fix: TTS voices — retry 10s, message on failure, refresh button

- Retry loop extended to 10s (50×200ms) for slow Android WebViews
- Show 'Nessuna voce disponibile' after timeout instead of infinite loader
- Show 'Voce non supportata dal browser' if speechSynthesis missing
- Reset to loading state on each settings open (fixes stale empty select)
- Add refresh button ↺ to force-reload voices manually

assets/js/app.js

@@ -9279,7 +9279,15 @@ function onTtsEngineChange(engine) {
 /** Populate voice selector from Web Speech API. Called on settings load and on voiceschanged. */
 function _initBrowserTtsVoices(selectedVoice) {
     const sel = document.getElementById('setting-tts-voice');
-    if (!sel || !window.speechSynthesis) return;
+    if (!sel) return;
+
+    if (!window.speechSynthesis) {
+        sel.innerHTML = '<option value="">— Voce non supportata dal browser —</option>';
+        return;
+    }
+
+    // Reset to loading state each time (settings page may be re-opened)
+    sel.innerHTML = '<option value="">— Caricamento voci… —</option>';
 
     const populate = () => {
         const voices = window.speechSynthesis.getVoices();
@@ -9291,7 +9299,7 @@ function _initBrowserTtsVoices(selectedVoice) {
         sel.innerHTML = sorted.map(v =>
             `<option value="${v.name}" ${v.name === selectedVoice ? 'selected' : ''}>${v.name} (${v.lang})${v.localService ? '' : ' ☁️'}</option>`
         ).join('');
-        // Auto-select Paola if no preference and it exists
+        // Auto-select first Italian voice if no preference set
         if (!selectedVoice) {
             const paola = sorted.find(v => v.name === 'Paola');
             const firstIt = sorted.find(v => v.lang.startsWith('it'));
@@ -9301,23 +9309,26 @@ function _initBrowserTtsVoices(selectedVoice) {
         return true;
     };
 
-    // Some browsers (Chrome) load voices async: onvoiceschanged fires once they're ready.
-    // But if voices are already loaded, onvoiceschanged never fires → try immediately too.
-    // Additionally, onvoiceschanged may have already fired before we assign the handler
-    // (race condition), so we also poll with short retries as fallback.
-    if (!populate()) {
-        // Register the event handler for browsers that fire it
-        if (window.speechSynthesis.onvoiceschanged !== undefined) {
+    // Try immediately (voices already cached from previous call)
+    if (populate()) return;
+
+    // onvoiceschanged fires in Firefox / some Chrome versions
     window.speechSynthesis.onvoiceschanged = () => { populate(); };
-        }
-        // Fallback retry loop: try every 200ms for up to 4 seconds
+
+    // Polling fallback: Chrome/WebView loads voices async (up to ~3s on desktop, longer on Android)
     let tries = 0;
     const interval = setInterval(() => {
         tries++;
-            if (populate() || tries >= 20) clearInterval(interval);
-        }, 200);
+        if (populate()) {
+            clearInterval(interval);
+        } else if (tries >= 50) { // 50 × 200ms = 10s
+            clearInterval(interval);
+            if (!window.speechSynthesis.getVoices().length) {
+                sel.innerHTML = '<option value="">— Nessuna voce disponibile su questo dispositivo —</option>';
             }
         }
+    }, 200);
+}
 
 /** Speak text using the browser Web Speech API (offline). */
 function _speakBrowser(text) {

index.html

@@ -930,10 +930,13 @@
                     <div id="tts-browser-section">
                         <div class="form-group">
                             <label>🗣️ Voce</label>
-                            <select id="setting-tts-voice" class="form-input">
+                            <div style="display:flex;gap:8px;align-items:center">
+                                <select id="setting-tts-voice" class="form-input" style="flex:1">
                                     <option value="">— Caricamento voci… —</option>
                                 </select>
-                            <p class="settings-hint">Le voci disponibili dipendono dal sistema operativo e dal browser. Su macOS/iOS è disponibile la voce <strong>Paola</strong> (italiano).</p>
+                                <button type="button" class="btn btn-secondary" style="padding:8px 12px;white-space:nowrap;flex-shrink:0" onclick="_initBrowserTtsVoices(document.getElementById('setting-tts-voice').value)">↺</button>
+                            </div>
+                            <p class="settings-hint">Le voci disponibili dipendono dal sistema operativo e dal browser. Su macOS/iOS è disponibile la voce <strong>Paola</strong> (italiano). Premi ↺ se la lista non si carica.</p>
                         </div>
                         <div class="form-group">
                             <label>⚡ Velocità: <span id="tts-rate-label">1.0</span>×</label>