ChordFlask v0.9.25

ChordFlask v0.9.25

Standalone and Debian reliability fixes

Bug Fixes

• Restored standalone and Debian chordflask-analyze for core Chordino/QM
  analysis using the bundled runtime, without a source checkout or development
  virtual environment.
• Fixed system FFmpeg selection in the frozen runtime and library-path
  conflicts when launching external FFmpeg, including MP4 conversion.
• Missing Vamp plugins now produce clearer installation, verification and
  retry instructions appropriate to the installation type.
• GUI analysis errors stay visible for the selected song and show the failure
  details and retry instructions.

Verification

• Added packaging, integration and regression tests for bundled CLI analysis,
  plugin diagnostics, GUI failures and frozen FFmpeg conversion.
• A Mint 22 / Python 3.12 DEB was practically verified on a separate Lubuntu
  26.04.1 LTS x86_64 system: MP4 conversion, Chordino/QM analysis and
  JSON/MIDI/MusicXML output succeeded with the normal CLI invocation. GUI
  analysis succeeded after a full process restart.

Known Limitations

• System FFmpeg remains required. Vamp plugins are not bundled or installed
  automatically; run the supplied plugin installer explicitly. Other optional
  helpers still require their documented external runtimes or source setup.
  Existing analyses remain displayable without optional generation runtimes.
• The Lubuntu result covers one tested installation, not arbitrary Linux
  distributions. An earlier intermittent GUI-worker tqdm._lock error has
  no proven root cause and is not claimed as independently fixed.
• Score remains Alpha / experimental, with the notation-quality limitations
  described in the Score documentation.

ChordFlask v0.9.24

ChordFlask v0.9.24

New Features

• Score (Alpha / experimental): generate piano transcriptions with Transkun
  and display the derived two-staff notation on desktop (1024 px and wider).
  The Grid | Lyrics | Score switch offers Score when valid prepared data
  exist. Measures follow playback, seeking and loops, with automatic page
  selection and scrolling when the notation line changes.
• Prepare → Score generates a score for the current song when the separate
  runtime is available. The chordflask-score CLI also supports non-recursive
  directory batches, --dry-run, CPU/CUDA selection and existing Demucs
  no-vocals stems. Generation accepts MP3/MP4 and requires existing QM rhythm
  analysis; it leaves chord and rhythm tracks unchanged.
• Score MIDI exports Transkun's original detected notes from the browser
  Export menu or with chordflask-export --format score-midi. All formats
  includes it when valid Score data are available. The existing accompaniment
  export is labeled Chord MIDI in the browser and remains --format midi
  in the CLI. Score MIDI export never runs transcription and is unaffected by
  chord edits or display transposition.

Improvements

• Optional CLI notation quantization: --quantization binary uses sixteenths
  only; the default mixed allows sixteenths and eighth-note triplets. Binary
  can reduce unwanted tuplets and rests but shifts genuine triplets. Changing
  the grid requires --replace; previous generations are retained. Original
  MIDI and QM playback timing are unchanged.
• Source, portable standalone and Debian players can display valid prepared
  scores and export Score MIDI without the generation runtime. Generation
  requires a separate Python 3.12 runtime, installed from a source checkout
  with make setup-score; make score-check checks its readiness. Transkun,
  Torch and model weights are not bundled with the player.
• Added optional Python 3.12 installation instructions using uv for Score and
  Lyrics V2/Qwen3, including interpreter selection and existing runtime
  environments. ChordFlask does not install Python or uv automatically.

Known Limitations

• Score remains experimental. Generated notation is often rhythmically
  cluttered and is not reliably readable or playable as a piano score.
  Overlapping notes can create many voices and rests; mixed straight/triplet
  rhythms have known notation limitations. Convincing MIDI playback does not
  establish notation quality.
• Notation requires one fixed meter and covers only the first through last
  measured QM beat. Notes outside that interval are omitted or clipped in
  notation; original MIDI is preserved. Hand separation uses a simple pitch
  split. Meter changes are unsupported.
• Score display and GUI preparation are desktop-only. Device, audio-source and
  quantization choices are CLI-only. CUDA operation and long-file performance
  have not been evaluated.

ChordFlask v0.9.23

ChordFlask v0.9.23

New Features

• Export beat-synchronized MIDI from the browser Export menu or with
  chordflask-export --format midi. The piano chord sequence uses the selected
  track, edits and transposition, with a tempo map matching analyzed beat times.
  MIDI is also included in All formats exports.

Improvements

• Updated README and Vamp documentation with separate installation and
  verification commands for the Debian package, portable standalone and source
  installation. Vamp plugins still need to be installed separately.
• Clarified the human-directed, AI-assisted development methodology and the
  maintainer's responsibility for evaluating and accepting results.

Bug Fixes

• Missing-Vamp messages now point to the installer for the running distribution:
  /opt/chordflask/install_vamp.sh for Debian, the bundled install_vamp.sh for
  portable standalone, and make plugins for source installations.
• The Debian package now requires curl | wget alongside FFmpeg for the Vamp
  installer.
• Updated the source-publication transparency check to accept the revised
  development description while still requiring disclosure of AI assistance
  and maintainer responsibility.

Known Limitations

• MIDI exports are chord sequences with fixed piano voicings, not full music
  transcriptions. Unknown chord labels remain silent; the result depends on the
  selected chord and beat analysis.

ChordFlask v0.9.22

ChordFlask v0.9.22

New Features

• Experimental Lyrics V2: generate lyrics locally from audio with
  chordflask-genlyrics --analyzer qwen3. Qwen3-ASR recognizes the words and
  Qwen3-ForcedAligner synchronizes them. Setup downloads the models into a
  separate optional runtime; generation then works offline.
• Keep Lyrics V1 (song.cho) and V2 (song.v2.cho) side by side. Desktop Lyrics
  lets you choose between them when both exist, and browser export uses the
  selected version. Existing valid Demucs vocals can be reused for V2 generation.
• Remove a song's registered Demucs stems and its owned stem files with
  chordflask-maintain stems remove song.mp3. Use --dry-run to preview the
  removals.
• Maintenance validation and storage reports now accept a single media file.
  Schema migration also supports a non-writing --dry-run preview.

Improvements

• On narrow screens, the chord grid now prefers to keep the playing row in the
  second visible row, retaining the preceding row for context. Short viewports
  and song boundaries still prioritize keeping the playing row visible.
• Analysis, Demucs and Lyrics dry-run summaries distinguish planned work from
  completed work. Analysis and export commands print their output paths; export
  also reports when it created missing analysis or completed only some formats.
• Lyrics generation accepts --replace; --force remains an alias. Replacement
  affects only the selected lyrics version. Missing-analysis messages in Lyrics
  and BTC point to chordflask-analyze.
• Maintenance reports flag analysis files without matching media for review.
  Ambiguous media filenames, cleanup refusals and permission errors have clearer
  guidance. Installation documentation clarifies which CLI helpers require a
  source-installed virtual environment.

Bug Fixes

• Adjusted desktop Firefox STEMS playback to avoid periodic speed and seek
  corrections associated with dropouts in local testing. Initial alignment and
  explicit seeking remain available.
• Lyrics retries now reject explicitly conflicting options instead of silently
  ignoring them in favor of saved generation settings.
• Lyrics generation preserves a target created while generation is running
  unless replacement was explicitly requested, and rejects symlink outputs.

Known Limitations

• Lyrics V2 transcription and singing alignment are experimental and can fail
  on some recordings. Unsupported alignment languages, including Thai, produce
  no V2 file. Failed generation leaves existing lyrics intact.
• V2 generation is CLI-only; desktop Prepare still generates V1. It needs the
  external Lyrics helper and Qwen runtime, Python 3.10–3.13 (3.12 recommended),
  and model downloads during setup. Python 3.14 is not supported by this setup.
  CPU inference can be slow and memory intensive; CUDA validation remains
  pending. Models and Torch are not bundled with the player.
• Desktop Chromium-based browsers remain recommended for STEMS playback.
  Firefox can still experience audible dropouts. Narrow layouts remain
  Grid-only, without Lyrics or Prepare controls.
• A single-media storage report covers direct .chordflask artifacts and the
  song's journal; it excludes Demucs stem directories and adjacent lyrics files.

ChordFlask v0.9.21

Full Changelog: v0.9.20...v0.9.21

ChordFlask v0.9.20

Full Changelog: v0.9.19...v0.9.20

ChordFlask v0.9.19

Full Changelog: v0.9.18...v0.9.19

ChordFlask v0.9.18

Full Changelog: v0.9.17...v0.9.18

ChordFlask v0.9.17

What's changed

• Added a unified desktop Prepare workflow for optional Lyrics, BTC, and Demucs producers. Only available runtimes are shown.
• Reworked optional runtime integration so Lyrics, BTC, and Demucs keep their heavy dependencies outside the ChordFlask application and standalone bundle.
• Improved synchronized Lyrics alignment: timed lyric lines now retain their own playback ranges and chord ownership more clearly.
• Added isolated Lyrics runtime setup and diagnostics with make setup-lyrics and make lyrics-check.
• Expanded Lyrics source support and documentation for local LRC, embedded lyrics, and LRCLIB, including matching, alignment, romanization, troubleshooting, and privacy behavior.
• Added substantially more regression coverage for Lyrics generation, alignment, runtime setup, and media preparation.
• Reworked and shortened the README, moving detailed feature documentation into dedicated guides.

Full Changelog: v0.9.16...v0.9.17

ChordFlask v0.9.16

Full Changelog: v0.9.15...v0.9.16