Yass Reloaded icon
Yass Reloaded Documentation
GitHub Wiki Releases

UltraStar Song Editor

Yass-Reloaded

Yass Reloaded 2026.7 Release Notes

Karaoke Editor

Yass Reloaded 2026.7 Release Notes

These release notes summarize the user-facing changes planned for 2026.7_release.

Highlights

  • New Pitch Shift workflow for songs that are tuned slightly too high or too low
  • New Set Pitch From Microphone action to set note heights by singing or humming
  • Better wizard subtitle and transcript handling, especially for YouTube-based song creation
  • More reliable Align to Melody and transcript-based note rebuilding
  • New Remove Reverb (MVSEP) helper for existing #VOCALS tracks
  • Better editor shortcuts on macOS and a setting for repeated Shift-Up / Shift-Down selection behavior
  • Several UI, Preferences, USDB compare, autosave, and localization fixes

Pitch Shift

  • New Pitch Shift… dialog in the editor’s Extras menu.
    • Open a song, choose Extras -> Pitch Shift..., then either analyze the tuning or enter a manual correction in cents.
    • Aubio is required for automatic analysis; manual corrections still work without Aubio.
    • Small offsets below 5 cents are reported as “no correction needed”.
  • Pitch correction is stored per song.
    • Yass writes the correction into the song’s #COMMENT data as pitchShiftCents.
    • Playback uses a temporary corrected audio file, so the original files are not changed just by trying the feature.
    • When a non-zero correction is active, a small cents field appears in the editor toolbar for quick adjustments.
  • The correction can also be made permanent.
    • Use Render to Files… in the Pitch Shift dialog.
    • FFmpeg is required for rendering.
    • Yass creates new shifted audio files, keeps the originals, updates the song to point to the new files, and removes the stored live correction so it is not applied twice.
  • Pitch lines and melody alignment now stay in sync with pitch-shifted playback.
    • Previously, the detected pitch view could apply the correction twice in some cases.

Set Pitch From Microphone

  • New Set Pitch From Microphone action in the editor.
    • Configure a microphone first in Extras -> Preferences... -> Editor -> Control.
    • If the microphone is too quiet or too sensitive, adjust Mic sensitivity on the same settings page.
    • Devices are detected at startup, so reconnecting a microphone may require restarting Yass Reloaded before it appears.
    • If no microphone is configured yet, the on-screen hint now names this path and can be dismissed with Esc.
  • To use it:
    • select one note or several notes
    • press Ctrl-Shift-M or choose the menu action
    • sing or hum the intended pitch
    • Yass applies the detected pitch to the selected note(s)
  • While listening:
    • Up / Down shifts the detection window by octave
    • selecting another note accepts the current pitch and continues on the new selection
    • pressing Ctrl-Shift-M again or Esc accepts and turns listening off
    • use undo afterwards if you want to discard the accepted pitch
  • Captured microphone audio is not stored. The feature only uses the detected pitch.

Wizard, Subtitles, And Lyrics

  • The Create Song Wizard can now use YouTube subtitles more reliably as transcript input.
    • Start from a YouTube URL as usual.
    • If subtitles are downloaded, Yass can keep the selected subtitle file for the final song creation step.
    • Manual subtitles are preferred over auto-generated subtitles when both are available.
  • Subtitle data can be used as a fallback transcript.
    • This helps when no full OpenAI or WhisperX transcription was run.
    • The fallback transcript can still be saved as yass-transcript.json for later reuse.
  • Subtitle, LRC, and LRCLib timing sources are handled more consistently.
    • All-caps lyric subtitles are normalized more cleanly.
    • Coarse line timings can be refined against separated vocals when available.
    • Wizard output can use the same transcript artifact model that editor transcription workflows already use.
  • The wizard logs subtitle and transcript decisions more clearly, which should make missing transcript artifacts easier to diagnose.

Align To Melody And Transcript Rebuilds

  • Align to Melody and transcript-based note rebuilding are more vocal-aware.
    • This especially helps lyrics coming from subtitles, local .lrc files, LRCLib, OpenAI, or WhisperX.
    • Yass can use vocal pitch and energy data to place words closer to the actually sung timing.
  • Better handling of song starts and page boundaries.
    • Songs with #GAP:0 can scan for the first real vocal entry instead of anchoring to intro dialog or bleed.
    • The first note after a page break can use a nearby unused vocal pitch island instead of leaving that sung bit unassigned.
  • Better handling of sustained and melismatic vocals.
    • Notes can be extended when the vocal energy clearly continues.
    • Extra stable pitch runs can be reclaimed instead of being dropped.
    • Long sung blocks are less likely to steal timing from the following words.
  • Manual editor alignment stays more predictable.
    • When you align an existing selected note, the pitch decision is based on that note’s original range rather than being pulled by a nearby lead-in beat.

MVSEP And Audio Tools

  • New Remove Reverb (MVSEP) action.
    • Configure your MVSEP API token first.
    • Open a song that already has a usable #VOCALS file.
    • Choose Extras -> Remove Reverb (MVSEP).
    • Yass uploads the current vocals track, downloads the de-reverbed result into the song folder, and asks whether it should become the new #VOCALS file.
  • Reverb Removal is kept separate from the normal MVSEP stem-separation model list, so it no longer appears as if it were a normal separation preset.
  • The reverb-removal status dialog supports progress output and cancellation in the same style as other separation actions.

Editor Shortcuts And Selection

  • macOS has dedicated editor shortcut mappings for arrow-key editing.
    • This avoids several combinations that macOS normally intercepts for Mission Control, Spaces, or window navigation.
    • The GitHub Pages shortcut reference now lists Windows/Linux and macOS variants side by side.
  • Repeated Shift-Up / Shift-Down selection behavior is configurable.
    • With the option enabled, repeated presses can extend the selection to the current word or page boundary.
    • With the option disabled, Shift-Up and Shift-Down stay note-by-note.
    • Configure this in Extras -> Preferences... -> Editor -> Keyboard.
  • The shortcut reference was updated for current note movement, resizing, alignment, microphone pitch, and lyrics actions.

USDB And Library Workflows

  • USDB compare/edit now detects solo-vs-duet mismatches.
    • If the local song and the selected USDB song are different solo/duet variants, Yass no longer tries to compare incompatible text.
    • Forced mismatches show a clear message.
    • Automatic compare can fall back to search so the correct variant can be selected.
  • The USDB Syncer metadata dialog is easier to use for long fields.
    • It is scrollable/resizable enough to show rows such as Video URL.
  • Background image refreshes are more consistent after metadata changes.

Preferences, Autosave, And UI Fixes

  • French and Spanish Preferences no longer crash when opening Extras -> Preferences.
    • Fixes GitHub issue #51.
    • The issue was a duplicated translated language-list key that overwrote the internal language choices.
  • Autosave is less intrusive during active editing.
    • The backup write path no longer performs a verify-reload step that could interrupt editing with a dialog.
    • The autosave interval is still configured in Extras -> Preferences... -> Editor -> Control.
    • 0 disables autosave; the default remains 300 seconds.
  • Tag chips in the song header look cleaner.
    • The remove button now stays next to the tag label instead of drifting far away.
  • The absolute pitch view recenters more sensibly after transcript rebuilds.
  • The transcript overwrite prompt no longer mentions WhisperX when the transcript came from another source.
  • New translated UI text was added for pitch shift, microphone pitch capture, mic sensitivity, MVSEP reverb removal, and related prompts.

Documentation

  • GitHub Pages now includes a dedicated Pitch Shift page.
  • The keyboard shortcut page now documents macOS-specific editor shortcuts.
  • The settings, editor, wizard, and tools pages were updated for:
    • Set Pitch From Microphone
    • mic sensitivity
    • autosave interval
    • repeated Shift-Up / Shift-Down selection behavior
    • MVSEP Reverb Removal
    • YouTube subtitle transcript fallback