Merge branch 'codex/review-1861' into codex/pr-queue-integration

# Conflicts:
#	CHANGELOG.md

CHANGELOG.md

@@ -16,6 +16,7 @@ the frozen-backend fallback mirror it for their toolchains.
 - Tilde-separated number ranges are spoken clearly without running their endpoints together (#1821) — thanks @flutterkage2k!
 - Voice modes use themed tabs, with Synthesize and Convert pinned below their scrolling forms (#1823)
 - Voice cloning now starts with a clear upload-or-record choice, reveals recording and reference details only when needed, and keeps sampling controls under Production Overrides (#1817)
+- The first-run welcome line uses an instruction accepted by OmniVoice and VoiceDesign engines (#1861) — thanks @psiberfunk!
 
 ### Changed
 
@@ -67,6 +68,8 @@ the frozen-backend fallback mirror it for their toolchains.
 - The notification count stays inside the title bar instead of clipping above the bell (#1823)
 - The workspace engine menu opens beside its button instead of at the opposite edge of the page (#1823)
 - Cloning reuses the dubbing language picker with flags, search, and single selection, opening above the pinned synthesis controls (#1823)
+- The first-run welcome line uses an instruction accepted by OmniVoice and VoiceDesign engines (#1861) — thanks @psiberfunk!
+
 
 ## [0.5.2] — 2026-09-02
 

frontend/src/App.jsx

@@ -1,3 +1,4 @@
+import { firstSoundRequest } from './utils/firstSound';
 import React, {
   useState,
   useRef,
@@ -699,16 +700,10 @@ function App() {
     if (!pending) return;
     (async () => {
       try {
-        const fd = new FormData();
-        fd.append('text', i18n.t('firstrun.first_sound_text'));
-        // Functional model prompt (not user-facing copy) — keeps the demo
-        // voice warm without depending on seeded profiles.
-        fd.append('instruct', 'A warm, friendly narrator voice, medium pace');
-        fd.append('num_step', '16');
-        const res = await apiFetch(`${API}/generate`, {
-          method: 'POST',
-          body: fd,
-        });
+        const res = await apiFetch(
+          `${API}/generate`,
+          firstSoundRequest(i18n.t('firstrun.first_sound_text')),
+        );
         const blob = await res.blob();
         await playBlobAudio(blob, { label: i18n.t('player.generated_audio') });
         toast.success(i18n.t('firstrun.first_sound_done'), { duration: 7000 });

frontend/src/test/firstSoundInstruct.test.js

@@ -0,0 +1,15 @@
+import { describe, it, expect } from 'vitest';
+import { firstSoundRequest } from '../utils/firstSound';
+
+describe('onboarding first-sound request', () => {
+  it('posts localized text and a nonempty voice-design instruction as multipart form data', () => {
+    const request = firstSoundRequest('Welcome to VoiceStudio');
+    expect(request.method).toBe('POST');
+    expect(request.body).toBeInstanceOf(FormData);
+    expect(Object.fromEntries(request.body)).toEqual({
+      text: 'Welcome to VoiceStudio',
+      instruct: 'middle-aged, low pitch',
+      num_step: '16',
+    });
+  });
+});

frontend/src/utils/firstSound.js

@@ -0,0 +1,10 @@
+import defaults from './firstSound.json';
+
+// Taxonomy tokens also provide the nonempty description VoiceDesign requires.
+export function firstSoundRequest(text) {
+  const body = new FormData();
+  body.append('text', text);
+  body.append('instruct', defaults.instruct);
+  body.append('num_step', String(defaults.num_step));
+  return { method: 'POST', body };
+}

frontend/src/utils/firstSound.json

@@ -0,0 +1,4 @@
+{
+  "instruct": "middle-aged, low pitch",
+  "num_step": 16
+}

tests/test_first_sound_instruct.py

@@ -0,0 +1,16 @@
+"""The actual onboarding defaults must pass the runtime engine validator."""
+import json
+from pathlib import Path
+
+from omnivoice.models.omnivoice import _resolve_instruct
+
+
+def test_first_sound_instruction_passes_runtime_taxonomy_validation():
+    defaults = json.loads(
+        (Path(__file__).parents[1] / "frontend/src/utils/firstSound.json").read_text()
+    )
+    instruct = defaults["instruct"]
+    # VoiceDesign requires a nonempty description; OmniVoice validates tokens.
+    assert instruct.strip()
+    assert _resolve_instruct(instruct) == instruct
+    assert _resolve_instruct(instruct, use_zh=True)