Labyricorn/voicebox
1
0
Fork 0
mirror of https://github.com/jamiepine/voicebox.git synced 2026-09-20 23:30:38 -07:00
Code Issues Packages Projects Releases Wiki Activity

feat(generate): restore instruct toggle in floating generate box (Qwen CustomVoice)

Browse Source 
Before 0.4 every engine was a cloning model, so the instruct UI in the
floating generate box applied the same way everywhere. Commit 3187344
hid the instruct toggle because the mix of new engines landing in 0.4
made it unclear which ones honored the kwarg. With Qwen CustomVoice
now shipping as the only engine actually tuned for instruct-style
control, bring the button back — conditionally, and only for that
engine.

Changes:
  • FloatingGenerateBox: SlidersHorizontal toggle button appears left
    of Generate when the box is expanded AND engine is
    qwen_custom_voice. Clicking it reveals an additive instruct
    textarea below the main text field (not a modal swap like the old
    version). State persists across engine switches so the toggle
    remembers its last position.
  • GenerationForm: narrow the instruct FormField's conditional from
    `qwen || qwen_custom_voice` to just `qwen_custom_voice`.
  • useGenerationForm: narrow supportsInstruct for the same reason.
    Base Qwen3-TTS accepts the kwarg but the model itself doesn't honor
    it — only CustomVoice was trained for instruction-based style
    control.

Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>
This commit is contained in:
James Pine
2026-04-16 03:09:55 -07:00
co-authored by Claude Opus 4.6
parent c9e5c5d9a7
commit 106aec46a8
3 changed files with 91 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files
app/src/lib/hooks/useGenerationForm.ts
+3 -1
View File
@@ -136,7 +136,9 @@ export function useGenerationForm(options: UseGenerationFormOptions = {}) {
const hasModelSizes =
engine === 'qwen' || engine === 'qwen_custom_voice' || engine === 'tada';
const supportsInstruct = engine === 'qwen' || engine === 'qwen_custom_voice';
// Only Qwen CustomVoice actually honors the instruct kwarg at model level.
// Base Qwen3-TTS accepts the kwarg but ignores it.
const supportsInstruct = engine === 'qwen_custom_voice';
const effectsChain = options.getEffectsChain?.();
// This now returns immediately with status="generating"
const result = await generation.mutateAsync({