Skip to content

0.1.575 Beta

Pre-release
Pre-release

Choose a tag to compare

View all tags
@github-actions github-actions released this 17 Oct 11:14
· 5 commits to master since this release
0.1.575-beta
fd8c843
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

New

  • Warning center — a warning icon appears next to Help whenever something is wrong; open it for a list of active warnings with details, dismiss one at a time or all at once. Localized, and any part of OpenUtau can raise one. First warning: duplicate voicebank folder names under different singer roots, where only the first was loaded silently (#2152). (4c955fb9)
  • DiffSinger group-aware duration models — these models now receive the word_div and word_dur inputs they need, chosen by dur_type in dsconfig.yaml (rel = group-aware, abs = legacy absolute). Voicebanks exported before the field existed stay on the absolute path, and a graph probe covers the attn voicebanks exported in between, so nothing that worked before stops working. (fd8c843e, #2451)

UI

  • Light theme borders lightened (#707070 → #C4C4C4); text boxes, checkboxes and toggles now follow the theme's border colours (c0c7bd55, 67dbf355)
  • Secondary text is readable in dark theme — uses ForegroundBrushSecondary instead of a border-mapped brush (c0c7bd55)
  • Hidden find bar no longer keeps reselecting notes on part load (8fdb4e9b, #2469)
  • WORLDLINE-R1.1 is in the default renderer list (e92d5e56)
  • Right-click deselects in the Pitch Point tool (87e2b6bb, #2491)

Fixes

Rendering

  • Playback and pre-render share one render lane — no more double rendering or progress-bar fighting (dc9270f8, #2518)
  • DiffSinger: missing vocoder falls back to the universal pc-nsf-hifigan (which gains SHFC) and offers to install it, instead of a stack trace (99f2f25f, #2508)
  • Transcription: GAME and RMVPE are offered in the missing-package prompt, then the transcription retries (99f2f25f, #2508)
  • VOICEVOX waveforms line up with the synthesized audio — head/tail pau counted, bad positionMs override removed, durations relative to the note so tempo changes don't skew them (0a7cde87, #2509)
  • Fixed NullReferenceException copying voicebank files for built-in resamplers (5c3c83bb)

Phonemizers

  • English C+V: fixed the VC alias condition and added contextual fallbacks, with tests (8879fbed #2517, 7b6cae8b #2527)
  • Arpasing+ / EN X-Sampa: fixed vv diphthongs (5714d927, #2531)
  • Thai VCCV: reverted the #2461 refactor (fa2a670c, #2516)
  • DiffSinger Filipino G2P: fixed a rule error (13184085, #2523)

Import

  • SVP: imported audio tracks are loaded properly — peaks exist, length comes from the SVP's declared duration, and tracks with missing files import as missing instead of vanishing (4ef547c6)

See Getting Started for how to use.

If you encountered issues, see FAQ

macOS says "This app is damaged"

Open a terminal and run xattr -rc /Applications/OpenUtau.app. Try opening OpenUtau again.

Assets 20
Loading