used within the\nstructure, by beam detection.\n\n##### Beam Detection\n\n[Beams](../moonlight/structure/beams.py) are currently detected from connected\ncomponents on an\n[eroded](https://en.wikipedia.org/wiki/Mathematical_morphology#Binary_morphology)\nstaves-removed image. These are attached to notes by a `BeamProcessor`.\n\n##### Vertical Line Detection\n\n[ColumnBasedVerticals](../moonlight/structure/verticals.py) detects all vertical\nlines in the image. These will later be used as either stems or barlines.\n\n#### Glyph Classification: 1-D Convolutional Model\n\nWe also run a glyph classifier as part of the TensorFlow graph, which outputs\npredictions.\n\n##### Staffline Extraction\n\nGlyphs are considered to lie on a black staff line, or halfway between staff\nlines. For OMR, extracted stafflines are slices of the image that are either\ncentered on an staff line, or halfway between staff lines. The line that the\nextracted staffline lies on may just be referred to a staffline, or a y position\nof the staff.\n\n[StafflineExtractor](../moonlight/staves/staffline_extractor.py) extracts these\nvertical slices of the image, and scales their height to a constant value\n(currently, 18 pixels tall). `StaffRemover` is used so that all extracted\nstafflines should look similar.\n\n##### Glyph Classification\n\n[Glyphs](concepts.md) are classified on small, horizontal slices (currently, 15\npixels wide) of the extracted staffline, a 1D convolutional model.\n\nA [GlyphClassifier](../moonlight/glyphs/base.py) outputs a Tensor\n`staffline_predictions` of shape `(num_staves, num_stafflines, width)`. The\nvalues are for the `Glyph.Type` enum. Value 0 (UNKNOWN_TYPE) is not used; value\n1 (NONE) corresponds to no glyph.\n\n### Post-Processing\n\n#### Page Construction\n\nOMR processing operates on [Page\nprotos](../moonlight/protobuf/musicscore.proto). The Page is first constructed\nby `BaseGlyphClassifier.get_page`, which populates the glyphs on each staff.\nStaff location information is then added by `StaffProcessor`.\n\nSingle `Glyph`s are created from consecutive runs in `staffline_predictions`\nthat are classified as the same glyph type.\n\nAdditional processors modify the `Page` in place, usually adding information\nfrom the `Structure`. Each page is run through\n[`page_processors.process()`](../moonlight/page_processors.py), and then the\nscore (containing all pages) is run through\n[`score_processors.process()`](../moonlight/score_processors.py).\n\n#### Stem Detection\n\n[Stems](../moonlight/structure/stems.py) finds stem candidates from the vertical\nlines, and adds a `Stem` to notehead `Glyph`s if the closest stem is close\nenough to the expected position. The `ScoreReader` considers multiple noteheads\nwith identical `Stem`s as a single chord. Stems will also be used as a negative\nsignal to avoid detecting barlines in the same area.\n\n#### Beam Processing\n\nBeams from the `Structure` that are close enough to a stem are added to one or\nmore notes by [BeamProcessor](../moonlight/structure/beam_processor.py).\n\n#### Barlines\n\n[Barlines](../moonlight/structure/barlines.py) are detected from the verticals\nif they have not already been used as a stem.\n\n#### Score Reading\n\nThe [ScoreReader](../moonlight/score/reader.py) is the only score processor. It\ncan potentially use state that lasts across multiple pages, such as the current\ntime in the score, which needs to persist for the entire score.\n\nStaves are scanned from left to right for glyphs. The `ScoreReader` manages a\nhierarchy of state, from the global `ScoreState` to the `MeasureState`, holding\nlocal state such as accidentals. Based on the preceding `Glyph`s, each notehead\n`Glyph` gets assigned a\n[`Note`](https://github.com/magenta/note-seq/blob/d7153cdb26758a69c2fa022782c5817970de7066/note_seq/protobuf/music.proto#L104)\nfield holding its musical value.\n\nAfterwards, the `Score` can be converted to a `NoteSequence` (just pulling out\nall of the `Note`s) or [MusicXML](../moonlight/conversions/musicxml.py).\n" }, { "path": "moonlight/BUILD", "content": "# Description:\n# Optical music recognition using TensorFlow.\n\npackage(\n default_visibility = [\"//moonlight:__subpackages__\"],\n)\n\nlicenses([\"notice\"]) # Apache 2.0\n\n# The OMR engine. Entry point for running OMR.\npy_library(\n name = \"engine\",\n srcs = [\"engine.py\"],\n srcs_version = \"PY2AND3\",\n deps = [\n \":image\",\n \":page_processors\",\n \":score_processors\",\n \"//moonlight/conversions\",\n \"//moonlight/glyphs:saved_classifier_fn\",\n \"//moonlight/protobuf:protobuf_py_pb2\",\n \"//moonlight/staves:base\",\n \"//moonlight/structure\",\n \"//moonlight/structure:beams\",\n \"//moonlight/structure:components\",\n \"//moonlight/structure:verticals\",\n # numpy dep\n # six dep\n # tensorflow dep\n ],\n)\n\n# The omr CLI for running locally on a single score.\npy_binary(\n name = \"omr\",\n srcs = [\"omr.py\"],\n srcs_version = \"PY2AND3\",\n deps = [\n \":engine\",\n # disable_tf2\n \"@com_google_protobuf//:protobuf_python\",\n # absl dep\n \"//moonlight/conversions\",\n \"//moonlight/glyphs:saved_classifier_fn\",\n # tensorflow dep\n ],\n)\n\npy_test(\n name = \"omr_endtoend_test\",\n size = \"large\",\n srcs = [\"omr_endtoend_test.py\"],\n data = [\"//moonlight/testdata:images\"],\n shard_count = 4,\n srcs_version = \"PY2AND3\",\n deps = [\n \":engine\",\n # disable_tf2\n # pillow dep\n # absl/testing dep\n # librosa dep\n # lxml dep\n \"//moonlight/conversions\",\n \"@magenta//protobuf:music_py_pb2\",\n # numpy dep\n # tensorflow.python.platform dep\n ],\n)\n\npy_test(\n name = \"omr_regression_test\",\n size = \"large\",\n srcs = [\"omr_regression_test.py\"],\n args = [\"--corpus_dir=../omr_regression_test_data\"],\n data = [\"@omr_regression_test_data\"],\n shard_count = 4,\n srcs_version = \"PY2AND3\",\n deps = [\n \":engine\",\n # disable_tf2\n # absl/testing dep\n \"//moonlight/protobuf:protobuf_py_pb2\",\n \"//moonlight/score:reader\",\n ],\n)\n\npy_library(\n name = \"image\",\n srcs = [\"image.py\"],\n srcs_version = \"PY2AND3\",\n deps = [], # tensorflow dep\n)\n\npy_library(\n name = \"page_processors\",\n srcs = [\"page_processors.py\"],\n srcs_version = \"PY2AND3\",\n deps = [\n \"//moonlight/glyphs:glyph_types\",\n \"//moonlight/glyphs:note_dots\",\n \"//moonlight/glyphs:repeated\",\n \"//moonlight/staves:staff_processor\",\n \"//moonlight/structure:barlines\",\n \"//moonlight/structure:beam_processor\",\n \"//moonlight/structure:section_barlines\",\n \"//moonlight/structure:stems\",\n ],\n)\n\npy_library(\n name = \"score_processors\",\n srcs = [\"score_processors.py\"],\n srcs_version = \"PY2AND3\",\n deps = [\"//moonlight/score:reader\"],\n)\n" }, { "path": "moonlight/conversions/BUILD", "content": "# Description:\n# Format conversions for OMR.\npackage(\n default_visibility = [\"//moonlight:__subpackages__\"],\n)\n\nlicenses([\"notice\"]) # Apache 2.0\n\npy_library(\n name = \"conversions\",\n srcs = [\"__init__.py\"],\n deps = [\n \":musicxml\",\n \":notesequence\",\n ],\n)\n\npy_library(\n name = \"musicxml\",\n srcs = [\"musicxml.py\"],\n deps = [\n # librosa dep\n # lxml dep\n \"//moonlight/protobuf:protobuf_py_pb2\",\n \"//moonlight/score:measures\",\n ],\n)\n\npy_test(\n name = \"musicxml_test\",\n srcs = [\"musicxml_test.py\"],\n deps = [\n \":musicxml\",\n # absl/testing dep\n \"//moonlight/protobuf:protobuf_py_pb2\",\n \"@magenta//protobuf:music_py_pb2\",\n ],\n)\n\npy_library(\n name = \"notesequence\",\n srcs = [\"notesequence.py\"],\n deps = [\n \"//moonlight/protobuf:protobuf_py_pb2\",\n \"@magenta//protobuf:music_py_pb2\",\n ],\n)\n" }, { "path": "moonlight/conversions/__init__.py", "content": "# Copyright 2018 Google LLC\n#\n# Licensed under the Apache License, Version 2.0 (the \"License\");\n# you may not use this file except in compliance with the License.\n# You may obtain a copy of the License at\n#\n# https://www.apache.org/licenses/LICENSE-2.0\n#\n# Unless required by applicable law or agreed to in writing, software\n# distributed under the License is distributed on an \"AS IS\" BASIS,\n# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n# See the License for the specific language governing permissions and\n# limitations under the License.\n\"\"\"OMR format conversions.\"\"\"\n# TODO(ringw): Score to MusicXML, preserving staves, etc.\n\nfrom __future__ import absolute_import\nfrom __future__ import division\nfrom __future__ import print_function\n\nfrom moonlight.conversions.musicxml import score_to_musicxml\nfrom moonlight.conversions.notesequence import page_to_notesequence\nfrom moonlight.conversions.notesequence import score_to_notesequence\n" }, { "path": "moonlight/conversions/musicxml.py", "content": "# Copyright 2018 Google LLC\n#\n# Licensed under the Apache License, Version 2.0 (the \"License\");\n# you may not use this file except in compliance with the License.\n# You may obtain a copy of the License at\n#\n# https://www.apache.org/licenses/LICENSE-2.0\n#\n# Unless required by applicable law or agreed to in writing, software\n# distributed under the License is distributed on an \"AS IS\" BASIS,\n# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n# See the License for the specific language governing permissions and\n# limitations under the License.\n\"\"\"Score to MusicXML conversion.\"\"\"\n# TODO(ringw): Key signature\n# TODO(ringw): Chords\n# TODO(ringw): Stems--MusicXML supports \"up\" or \"down\".\n# TODO(ringw): Accurate layout of pages, staves, and measures.\n# TODO(ringw): Barline types.\n\nfrom __future__ import absolute_import\nfrom __future__ import division\nfrom __future__ import print_function\n\nimport copy\nimport re\n\nimport librosa\nfrom lxml import etree\nfrom moonlight.protobuf import musicscore_pb2\nfrom moonlight.score import measures\nfrom six import moves\n\nDOCTYPE = ('\\n')\nMUSICXML_VERSION = '3.0'\n\n# Number of divisions (duration units) per quarter note.\nDIVISIONS = 1024\n\n# TODO(ringw): Detect the actual time signature.\nTIME_SIGNATURE = etree.Element('time', symbol='common')\netree.SubElement(TIME_SIGNATURE, 'beats').text = '4'\netree.SubElement(TIME_SIGNATURE, 'beat-type').text = '4'\n\n# Note types.\nHALF = 'half'\nWHOLE = 'whole'\n# Indexed by the number of beams on a filled note. Each beam halves the duration\n# of the note.\nFILLED = [\n 'quarter', 'eighth', '16th', '32nd', '64th', '128th', '256th', '512th',\n '1024th'\n]\n\n# Maps the ASCII accidental names to the actual pitch alteration.\nACCIDENTAL_TO_ALTER = {'': 0, '#': 1, 'b': -1}\n\n\ndef score_to_musicxml(score):\n \"\"\"Converts a `tensorflow.moonlight.Score` to MusicXML.\n\n Args:\n score: The OMR score.\n\n Returns:\n XML text.\n \"\"\"\n musicxml = MusicXMLScore(score)\n measure_num = 0\n previous_note_start_time = 0\n previous_note_end_time = 0\n\n for page in score.page:\n for system in page.system:\n system_measures = measures.Measures(system)\n for system_measure_num in moves.xrange(system_measures.size()):\n for staff_num, staff in enumerate(system.staff):\n # Produce the measure, even if there are no glyphs.\n measure = musicxml.get_measure(staff_num, measure_num)\n\n for glyph in staff.glyph:\n if system_measures.get_measure(glyph) == system_measure_num:\n clef = _glyph_to_clef(glyph)\n if clef is not None:\n attributes = _get_attributes(measure)\n if attributes.find('clef') is not None:\n attributes.remove(attributes.find('clef'))\n attributes.append(clef)\n note = _glyph_to_note(glyph)\n if note is not None:\n if (glyph.note.start_time == previous_note_start_time and\n glyph.note.end_time == previous_note_end_time):\n position = note.index(note.find('pitch'))\n chord = etree.Element('chord')\n note.insert(position, chord)\n previous_note_start_time = glyph.note.start_time\n previous_note_end_time = glyph.note.end_time\n measure.append(note)\n measure_num += 1\n # Add and