crisp_notation_cli 0.12.0

The crisp_notation command-line tool — inspect scores, convert MusicXML/MEI/kern/MIDI/MuseScore/GP/ABC, export LilyPond and braille, render SVG/PNG, and run OMR.

Changelog

0.12.0 - 2026-10-19

Built on crisp_notation_core 0.12.0: the LilyPond reader fixes of this round reach convert and render (419 of 433 Mutopia files agree with LilyPond's MIDI), and MIDI export plays measure-repeat bars.

0.11.0 - 2026-10-18

Built on crisp_notation_core 0.11.0: tighter, LilyPond-like spacing in multi-staff PDF and SVG (the first system alone is indented for the instrument names), endings, dynamics above vocal lines, stated beams over barlines without stem clashes, and the LilyPond, MusicXML and MEI reader fixes of this round.

0.10.0 - 2026-10-17

Built on crisp_notation_core 0.10.0: MusicXML beams as written, staves and systems spaced by their ink (voice-and-piano PDFs take about a third fewer pages), volta labels and open endings, and the LilyPond q, \skip and \incipit fixes.

• A line-broken instrument name is sized by its widest line.
• render <in> <out.pdf> --text-font <path.ttf|.otf> now applies to PDF output too: the layout measures text in the font and the PDF embeds it (with a ToUnicode map), so lyrics and titles outside Latin-1 print instead of turning into ?. Without --text-font a PDF still prints Times.
• render <in> <out.svg> --text-font <path.ttf|.otf> measures lyrics and markings in the given font (via crisp_notation_core's textWidthFromFont) instead of the wide generic estimate, and embeds it in the SVG as a second @font-face (suppressed by --no-embed-font, same as the engraving font).
• PDF output measures lyrics and text in Times, the face it prints, so songs set tighter (fewer systems) than the safe SVG estimate allows.

0.9.0 - 2026-10-17

Built on crisp_notation_core 0.9.0; convert writes a piano as one two-staff MusicXML part.

• render <voice.mei> <out.svg|pdf> --mensural engraves white mensural notation.

0.8.0 - 2026-10-17

Built on crisp_notation_core 0.8.0.

• --style <sheet.json> renders SVG and PDF in a house style.
• render <chant.gabc> <out.svg|pdf> --square engraves Gregorian chant in square notation (lines of --width spaces as SVG, pages as PDF).
• render <in> <out.pdf> writes paged vector PDF — A4 or --paper letter, --staff-pt points per staff space, the title on page one, every part of a multi-part score; --font names the music font to embed.

0.7.2 - 2026-10-16

Built on crisp_notation_core 0.7.2: render and convert read MusicXML <forward> (voices after a gap were placed too early in 60% of real files), keep both halves of cross-staff piano chords, align multi-staff barlines, and render draws braces, brackets and a grand staff's joined barlines in SVG, without a metronome mark for a playback-only tempo.

0.7.1 - 2026-10-13

Built on crisp_notation_core 0.7.1: render no longer doubles the closing barline of a piece that ends on a repeat (#19), stops the staff at an explicit closing barline, draws start repeats without a stray barline before them, and keeps note names on every wrapped line. convert reads MusicXML keys beyond seven sharps or flats.

0.7.0 - 2026-10-06

Built on crisp_notation_core 0.7.0, with fixes for GitHub issues #8–#14:

• render draws staccatissimo and quarter-tone accidentals (they had no codepoint) and metronome marks.
• render places slurs over mixed stems above, puts dynamics and touching hairpins on one line, and sits chord tremolos on the free stem.
• convert keeps multi-measure rests and their text, counting the covered bars once.
• A live test runs these through the compiled binary.

0.6.0 - 2026-10-06

Built on crisp_notation_core 0.6.0.

• LilyPond input: info, convert, timeline and render read .ly files (--from ly); render draws every staff of a multi-staff score. This uses core's LilyPond reader, which now matches LilyPond's own MIDI output on 123 of 374 Mutopia files, up from 13.
• MusicXML written by convert names its encoder (<software>crisp_notation</software>).
• Cross-staff beams in render slant with the figure and draw 16th-note and shorter beam levels.

0.5.0 - 2026-10-05

Built on crisp_notation_core 0.5.0. Every import, export and render picks up its fixes. Highlights for the CLI: MusicXML tempo text and metronome marks split across <direction-type> blocks are both kept, and timeline now repeats a section closed by an end repeat with no start repeat. Ties, slurs and beams in render follow standard engraving in chords and two-voice bars, and LilyPond and kern files with text over a rest render instead of failing.

• GABC (Gregorian chant) input: info, convert and render read .gabc files, via core's new clean-room reader.
• kern breve, long and maxima durations (0, 00, 000) are parsed.
• Live regression tests run each of these fixes through the compiled binary.

0.4.3 - 2026-07-15

• First pub.dev release. Install with dart pub global activate crisp_notation_cli to get the crisp_notation command (inspect, convert between MusicXML/MEI/kern/MIDI/MuseScore/GP/ABC with LilyPond & braille export, render to SVG/PNG, and OMR). Added a usage example. render ... .png needs the Flutter SDK; omr needs the native libcrispembed — SVG/convert/info are pure Dart.

0.4.1 - 2026-07-14

• Sheet title rendering: multi-staff render outputs include imported title/composer page text by default; pass --no-title to suppress it.

• Offline render for the standalone binary (hardening G1): the Bravura SMuFL metadata is now embedded in the CLI (DEFLATE-compressed + base64, ~136 KB, inflated at runtime via the existing pure-Dart inflate), so a released crisp_notation binary renders SVG without the repo checkout or a --metadata path. Resolution order is unchanged (--metadata → repo → embedded); when it falls back to the embedded copy there is no font file, so the SVG references the engraving font by name instead of inlining it. Regenerate the embed with tool/embed_metadata.dart.

• Braille music output: convert --to brl (or a .brl output path) writes Unicode braille-music notation via the new core scoreToBraille.

0.4.0 - 2026-07-13

First tagged release. The crisp_notation CLI: inspect scores; convert between MusicXML / .mxl / MEI / **kern / MIDI / MuseScore / the .gp tablature family / ABC (plus LilyPond export); render to SVG or PNG (notation or tab); and optical music recognition (omr) routing three CrispEmbed engines (Sheet Music Transformer, Polyphonic-TrOMR, Flova) through one command, with full-page segmentation (--page) and model auto-download by name. The OMR pipeline is also exposed as a reusable library (package:crisp_notation_cli/omr.dart).