fix SSE cleanup, filter add popover to completed only, derive isGenerating from pending set, handle not_found status

Generations now return immediately with a 'generating' status and appear
in history right away. TTS runs in a serial background queue to avoid
GPU contention. Users can kick off multiple generations without blocking.

- Async POST /generate creates DB record immediately, queues TTS work
- Serial generation queue prevents concurrent GPU access (Metal/CUDA/CPU)
- SSE endpoint GET /generate/{id}/status for real-time completion tracking
- Retry endpoint POST /generate/{id}/retry for failed generations
- Store engine and model_size on generation records for retry support
- History cards show animated loader (react-loaders) for generating/playing
- Failed generations show retry button instead of actions menu
- Model downloads happen inline in the queue instead of rejecting with 202
- Stale 'generating' records marked as failed on server startup
- Autoplay on generate setting (default: on)
- Show engine name on generation cards
- Remove sidebar generation spinner
- Checkbox alignment fix in settings

app/src/App.tsx

@@ -94,10 +94,11 @@ function App() {
 
     serverStartingRef.current = true;
     const isRemote = useServerStore.getState().mode === 'remote';
+    const customModelsDir = useServerStore.getState().customModelsDir;
     console.log(`Production mode: Starting bundled server... (remote: ${isRemote})`);
 
     platform.lifecycle
-      .startServer(isRemote)
+      .startServer(isRemote, customModelsDir)
       .then((serverUrl) => {
         console.log('Server is ready at:', serverUrl);
         // Update the server URL in the store with the dynamically assigned port

app/src/components/AudioPlayer/AudioPlayer.tsx

@@ -7,8 +7,8 @@ import { Slider } from '@/components/ui/slider';
 import { apiClient } from '@/lib/api/client';
 import { formatAudioDuration } from '@/lib/utils/audio';
 import { debug } from '@/lib/utils/debug';
-import { usePlayerStore } from '@/stores/playerStore';
 import { usePlatform } from '@/platform/PlatformContext';
+import { usePlayerStore } from '@/stores/playerStore';
 
 export function AudioPlayer() {
   const platform = usePlatform();
@@ -833,11 +833,7 @@ export function AudioPlayer() {
             className="shrink-0"
             title={duration === 0 && !isLoading ? 'Audio not loaded' : ''}
             aria-label={
-              duration === 0 && !isLoading
-                ? 'Audio not loaded'
-                : isPlaying
-                  ? 'Pause'
-                  : 'Play'
+              duration === 0 && !isLoading ? 'Audio not loaded' : isPlaying ? 'Pause' : 'Play'
             }
           >
             {isPlaying ? <Pause className="h-5 w-5" /> : <Play className="h-5 w-5" />}
@@ -872,7 +868,9 @@ export function AudioPlayer() {
 
           {/* Title */}
           {title && (
-            <div className="text-sm font-medium truncate max-w-[200px] shrink-0">{title}</div>
+            <div className="text-sm font-medium truncate max-w-[200px] shrink-0 hidden lg:block">
+              {title}
+            </div>
           )}
 
           {/* Loop Button */}
@@ -888,7 +886,11 @@ export function AudioPlayer() {
           </Button>
 
           {/* Volume Control */}
-          <div className="flex items-center gap-2 shrink-0 w-[120px]" role="group" aria-label="Volume">
+          <div
+            className="flex items-center gap-2 shrink-0 w-[120px]"
+            role="group"
+            aria-label="Volume"
+          >
             <Button
               variant="ghost"
               size="icon"

app/src/components/AudioTab/AudioTab.tsx

@@ -23,8 +23,8 @@ import {
 import { apiClient } from '@/lib/api/client';
 import { BOTTOM_SAFE_AREA_PADDING } from '@/lib/constants/ui';
 import { cn } from '@/lib/utils/cn';
-import { usePlayerStore } from '@/stores/playerStore';
 import { usePlatform } from '@/platform/PlatformContext';
+import { usePlayerStore } from '@/stores/playerStore';
 
 interface AudioDevice {
   id: string;
@@ -129,7 +129,7 @@ export function AudioTab() {
     if (await confirm('Delete this channel?')) {
       deleteChannel.mutate(channelId);
     }
-  }
+  };
 
   const allChannels = channels || [];
   const allDevices = devices || [];
@@ -168,7 +168,7 @@ export function AudioTab() {
               </Button>
             </div>
           ) : (
-            <div className="space-y-3 p-2">
+            <div className="space-y-3">
               {allChannels.map((channel) => {
                 const isSelected = selectedChannelId === channel.id;
                 return (
@@ -343,7 +343,9 @@ export function AudioTab() {
             <div className="flex flex-col items-center justify-center py-12 border-2 border-dashed border-muted rounded-md">
               <CheckCircle2 className="h-12 w-12 text-muted-foreground mb-4" />
               <p className="text-muted-foreground text-center">
-                {platform.metadata.isTauri ? 'No audio devices found' : 'Audio device selection requires Tauri'}
+                {platform.metadata.isTauri
+                  ? 'No audio devices found'
+                  : 'Audio device selection requires Tauri'}
               </p>
             </div>
           )}

app/src/components/Generation/FloatingGenerateBox.tsx

@@ -12,12 +12,12 @@ import {
   SelectValue,
 } from '@/components/ui/select';
 import { Textarea } from '@/components/ui/textarea';
-import { useToast } from '@/components/ui/use-toast';
 import { getLanguageOptionsForEngine, type LanguageCode } from '@/lib/constants/languages';
 import { useGenerationForm } from '@/lib/hooks/useGenerationForm';
 import { useProfile, useProfiles } from '@/lib/hooks/useProfiles';
-import { useAddStoryItem, useStory } from '@/lib/hooks/useStories';
+import { useStory } from '@/lib/hooks/useStories';
 import { cn } from '@/lib/utils/cn';
+import { useGenerationStore } from '@/stores/generationStore';
 import { useStoryStore } from '@/stores/storyStore';
 import { useUIStore } from '@/stores/uiStore';
 import { ParalinguisticInput } from './ParalinguisticInput';
@@ -44,8 +44,7 @@ export function FloatingGenerateBox({
   const selectedStoryId = useStoryStore((state) => state.selectedStoryId);
   const trackEditorHeight = useStoryStore((state) => state.trackEditorHeight);
   const { data: currentStory } = useStory(selectedStoryId);
-  const addStoryItem = useAddStoryItem();
-  const { toast } = useToast();
+  const addPendingStoryAdd = useGenerationStore((s) => s.addPendingStoryAdd);
 
   // Calculate if track editor is visible (on stories route with items)
   const hasTrackEditor = isStoriesRoute && currentStory && currentStory.items.length > 0;
@@ -53,25 +52,9 @@ export function FloatingGenerateBox({
   const { form, handleSubmit, isPending } = useGenerationForm({
     onSuccess: async (generationId) => {
       setIsExpanded(false);
-      // If on stories route and a story is selected, add generation to story
+      // Defer the story add until TTS completes — useGenerationProgress handles it
       if (isStoriesRoute && selectedStoryId && generationId) {
-        try {
-          await addStoryItem.mutateAsync({
-            storyId: selectedStoryId,
-            data: { generation_id: generationId },
-          });
-          toast({
-            title: 'Added to story',
-            description: `Generation added to "${currentStory?.name || 'story'}"`,
-          });
-        } catch (error) {
-          toast({
-            title: 'Failed to add to story',
-            description:
-              error instanceof Error ? error.message : 'Could not add generation to story',
-            variant: 'destructive',
-          });
-        }
+        addPendingStoryAdd(generationId, selectedStoryId);
       }
     },
   });
@@ -182,7 +165,7 @@ export function FloatingGenerateBox({
         isStoriesRoute
           ? // Position aligned with story list: after sidebar + padding, width 360px
             'left-[calc(5rem+2rem)] w-[360px]'
-          : 'left-[calc(5rem+2rem)] w-[calc((100%-5rem-4rem)/2-1rem)]',
+          : 'left-[calc(5rem+2rem)] right-8 lg:right-auto lg:w-[calc((100%-5rem-4rem)/2-1rem)]',
       )}
       style={{
         // On stories route: offset by track editor height when visible

app/src/components/History/HistoryTable.tsx

@@ -1,13 +1,15 @@
+import { useQueryClient } from '@tanstack/react-query';
 import {
-  AudioWaveform,
   Download,
   FileArchive,
   Loader2,
   MoreHorizontal,
   Play,
+  RotateCcw,
   Trash2,
 } from 'lucide-react';
 import { useEffect, useRef, useState } from 'react';
+import Loader from 'react-loaders';
 import { Button } from '@/components/ui/button';
 import {
   Dialog,
@@ -36,7 +38,8 @@ import {
   useImportGeneration,
 } from '@/lib/hooks/useHistory';
 import { cn } from '@/lib/utils/cn';
-import { formatDate, formatDuration } from '@/lib/utils/format';
+import { formatDate, formatDuration, formatEngineName } from '@/lib/utils/format';
+import { useGenerationStore } from '@/stores/generationStore';
 import { usePlayerStore } from '@/stores/playerStore';
 
 // OLD TABLE-BASED COMPONENT - REMOVED (can be found in git history)
@@ -54,9 +57,12 @@ export function HistoryTable() {
   const [importDialogOpen, setImportDialogOpen] = useState(false);
   const [selectedFile, setSelectedFile] = useState<File | null>(null);
   const [deleteDialogOpen, setDeleteDialogOpen] = useState(false);
-  const [generationToDelete, setGenerationToDelete] = useState<{ id: string; name: string } | null>(null);
+  const [generationToDelete, setGenerationToDelete] = useState<{ id: string; name: string } | null>(
+    null,
+  );
   const limit = 20;
   const { toast } = useToast();
+  const queryClient = useQueryClient();
 
   const {
     data: historyData,
@@ -71,6 +77,7 @@ export function HistoryTable() {
   const exportGeneration = useExportGeneration();
   const exportGenerationAudio = useExportGenerationAudio();
   const importGeneration = useImportGeneration();
+  const addPendingGeneration = useGenerationStore((state) => state.addPendingGeneration);
   const setAudioWithAutoPlay = usePlayerStore((state) => state.setAudioWithAutoPlay);
   const restartCurrentAudio = usePlayerStore((state) => state.restartCurrentAudio);
   const currentAudioId = usePlayerStore((state) => state.audioId);
@@ -194,6 +201,20 @@ export function HistoryTable() {
     }
   };
 
+  const handleRetry = async (generationId: string) => {
+    try {
+      const result = await apiClient.retryGeneration(generationId);
+      addPendingGeneration(result.id);
+      queryClient.invalidateQueries({ queryKey: ['history'] });
+    } catch (error) {
+      toast({
+        title: 'Retry failed',
+        description: error instanceof Error ? error.message : 'Could not retry generation',
+        variant: 'destructive',
+      });
+    }
+  };
+
   const handleImportConfirm = () => {
     if (selectedFile) {
       importGeneration.mutate(selectedFile, {
@@ -250,22 +271,30 @@ export function HistoryTable() {
           >
             {history.map((gen) => {
               const isCurrentlyPlaying = currentAudioId === gen.id && isPlaying;
+              const isGenerating = gen.status === 'generating';
+              const isFailed = gen.status === 'failed';
+              const isPlayable = !isGenerating && !isFailed;
               return (
                 <div
                   key={gen.id}
-                  role="button"
-                  tabIndex={0}
+                  role={isPlayable ? 'button' : undefined}
+                  tabIndex={isPlayable ? 0 : undefined}
                   className={cn(
-                    'flex items-stretch gap-4 h-26 border rounded-md p-3 bg-card hover:bg-muted/70 transition-colors text-left w-full',
+                    'flex items-stretch gap-4 h-26 border rounded-md p-3 bg-card transition-colors text-left w-full',
+                    isPlayable && 'hover:bg-muted/70 cursor-pointer',
                     isCurrentlyPlaying && 'bg-muted/70',
                   )}
                   aria-label={
-                    isCurrentlyPlaying
-                      ? `Sample from ${gen.profile_name}, ${formatDuration(gen.duration)}, ${formatDate(gen.created_at)}. Playing. Press Enter to restart.`
-                      : `Sample from ${gen.profile_name}, ${formatDuration(gen.duration)}, ${formatDate(gen.created_at)}. Press Enter to play.`
+                    isGenerating
+                      ? `Generating speech for ${gen.profile_name}...`
+                      : isFailed
+                        ? `Generation failed for ${gen.profile_name}`
+                        : isCurrentlyPlaying
+                          ? `Sample from ${gen.profile_name}, ${formatDuration(gen.duration ?? 0)}, ${formatDate(gen.created_at)}. Playing. Press Enter to restart.`
+                          : `Sample from ${gen.profile_name}, ${formatDuration(gen.duration ?? 0)}, ${formatDate(gen.created_at)}. Press Enter to play.`
                   }
                   onMouseDown={(e) => {
-                    // Don't trigger play if clicking on textarea or if text is selected
+                    if (!isPlayable) return;
                     const target = e.target as HTMLElement;
                     if (target.closest('textarea') || window.getSelection()?.toString()) {
                       return;
@@ -273,6 +302,7 @@ export function HistoryTable() {
                     handlePlay(gen.id, gen.text, gen.profile_id);
                   }}
                   onKeyDown={(e) => {
+                    if (!isPlayable) return;
                     const target = e.target as HTMLElement;
                     if (target.closest('textarea') || target.closest('button')) return;
                     if (e.key === 'Enter' || e.key === ' ') {
@@ -281,9 +311,14 @@ export function HistoryTable() {
                     }
                   }}
                 >
-                  {/* Waveform icon */}
-                  <div className="flex items-center shrink-0">
-                    <AudioWaveform className="h-5 w-5 text-muted-foreground" />
+                  {/* Status icon */}
+                  <div className="flex items-center shrink-0 w-10 justify-center overflow-hidden">
+                    <div className="scale-50">
+                      <Loader
+                        type={isGenerating ? 'line-scale' : 'line-scale-pulse-out-rapid'}
+                        active={isGenerating || isCurrentlyPlaying}
+                      />
+                    </div>
                   </div>
 
                   {/* Left side - Meta information */}
@@ -294,11 +329,22 @@ export function HistoryTable() {
                     <div className="flex items-center gap-2">
                       <span className="text-xs text-muted-foreground">{gen.language}</span>
                       <span className="text-xs text-muted-foreground">
-                        {formatDuration(gen.duration)}
+                        {formatEngineName(gen.engine, gen.model_size)}
                       </span>
+                      {isFailed ? (
+                        <span className="text-xs text-destructive">Failed</span>
+                      ) : !isGenerating ? (
+                        <span className="text-xs text-muted-foreground">
+                          {formatDuration(gen.duration ?? 0)}
+                        </span>
+                      ) : null}
                     </div>
                     <div className="text-xs text-muted-foreground">
-                      {formatDate(gen.created_at)}
+                      {isGenerating ? (
+                        <span className="text-accent">Generating...</span>
+                      ) : (
+                        formatDate(gen.created_at)
+                      )}
                     </div>
                   </div>
 
@@ -308,16 +354,27 @@ export function HistoryTable() {
                       value={gen.text}
                       className="flex-1 resize-none text-sm text-muted-foreground select-text"
                       readOnly
-                      aria-label={`Transcript for sample from ${gen.profile_name}, ${formatDuration(gen.duration)}`}
+                      aria-label={`Transcript for sample from ${gen.profile_name}, ${formatDuration(gen.duration ?? 0)}`}
                     />
                   </div>
 
-                  {/* Far right - Ellipsis actions */}
+                  {/* Far right - Actions */}
                   <div
-                    className="w-10 shrink-0 flex justify-end"
+                    className="w-10 shrink-0 flex justify-end items-center"
                     onMouseDown={(e) => e.stopPropagation()}
                     onClick={(e) => e.stopPropagation()}
                   >
+                    {isFailed ? (
+                      <Button
+                        variant="ghost"
+                        size="icon"
+                        className="h-8 w-8"
+                        aria-label="Retry generation"
+                        onClick={() => handleRetry(gen.id)}
+                      >
+                        <RotateCcw className="h-4 w-4" />
+                      </Button>
+                    ) : isPlayable ? (
                       <DropdownMenu>
                         <DropdownMenuTrigger asChild>
                           <Button
@@ -360,6 +417,7 @@ export function HistoryTable() {
                           </DropdownMenuItem>
                         </DropdownMenuContent>
                       </DropdownMenu>
+                    ) : null}
                   </div>
                 </div>
               );
@@ -387,7 +445,8 @@ export function HistoryTable() {
           <DialogHeader>
             <DialogTitle>Delete Generation</DialogTitle>
             <DialogDescription>
-              Are you sure you want to delete this generation from "{generationToDelete?.name}"? This action cannot be undone.
+              Are you sure you want to delete this generation from "{generationToDelete?.name}"?
+              This action cannot be undone.
             </DialogDescription>
           </DialogHeader>
           <DialogFooter>

app/src/components/MainEditor/MainEditor.tsx

@@ -13,7 +13,7 @@ import {
 } from '@/components/ui/dialog';
 import { useToast } from '@/components/ui/use-toast';
 import { ProfileList } from '@/components/VoiceProfiles/ProfileList';
-import { BOTTOM_SAFE_AREA_PADDING } from '@/lib/constants/ui';
+
 import { useImportProfile } from '@/lib/hooks/useProfiles';
 import { cn } from '@/lib/utils/cn';
 import { usePlayerStore } from '@/stores/playerStore';
@@ -77,9 +77,9 @@ export function MainEditor() {
 
   return (
     // Main view: Profiles top left, Generator bottom left, History right
-    <div className="grid grid-cols-1 lg:grid-cols-2 gap-6 h-full min-h-0 overflow-hidden relative">
+    <div className="grid grid-cols-1 lg:grid-cols-2 lg:gap-6 h-full min-h-0 overflow-hidden relative">
       {/* Left Column */}
-      <div className="flex flex-col min-h-0 overflow-hidden relative">
+      <div className="flex flex-col min-h-0 overflow-hidden relative lg:overflow-hidden">
         {/* Scroll Mask - Always visible, behind content */}
         <div className="absolute top-0 left-0 right-0 h-16 bg-gradient-to-b from-background to-transparent z-0 pointer-events-none" />
 
@@ -110,10 +110,7 @@ export function MainEditor() {
         {/* Scrollable Content */}
         <div
           ref={scrollRef}
-          className={cn(
-            'flex-1 min-h-0 overflow-y-auto pt-14',
-            isPlayerVisible ? BOTTOM_SAFE_AREA_PADDING : 'pb-4',
-          )}
+          className={cn('flex-1 min-h-0 overflow-y-auto pt-14 pb-4', isPlayerVisible && 'lg:pb-32')}
         >
           <div className="flex flex-col gap-6">
             <div className="shrink-0 flex flex-col">
@@ -123,6 +120,9 @@ export function MainEditor() {
         </div>
       </div>
 
+      {/* Divider - single column only */}
+      {/* <div className="border-t border-border -my-3 lg:hidden" /> */}
+
       {/* Right Column - History */}
       <div className="flex flex-col min-h-0 overflow-hidden">
         <HistoryTable />

app/src/components/ModelsTab/ModelsTab.tsx

@@ -2,7 +2,7 @@ import { ModelManagement } from '@/components/ServerSettings/ModelManagement';
 
 export function ModelsTab() {
   return (
-    <div className="h-full flex flex-col p-4">
+    <div className="h-full flex flex-col">
       <ModelManagement />
     </div>
   );

app/src/components/ServerSettings/ConnectionForm.tsx

@@ -124,6 +124,7 @@ export function ConnectionForm() {
           <div className="flex items-start space-x-3">
             <Checkbox
               id="keepServerRunning"
+              className="mt-[6px]"
               checked={keepServerRunningOnClose}
               onCheckedChange={(checked: boolean) => {
                 setKeepServerRunningOnClose(checked);
@@ -158,6 +159,7 @@ export function ConnectionForm() {
             <div className="flex items-start space-x-3">
               <Checkbox
                 id="allowNetworkAccess"
+                className="mt-[6px]"
                 checked={mode === 'remote'}
                 onCheckedChange={(checked: boolean) => {
                   setMode(checked ? 'remote' : 'local');

app/src/components/ServerSettings/GenerationSettings.tsx

@@ -1,4 +1,5 @@
 import { Card, CardContent, CardDescription, CardHeader, CardTitle } from '@/components/ui/card';
+import { Checkbox } from '@/components/ui/checkbox';
 import { Slider } from '@/components/ui/slider';
 import { useServerStore } from '@/stores/serverStore';
 
@@ -7,6 +8,10 @@ export function GenerationSettings() {
   const setMaxChunkChars = useServerStore((state) => state.setMaxChunkChars);
   const crossfadeMs = useServerStore((state) => state.crossfadeMs);
   const setCrossfadeMs = useServerStore((state) => state.setCrossfadeMs);
+  const normalizeAudio = useServerStore((state) => state.normalizeAudio);
+  const setNormalizeAudio = useServerStore((state) => state.setNormalizeAudio);
+  const autoplayOnGenerate = useServerStore((state) => state.autoplayOnGenerate);
+  const setAutoplayOnGenerate = useServerStore((state) => state.setAutoplayOnGenerate);
 
   return (
     <Card role="region" aria-label="Generation Settings" tabIndex={0}>
@@ -32,7 +37,7 @@ export function GenerationSettings() {
               value={[maxChunkChars]}
               onValueChange={([value]) => setMaxChunkChars(value)}
               min={100}
-              max={2000}
+              max={5000}
               step={50}
               aria-label="Auto-chunking character limit"
             />
@@ -64,6 +69,46 @@ export function GenerationSettings() {
               Blends audio between chunks to smooth transitions. Set to 0 for a hard cut.
             </p>
           </div>
+
+          <div className="flex items-start gap-3">
+            <Checkbox
+              id="normalizeAudio"
+              checked={normalizeAudio}
+              onCheckedChange={setNormalizeAudio}
+              className="mt-[6px]"
+            />
+            <div className="space-y-1">
+              <label
+                htmlFor="normalizeAudio"
+                className="text-sm font-medium leading-none cursor-pointer"
+              >
+                Normalize audio
+              </label>
+              <p className="text-sm text-muted-foreground">
+                Adjusts output volume to a consistent level across generations.
+              </p>
+            </div>
+          </div>
+
+          <div className="flex items-start gap-3">
+            <Checkbox
+              id="autoplayOnGenerate"
+              checked={autoplayOnGenerate}
+              onCheckedChange={setAutoplayOnGenerate}
+              className="mt-[6px]"
+            />
+            <div className="space-y-1">
+              <label
+                htmlFor="autoplayOnGenerate"
+                className="text-sm font-medium leading-none cursor-pointer"
+              >
+                Autoplay on generate
+              </label>
+              <p className="text-sm text-muted-foreground">
+                Automatically play audio when a generation completes.
+              </p>
+            </div>
+          </div>
         </div>
       </CardContent>
     </Card>

app/src/components/ServerSettings/ModelManagement.tsx

@@ -7,6 +7,7 @@ import {
   CircleX,
   Download,
   ExternalLink,
+  FolderOpen,
   HardDrive,
   Heart,
   Loader2,
@@ -41,6 +42,8 @@ import { useToast } from '@/components/ui/use-toast';
 import { apiClient } from '@/lib/api/client';
 import type { ActiveDownloadTask, HuggingFaceModelInfo, ModelStatus } from '@/lib/api/types';
 import { useModelDownloadToast } from '@/lib/hooks/useModelDownloadToast';
+import { usePlatform } from '@/platform/PlatformContext';
+import { useServerStore } from '@/stores/serverStore';
 
 async function fetchHuggingFaceModelInfo(repoId: string): Promise<HuggingFaceModelInfo> {
   const response = await fetch(`https://huggingface.co/api/models/${repoId}`);
@@ -48,6 +51,29 @@ async function fetchHuggingFaceModelInfo(repoId: string): Promise<HuggingFaceMod
   return response.json();
 }
 
+const MODEL_DESCRIPTIONS: Record<string, string> = {
+  'qwen-tts-1.7B':
+    'High-quality multilingual TTS by Alibaba. Supports 10 languages with natural prosody and voice cloning from short reference audio.',
+  'qwen-tts-0.6B':
+    'Lightweight version of Qwen TTS. Same language support with faster inference, ideal for lower-end hardware.',
+  luxtts:
+    'Lightweight ZipVoice-based TTS designed for high quality voice cloning and 48kHz speech generation at speeds exceeding 150x realtime.',
+  'chatterbox-tts':
+    'Production-grade open source TTS by Resemble AI. Supports 23 languages with voice cloning and emotion exaggeration control.',
+  'chatterbox-turbo':
+    'Streamlined 350M parameter TTS by Resemble AI. High-quality English speech with less compute and VRAM than larger models.',
+  'whisper-base':
+    'Smallest Whisper model (74M parameters). Fast transcription with moderate accuracy.',
+  'whisper-small':
+    'Whisper Small (244M parameters). Good balance of speed and accuracy for transcription.',
+  'whisper-medium':
+    'Whisper Medium (769M parameters). Higher accuracy transcription at moderate speed.',
+  'whisper-large':
+    'Whisper Large (1.5B parameters). Best accuracy for speech-to-text across multiple languages.',
+  'whisper-turbo':
+    'Whisper Large v3 Turbo. Pruned for significantly faster inference while maintaining near-large accuracy.',
+};
+
 function formatDownloads(n: number): string {
   if (n >= 1_000_000) return `${(n / 1_000_000).toFixed(1)}M`;
   if (n >= 1_000) return `${(n / 1_000).toFixed(1)}k`;
@@ -85,6 +111,18 @@ function formatBytes(bytes: number): string {
 export function ModelManagement() {
   const { toast } = useToast();
   const queryClient = useQueryClient();
+  const platform = usePlatform();
+  const customModelsDir = useServerStore((state) => state.customModelsDir);
+  const setCustomModelsDir = useServerStore((state) => state.setCustomModelsDir);
+  const [migrating, setMigrating] = useState(false);
+  const [migrationProgress, setMigrationProgress] = useState<{
+    current: number;
+    total: number;
+    progress: number;
+    filename?: string;
+    status: string;
+  } | null>(null);
+  const [pendingMigrateDir, setPendingMigrateDir] = useState<string | null>(null);
   const [downloadingModel, setDownloadingModel] = useState<string | null>(null);
   const [downloadingDisplayName, setDownloadingDisplayName] = useState<string | null>(null);
   const [consoleOpen, setConsoleOpen] = useState(false);
@@ -104,6 +142,12 @@ export function ModelManagement() {
     refetchInterval: 5000,
   });
 
+  const { data: cacheDir } = useQuery({
+    queryKey: ['modelsCacheDir'],
+    queryFn: () => apiClient.getModelsCacheDir(),
+    staleTime: 1000 * 60 * 5,
+  });
+
   const { data: activeTasks } = useQuery({
     queryKey: ['activeTasks'],
     queryFn: () => apiClient.getActiveTasks(),
@@ -382,6 +426,87 @@ export function ModelManagement() {
         </p>
       </div>
 
+      {/* Model storage location */}
+      {platform.metadata.isTauri && cacheDir && (
+        <div className="shrink-0 pb-4 border-b mb-4">
+          <div className="flex items-center justify-between gap-2">
+            <div className="min-w-0">
+              <span className="text-xs text-muted-foreground">Storage location</span>
+              <p
+                className="text-xs font-mono text-muted-foreground/70 truncate"
+                title={cacheDir.path}
+              >
+                {cacheDir.path}
+              </p>
+            </div>
+            <div className="flex items-center gap-1 shrink-0">
+              <Button
+                variant="ghost"
+                size="sm"
+                className="text-xs text-muted-foreground h-7 px-2"
+                onClick={async () => {
+                  try {
+                    const { open } = await import('@tauri-apps/plugin-shell');
+                    await open(cacheDir.path);
+                  } catch {
+                    toast({ title: 'Failed to open model folder', variant: 'destructive' });
+                  }
+                }}
+              >
+                <FolderOpen className="h-3 w-3" />
+                Open
+              </Button>
+              <Button
+                variant="ghost"
+                size="sm"
+                className="text-xs text-muted-foreground h-7 px-2"
+                onClick={async () => {
+                  try {
+                    const { open: openDialog } = await import('@tauri-apps/plugin-dialog');
+                    const selected = await openDialog({
+                      directory: true,
+                      title: 'Choose model storage folder',
+                    });
+                    if (!selected) return;
+                    const newDir =
+                      typeof selected === 'string' ? selected : (selected as { path: string }).path;
+                    if (!newDir) return;
+                    setPendingMigrateDir(newDir);
+                  } catch {
+                    toast({ title: 'Failed to open folder picker', variant: 'destructive' });
+                  }
+                }}
+                disabled={migrating}
+              >
+                {migrating ? (
+                  <Loader2 className="h-3 w-3 animate-spin" />
+                ) : (
+                  <FolderOpen className="h-3 w-3" />
+                )}
+                {migrating ? 'Migrating...' : 'Change'}
+              </Button>
+              {customModelsDir && (
+                <Button
+                  variant="ghost"
+                  size="sm"
+                  className="text-xs text-muted-foreground h-7 px-2"
+                  disabled={migrating}
+                  onClick={async () => {
+                    setCustomModelsDir(null);
+                    toast({ title: 'Reset to default location. Restarting server...' });
+                    await platform.lifecycle.restartServer('');
+                    queryClient.invalidateQueries();
+                  }}
+                >
+                  <RotateCcw className="h-3 w-3" />
+                  Reset
+                </Button>
+              )}
+            </div>
+          </div>
+        </div>
+      )}
+
       {/* Model list */}
       {isLoading ? (
         <div className="flex items-center justify-center py-16">
@@ -457,9 +582,7 @@ export function ModelManagement() {
                             {formatSize(model.size_mb)}
                           </span>
                         )}
-                        {!model.downloaded && !isDownloading && !hasError && (
-                          <span className="text-xs text-muted-foreground/60">Not downloaded</span>
-                        )}
+
                         <ChevronRight className="h-4 w-4 text-muted-foreground/40 group-hover:text-muted-foreground transition-colors" />
                       </div>
                     </button>
@@ -571,13 +694,6 @@ export function ModelManagement() {
                       Error
                     </Badge>
                   )}
-                  {!freshSelectedModel.downloaded &&
-                    !selectedState?.isDownloading &&
-                    !selectedState?.hasError && (
-                      <Badge variant="outline" className="text-xs text-muted-foreground">
-                        Not downloaded
-                      </Badge>
-                    )}
                 </div>
 
                 {/* HuggingFace model card info */}
@@ -588,6 +704,13 @@ export function ModelManagement() {
                   </div>
                 )}
 
+                {/* Description */}
+                {MODEL_DESCRIPTIONS[freshSelectedModel.model_name] && (
+                  <p className="text-xs text-muted-foreground leading-relaxed">
+                    {MODEL_DESCRIPTIONS[freshSelectedModel.model_name]}
+                  </p>
+                )}
+
                 {hfModelInfo && (
                   <div className="space-y-3">
                     {/* Pipeline tag + author */}
@@ -810,6 +933,126 @@ export function ModelManagement() {
           </AlertDialogFooter>
         </AlertDialogContent>
       </AlertDialog>
+
+      {/* Migration confirmation dialog */}
+      <AlertDialog
+        open={!!pendingMigrateDir}
+        onOpenChange={(open) => !open && setPendingMigrateDir(null)}
+      >
+        <AlertDialogContent>
+          <AlertDialogHeader>
+            <AlertDialogTitle>Move models to new location?</AlertDialogTitle>
+            <AlertDialogDescription>
+              The server will shut down while models are being moved to the new folder. It will
+              restart automatically once the migration is complete.
+            </AlertDialogDescription>
+          </AlertDialogHeader>
+          <div
+            className="text-xs font-mono text-muted-foreground bg-muted/50 rounded px-3 py-2 truncate"
+            title={pendingMigrateDir ?? ''}
+          >
+            {pendingMigrateDir}
+          </div>
+          <AlertDialogFooter>
+            <AlertDialogCancel>Cancel</AlertDialogCancel>
+            <AlertDialogAction
+              onClick={async () => {
+                if (!pendingMigrateDir) return;
+                const newDir = pendingMigrateDir;
+                setPendingMigrateDir(null);
+                setMigrating(true);
+                setMigrationProgress({
+                  current: 0,
+                  total: 0,
+                  progress: 0,
+                  status: 'downloading',
+                  filename: 'Preparing...',
+                });
+                try {
+                  // Start the migration (background task)
+                  await apiClient.migrateModels(newDir);
+
+                  // Connect to SSE for progress
+                  await new Promise<void>((resolve, reject) => {
+                    const es = new EventSource(apiClient.getMigrationProgressUrl());
+                    es.onmessage = (event) => {
+                      try {
+                        const data = JSON.parse(event.data);
+                        setMigrationProgress(data);
+                        if (data.status === 'complete') {
+                          es.close();
+                          resolve();
+                        } else if (data.status === 'error') {
+                          es.close();
+                          reject(new Error(data.error || 'Migration failed'));
+                        }
+                      } catch {
+                        /* ignore parse errors */
+                      }
+                    };
+                    es.onerror = () => {
+                      es.close();
+                      reject(new Error('Lost connection during migration'));
+                    };
+                  });
+
+                  setCustomModelsDir(newDir);
+                  setMigrationProgress({
+                    current: 1,
+                    total: 1,
+                    progress: 100,
+                    status: 'complete',
+                    filename: 'Restarting server...',
+                  });
+                  await platform.lifecycle.restartServer(newDir);
+                  queryClient.invalidateQueries();
+                  toast({ title: 'Models moved successfully' });
+                } catch (e) {
+                  toast({
+                    title: 'Migration failed',
+                    description: e instanceof Error ? e.message : 'Failed to migrate models',
+                    variant: 'destructive',
+                  });
+                } finally {
+                  setMigrating(false);
+                  setMigrationProgress(null);
+                }
+              }}
+            >
+              Move Models
+            </AlertDialogAction>
+          </AlertDialogFooter>
+        </AlertDialogContent>
+      </AlertDialog>
+
+      {/* Migration progress overlay */}
+      {migrating && migrationProgress && (
+        <div className="fixed inset-0 z-50 bg-background/95 backdrop-blur-sm flex items-center justify-center">
+          <div className="w-full max-w-md px-8 space-y-6 text-center">
+            <div className="space-y-2">
+              <Loader2 className="h-8 w-8 animate-spin mx-auto text-muted-foreground" />
+              <h2 className="text-lg font-semibold">Moving models</h2>
+              <p className="text-sm text-muted-foreground">
+                {migrationProgress.status === 'complete'
+                  ? 'Restarting server...'
+                  : 'The server is offline while models are being moved.'}
+              </p>
+            </div>
+            {migrationProgress.total > 0 && (
+              <div className="space-y-2">
+                <Progress value={migrationProgress.progress} className="h-2" />
+                <div className="flex justify-between text-xs text-muted-foreground">
+                  <span className="truncate max-w-[60%]">{migrationProgress.filename}</span>
+                  <span>
+                    {formatBytes(migrationProgress.current)} /{' '}
+                    {formatBytes(migrationProgress.total)}
+                  </span>
+                </div>
+              </div>
+            )}
+          </div>
+        </div>
+      )}
     </div>
   );
 }

app/src/components/ServerTab/ServerTab.tsx

@@ -2,12 +2,18 @@ import { ConnectionForm } from '@/components/ServerSettings/ConnectionForm';
 import { GenerationSettings } from '@/components/ServerSettings/GenerationSettings';
 import { GpuAcceleration } from '@/components/ServerSettings/GpuAcceleration';
 import { UpdateStatus } from '@/components/ServerSettings/UpdateStatus';
+import { BOTTOM_SAFE_AREA_PADDING } from '@/lib/constants/ui';
+import { cn } from '@/lib/utils/cn';
 import { usePlatform } from '@/platform/PlatformContext';
+import { usePlayerStore } from '@/stores/playerStore';
 
 export function ServerTab() {
   const platform = usePlatform();
+  const isPlayerVisible = !!usePlayerStore((state) => state.audioUrl);
   return (
-    <div className="overflow-y-auto flex flex-col">
+    <div
+      className={cn('overflow-y-auto flex flex-col', isPlayerVisible && BOTTOM_SAFE_AREA_PADDING)}
+    >
       <div className="grid gap-4 md:grid-cols-2">
         <ConnectionForm />
         <GenerationSettings />

app/src/components/Sidebar.tsx

@@ -1,9 +1,9 @@
 import { Link, useMatchRoute } from '@tanstack/react-router';
-import { Box, BookOpen, Loader2, Mic, Server, Speaker, Volume2 } from 'lucide-react';
+import { BookOpen, Box, Mic, Server, Speaker, Volume2 } from 'lucide-react';
 import voiceboxLogo from '@/assets/voicebox-logo.png';
 import { cn } from '@/lib/utils/cn';
-import { useGenerationStore } from '@/stores/generationStore';
 import { usePlayerStore } from '@/stores/playerStore';
+import { version } from '../../package.json';
 
 interface SidebarProps {
   isMacOS?: boolean;
@@ -19,10 +19,8 @@ const tabs = [
 ];
 
 export function Sidebar({ isMacOS }: SidebarProps) {
-  const isGenerating = useGenerationStore((state) => state.isGenerating);
-  const audioUrl = usePlayerStore((state) => state.audioUrl);
-  const isPlayerVisible = !!audioUrl;
   const matchRoute = useMatchRoute();
+  const isPlayerOpen = !!usePlayerStore((s) => s.audioUrl);
 
   return (
     <div
@@ -42,9 +40,7 @@ export function Sidebar({ isMacOS }: SidebarProps) {
           const Icon = tab.icon;
           // For index route, use exact match; for others, use default matching
           const isActive =
-            tab.path === '/'
-              ? matchRoute({ to: '/', exact: true })
-              : matchRoute({ to: tab.path });
+            tab.path === '/' ? matchRoute({ to: '/', exact: true }) : matchRoute({ to: tab.path });
 
           return (
             <Link
@@ -64,20 +60,13 @@ export function Sidebar({ isMacOS }: SidebarProps) {
         })}
       </div>
 
-      {/* Spacer to push loader to bottom */}
-      <div className="flex-1" />
-
-      {/* Generation Loader */}
-      {isGenerating && (
+      {/* Version */}
       <div
-          className={cn(
-            'w-full flex items-center justify-center transition-all duration-200',
-            isPlayerVisible ? 'mb-[120px]' : 'mb-0',
-          )}
+        className="mt-auto text-[10px] text-muted-foreground/50 transition-all duration-300"
+        style={{ paddingBottom: isPlayerOpen ? '7rem' : undefined }}
       >
-          <Loader2 className="h-6 w-6 text-accent animate-spin" />
+        v{version}
       </div>
-      )}
     </div>
   );
 }

app/src/components/StoriesTab/StoryContent.tsx

@@ -13,8 +13,11 @@ import {
   sortableKeyboardCoordinates,
   verticalListSortingStrategy,
 } from '@dnd-kit/sortable';
+import { Link } from '@tanstack/react-router';
+import { AnimatePresence, motion } from 'framer-motion';
 import { Download, Plus } from 'lucide-react';
 import { useEffect, useMemo, useRef, useState } from 'react';
+import Loader from 'react-loaders';
 import { Button } from '@/components/ui/button';
 import { Input } from '@/components/ui/input';
 import { Popover, PopoverContent, PopoverTrigger } from '@/components/ui/popover';
@@ -28,6 +31,7 @@ import {
   useStory,
 } from '@/lib/hooks/useStories';
 import { useStoryPlayback } from '@/lib/hooks/useStoryPlayback';
+import { useGenerationStore } from '@/stores/generationStore';
 import { useStoryStore } from '@/stores/storyStore';
 import { SortableStoryChatItem } from './StoryChatItem';
 
@@ -40,6 +44,7 @@ export function StoryContent() {
   const addStoryItem = useAddStoryItem();
   const { toast } = useToast();
   const scrollRef = useRef<HTMLDivElement>(null);
+  const pendingCount = useGenerationStore((s) => s.pendingGenerationIds.size);
 
   // Add generation popover state
   const [searchQuery, setSearchQuery] = useState('');
@@ -53,9 +58,9 @@ export function StoryContent() {
     const query = searchQuery.toLowerCase();
     return historyData.items.filter(
       (gen) =>
+        gen.status === 'completed' &&
         !storyGenerationIds.has(gen.id) &&
-        (gen.text.toLowerCase().includes(query) ||
-          gen.profile_name.toLowerCase().includes(query)),
+        (gen.text.toLowerCase().includes(query) || gen.profile_name.toLowerCase().includes(query)),
     );
   }, [historyData, story, searchQuery]);
 
@@ -267,7 +272,31 @@ export function StoryContent() {
             <p className="text-sm text-muted-foreground mt-1">{story.description}</p>
           )}
         </div>
-        <div className="flex gap-2">
+        <div className="flex gap-2 items-center">
+          <AnimatePresence>
+            {pendingCount > 0 && (
+              <motion.div
+                initial={{ opacity: 0, scale: 0.9, width: 0 }}
+                animate={{ opacity: 1, scale: 1, width: 'auto' }}
+                exit={{ opacity: 0, scale: 0.9, width: 0 }}
+                transition={{ duration: 0.2 }}
+              >
+                <Link
+                  to="/"
+                  className="flex items-center gap-2 h-8 pl-1.5 pr-3 rounded-full bg-card border border-border hover:bg-muted/50 transition-all duration-200 cursor-pointer"
+                >
+                  <div className="shrink-0 w-10 h-5 overflow-hidden flex items-center justify-center">
+                    <div className="scale-[0.45]">
+                      <Loader type="line-scale" active />
+                    </div>
+                  </div>
+                  <span className="text-xs text-muted-foreground whitespace-nowrap">
+                    Generating {pendingCount} {pendingCount === 1 ? 'audio' : 'audios'}
+                  </span>
+                </Link>
+              </motion.div>
+            )}
+          </AnimatePresence>
           <Popover open={isAddOpen} onOpenChange={setIsAddOpen}>
             <PopoverTrigger asChild>
               <Button variant="outline" size="sm">
@@ -287,9 +316,7 @@ export function StoryContent() {
               <div className="max-h-60 overflow-y-auto">
                 {availableGenerations.length === 0 ? (
                   <div className="p-4 text-center text-sm text-muted-foreground">
-                    {searchQuery
-                      ? 'No matching generations found'
-                      : 'No available generations'}
+                    {searchQuery ? 'No matching generations found' : 'No available generations'}
                   </div>
                 ) : (
                   availableGenerations.map((gen) => (

app/src/components/VoiceProfiles/ProfileCard.tsx

@@ -78,7 +78,7 @@ export function ProfileCard({ profile }: ProfileCardProps) {
     <>
       <Card
         className={cn(
-          'cursor-pointer hover:shadow-md transition-all flex flex-col',
+          'cursor-pointer hover:shadow-md transition-all flex flex-col h-[162px]',
           isSelected && 'ring-2 ring-primary shadow-md',
         )}
         onClick={handleSelect}

app/src/components/VoiceProfiles/ProfileList.tsx

@@ -41,9 +41,11 @@ export function ProfileList() {
             </CardContent>
           </Card>
         ) : (
-          <div className="grid gap-4 grid-cols-3 auto-rows-auto p-1 pb-[150px]">
+          <div className="flex gap-4 overflow-x-auto p-1 pb-1 lg:grid lg:grid-cols-3 lg:auto-rows-auto lg:overflow-x-visible lg:pb-[150px]">
             {allProfiles.map((profile) => (
-              <ProfileCard key={profile.id} profile={profile} />
+              <div key={profile.id} className="shrink-0 w-[200px] lg:w-auto lg:shrink">
+                <ProfileCard profile={profile} />
+              </div>
             ))}
           </div>
         )}

app/src/components/ui/checkbox.tsx

@@ -1,5 +1,5 @@
-import * as React from 'react';
 import { Check } from 'lucide-react';
+import * as React from 'react';
 import { cn } from '@/lib/utils/cn';
 
 export interface CheckboxProps {

app/src/index.css

@@ -1,4 +1,5 @@
 @import "tailwindcss" source(".");
+@import "loaders.css/loaders.min.css";
 
 @theme {
   --radius-sm: calc(var(--radius) - 4px);
@@ -155,3 +156,18 @@
   animation: fadeIn 0.5s ease-out 0.15s forwards;
   opacity: 0;
 }
+
+/* react-loaders */
+.line-scale-pulse-out-rapid > div,
+.line-scale > div {
+  background-color: hsl(var(--accent)) !important;
+}
+
+.loader-hidden {
+  display: block;
+}
+
+.loader-hidden > div > div {
+  animation-play-state: paused !important;
+  background-color: hsl(var(--muted-foreground)) !important;
+}

app/src/lib/api/client.ts

@@ -200,6 +200,12 @@ class ApiClient {
     });
   }
 
+  async retryGeneration(generationId: string): Promise<GenerationResponse> {
+    return this.request<GenerationResponse>(`/generate/${generationId}/retry`, {
+      method: 'POST',
+    });
+  }
+
   // History
   async listHistory(query?: HistoryQuery): Promise<HistoryListResponse> {
     const params = new URLSearchParams();
@@ -278,6 +284,11 @@ class ApiClient {
     return response.json();
   }
 
+  // Generation status SSE
+  getGenerationStatusUrl(generationId: string): string {
+    return `${this.getBaseUrl()}/generate/${generationId}/status`;
+  }
+
   // Audio
   getAudioUrl(audioId: string): string {
     return `${this.getBaseUrl()}/audio/${audioId}`;
@@ -316,6 +327,21 @@ class ApiClient {
     return this.request<ModelStatusListResponse>('/models/status');
   }
 
+  async getModelsCacheDir(): Promise<{ path: string }> {
+    return this.request<{ path: string }>('/models/cache-dir');
+  }
+
+  async migrateModels(destination: string): Promise<{ source: string; destination: string }> {
+    return this.request('/models/migrate', {
+      method: 'POST',
+      body: JSON.stringify({ destination }),
+    });
+  }
+
+  getMigrationProgressUrl(): string {
+    return `${this.getBaseUrl()}/models/migrate/progress`;
+  }
+
   async triggerModelDownload(modelName: string): Promise<{ message: string }> {
     console.log(
       '[API] triggerModelDownload called for:',

app/src/lib/api/types.ts

@@ -38,6 +38,7 @@ export interface GenerationRequest {
   instruct?: string;
   max_chunk_chars?: number;
   crossfade_ms?: number;
+  normalize?: boolean;
 }
 
 export interface GenerationResponse {
@@ -45,9 +46,14 @@ export interface GenerationResponse {
   profile_id: string;
   text: string;
   language: string;
-  audio_path: string;
-  duration: number;
+  audio_path?: string;
+  duration?: number;
   seed?: number;
+  instruct?: string;
+  engine?: string;
+  model_size?: string;
+  status: 'generating' | 'completed' | 'failed';
+  error?: string;
   created_at: string;
 }
 

app/src/lib/hooks/useGenerationForm.ts

@@ -8,7 +8,6 @@ import { LANGUAGE_CODES, type LanguageCode } from '@/lib/constants/languages';
 import { useGeneration } from '@/lib/hooks/useGeneration';
 import { useModelDownloadToast } from '@/lib/hooks/useModelDownloadToast';
 import { useGenerationStore } from '@/stores/generationStore';
-import { usePlayerStore } from '@/stores/playerStore';
 import { useServerStore } from '@/stores/serverStore';
 
 const generationSchema = z.object({
@@ -30,10 +29,10 @@ interface UseGenerationFormOptions {
 export function useGenerationForm(options: UseGenerationFormOptions = {}) {
   const { toast } = useToast();
   const generation = useGeneration();
-  const setAudioWithAutoPlay = usePlayerStore((state) => state.setAudioWithAutoPlay);
-  const setIsGenerating = useGenerationStore((state) => state.setIsGenerating);
+  const addPendingGeneration = useGenerationStore((state) => state.addPendingGeneration);
   const maxChunkChars = useServerStore((state) => state.maxChunkChars);
   const crossfadeMs = useServerStore((state) => state.crossfadeMs);
+  const normalizeAudio = useServerStore((state) => state.normalizeAudio);
   const [downloadingModelName, setDownloadingModelName] = useState<string | null>(null);
   const [downloadingDisplayName, setDownloadingDisplayName] = useState<string | null>(null);
 
@@ -70,8 +69,6 @@ export function useGenerationForm(options: UseGenerationFormOptions = {}) {
     }
 
     try {
-      setIsGenerating(true);
-
       const engine = data.engine || 'qwen';
       const modelName =
         engine === 'luxtts'
@@ -92,6 +89,7 @@ export function useGenerationForm(options: UseGenerationFormOptions = {}) {
                 ? 'Qwen TTS 1.7B'
                 : 'Qwen TTS 0.6B';
 
+      // Check if model needs downloading
       try {
         const modelStatus = await apiClient.getModelStatus();
         const model = modelStatus.models.find((m) => m.model_name === modelName);
@@ -105,6 +103,7 @@ export function useGenerationForm(options: UseGenerationFormOptions = {}) {
       }
 
       const isQwen = engine === 'qwen';
+      // This now returns immediately with status="generating"
       const result = await generation.mutateAsync({
         profile_id: selectedProfileId,
         text: data.text,
@@ -115,16 +114,13 @@ export function useGenerationForm(options: UseGenerationFormOptions = {}) {
         instruct: isQwen ? data.instruct || undefined : undefined,
         max_chunk_chars: maxChunkChars,
         crossfade_ms: crossfadeMs,
+        normalize: normalizeAudio,
       });
 
-      toast({
-        title: 'Generation complete!',
-        description: `Audio generated (${result.duration.toFixed(2)}s)`,
-      });
-
-      const audioUrl = apiClient.getAudioUrl(result.id);
-      setAudioWithAutoPlay(audioUrl, result.id, selectedProfileId, data.text.substring(0, 50));
+      // Track this generation for SSE status updates
+      addPendingGeneration(result.id);
 
+      // Reset form immediately — user can start typing again
       form.reset({
         text: '',
         language: data.language,
@@ -141,7 +137,6 @@ export function useGenerationForm(options: UseGenerationFormOptions = {}) {
         variant: 'destructive',
       });
     } finally {
-      setIsGenerating(false);
       setDownloadingModelName(null);
       setDownloadingDisplayName(null);
     }

app/src/lib/hooks/useGenerationProgress.ts

@@ -0,0 +1,154 @@
+import { useQueryClient } from '@tanstack/react-query';
+import { useEffect, useRef } from 'react';
+import { useToast } from '@/components/ui/use-toast';
+import { apiClient } from '@/lib/api/client';
+import { useGenerationStore } from '@/stores/generationStore';
+import { usePlayerStore } from '@/stores/playerStore';
+import { useServerStore } from '@/stores/serverStore';
+
+interface GenerationStatusEvent {
+  id: string;
+  status: 'generating' | 'completed' | 'failed' | 'not_found';
+  duration?: number;
+  error?: string;
+}
+
+/**
+ * Subscribes to SSE for all pending generations. When a generation completes,
+ * invalidates the history query, removes it from pending, and auto-plays
+ * if the player is idle.
+ */
+export function useGenerationProgress() {
+  const queryClient = useQueryClient();
+  const { toast } = useToast();
+  const pendingIds = useGenerationStore((s) => s.pendingGenerationIds);
+  const removePendingGeneration = useGenerationStore((s) => s.removePendingGeneration);
+  const removePendingStoryAdd = useGenerationStore((s) => s.removePendingStoryAdd);
+  const isPlaying = usePlayerStore((s) => s.isPlaying);
+  const setAudioWithAutoPlay = usePlayerStore((s) => s.setAudioWithAutoPlay);
+  const autoplayOnGenerate = useServerStore((s) => s.autoplayOnGenerate);
+
+  // Keep refs to avoid stale closures in EventSource handlers
+  const isPlayingRef = useRef(isPlaying);
+  const autoplayRef = useRef(autoplayOnGenerate);
+  isPlayingRef.current = isPlaying;
+  autoplayRef.current = autoplayOnGenerate;
+
+  // Track active EventSource instances
+  const eventSourcesRef = useRef<Map<string, EventSource>>(new Map());
+
+  // Unmount-only cleanup — close all SSE connections when the hook is torn down
+  useEffect(() => {
+    const sources = eventSourcesRef.current;
+    return () => {
+      for (const source of sources.values()) {
+        source.close();
+      }
+      sources.clear();
+    };
+  }, []);
+
+  useEffect(() => {
+    const currentSources = eventSourcesRef.current;
+
+    // Close SSE connections for IDs no longer pending
+    for (const [id, source] of currentSources.entries()) {
+      if (!pendingIds.has(id)) {
+        source.close();
+        currentSources.delete(id);
+      }
+    }
+
+    // Open SSE connections for new pending IDs
+    for (const id of pendingIds) {
+      if (currentSources.has(id)) continue;
+
+      const url = apiClient.getGenerationStatusUrl(id);
+      const source = new EventSource(url);
+
+      source.onmessage = (event) => {
+        try {
+          const data: GenerationStatusEvent = JSON.parse(event.data);
+
+          if (data.status === 'completed') {
+            source.close();
+            currentSources.delete(id);
+            removePendingGeneration(id);
+
+            // Refresh history to pick up the completed generation
+            queryClient.invalidateQueries({ queryKey: ['history'] });
+
+            // If this generation was queued for a story, add it now
+            const storyId = removePendingStoryAdd(id);
+            if (storyId) {
+              apiClient
+                .addStoryItem(storyId, { generation_id: id })
+                .then(() => {
+                  queryClient.invalidateQueries({ queryKey: ['stories'] });
+                  queryClient.invalidateQueries({ queryKey: ['stories', storyId] });
+                  toast({
+                    title: 'Added to story',
+                    description: data.duration
+                      ? `Audio generated (${data.duration.toFixed(2)}s) and added to story`
+                      : 'Audio generated and added to story',
+                  });
+                })
+                .catch(() => {
+                  toast({
+                    title: 'Generation complete',
+                    description: 'Audio generated but failed to add to story',
+                    variant: 'destructive',
+                  });
+                });
+            } else {
+              // toast({
+              //   title: 'Generation complete!',
+              //   description: data.duration
+              //     ? `Audio generated (${data.duration.toFixed(2)}s)`
+              //     : 'Audio generated',
+              // });
+            }
+
+            // Auto-play if enabled and nothing is currently playing
+            if (autoplayRef.current && !isPlayingRef.current) {
+              const genAudioUrl = apiClient.getAudioUrl(id);
+              setAudioWithAutoPlay(genAudioUrl, id, '', '');
+            }
+          } else if (data.status === 'failed' || data.status === 'not_found') {
+            source.close();
+            currentSources.delete(id);
+            removePendingGeneration(id);
+            removePendingStoryAdd(id);
+
+            queryClient.invalidateQueries({ queryKey: ['history'] });
+
+            toast({
+              title: data.status === 'not_found' ? 'Generation not found' : 'Generation failed',
+              description: data.error || 'An error occurred during generation',
+              variant: 'destructive',
+            });
+          }
+        } catch {
+          // Ignore parse errors from heartbeats etc
+        }
+      };
+
+      source.onerror = () => {
+        // EventSource auto-reconnects, but if we get repeated errors
+        // just clean up
+        source.close();
+        currentSources.delete(id);
+        removePendingGeneration(id);
+      };
+
+      currentSources.set(id, source);
+    }
+  }, [
+    pendingIds,
+    removePendingGeneration,
+    removePendingStoryAdd,
+    queryClient,
+    toast,
+    setAudioWithAutoPlay,
+  ]);
+}

app/src/lib/hooks/useRestoreActiveTasks.tsx

@@ -1,10 +1,10 @@
 import { useCallback, useEffect, useRef, useState } from 'react';
 import { apiClient } from '@/lib/api/client';
-import { useGenerationStore } from '@/stores/generationStore';
 import type { ActiveDownloadTask } from '@/lib/api/types';
+import { useGenerationStore } from '@/stores/generationStore';
 
 // Polling interval in milliseconds
-const POLL_INTERVAL = 2000;
+const POLL_INTERVAL = 30000;
 
 /**
  * Hook to monitor active tasks (downloads and generations).
@@ -15,8 +15,8 @@ const POLL_INTERVAL = 2000;
  */
 export function useRestoreActiveTasks() {
   const [activeDownloads, setActiveDownloads] = useState<ActiveDownloadTask[]>([]);
-  const setIsGenerating = useGenerationStore((state) => state.setIsGenerating);
   const setActiveGenerationId = useGenerationStore((state) => state.setActiveGenerationId);
+  const addPendingGeneration = useGenerationStore((state) => state.addPendingGeneration);
 
   // Track which downloads we've seen to detect new ones
   const seenDownloadsRef = useRef<Set<string>>(new Set());
@@ -25,15 +25,15 @@ export function useRestoreActiveTasks() {
     try {
       const tasks = await apiClient.getActiveTasks();
 
-      // Update generation state
+      // Restore pending generations (e.g., after page refresh)
       if (tasks.generations.length > 0) {
-        setIsGenerating(true);
         setActiveGenerationId(tasks.generations[0].task_id);
+        for (const gen of tasks.generations) {
+          addPendingGeneration(gen.task_id);
+        }
       } else {
-        // Only clear if we were tracking a generation
         const currentId = useGenerationStore.getState().activeGenerationId;
         if (currentId) {
-          setIsGenerating(false);
           setActiveGenerationId(null);
         }
       }
@@ -59,7 +59,7 @@ export function useRestoreActiveTasks() {
       // Silently fail - server might be temporarily unavailable
       console.debug('Failed to fetch active tasks:', error);
     }
-  }, [setIsGenerating, setActiveGenerationId]);
+  }, [setActiveGenerationId, addPendingGeneration]);
 
   useEffect(() => {
     // Fetch immediately on mount

app/src/lib/utils/format.ts

@@ -25,6 +25,21 @@ export function formatDate(date: string | Date): string {
   return formatDistance(dateObj, new Date(), { addSuffix: true }).replace(/^about /i, '');
 }
 
+const ENGINE_DISPLAY_NAMES: Record<string, string> = {
+  qwen: 'Qwen',
+  luxtts: 'LuxTTS',
+  chatterbox: 'Chatterbox',
+  chatterbox_turbo: 'Chatterbox Turbo',
+};
+
+export function formatEngineName(engine?: string, modelSize?: string): string {
+  const name = ENGINE_DISPLAY_NAMES[engine ?? 'qwen'] ?? engine ?? 'Qwen';
+  if (engine === 'qwen' && modelSize) {
+    return `${name} ${modelSize}`;
+  }
+  return name;
+}
+
 export function formatFileSize(bytes: number): string {
   if (bytes === 0) return '0 Bytes';
   const k = 1024;

app/src/platform/types.ts

@@ -49,9 +49,9 @@ export interface PlatformAudio {
 }
 
 export interface PlatformLifecycle {
-  startServer(remote?: boolean): Promise<string>;
+  startServer(remote?: boolean, modelsDir?: string | null): Promise<string>;
   stopServer(): Promise<void>;
-  restartServer(): Promise<string>;
+  restartServer(modelsDir?: string | null): Promise<string>;
   setKeepServerRunning(keep: boolean): Promise<void>;
   setupWindowCloseHandler(): Promise<void>;
   onServerReady?: () => void;

app/src/router.tsx

@@ -8,8 +8,10 @@ import { Sidebar } from '@/components/Sidebar';
 import { StoriesTab } from '@/components/StoriesTab/StoriesTab';
 import { Toaster } from '@/components/ui/toaster';
 import { VoicesTab } from '@/components/VoicesTab/VoicesTab';
+import { useGenerationProgress } from '@/lib/hooks/useGenerationProgress';
 import { useModelDownloadToast } from '@/lib/hooks/useModelDownloadToast';
 import { MODEL_DISPLAY_NAMES, useRestoreActiveTasks } from '@/lib/hooks/useRestoreActiveTasks';
+
 // Simple platform check that works in both web and Tauri
 const isMacOS = () => navigator.platform.toLowerCase().includes('mac');
 
@@ -18,6 +20,9 @@ function RootLayout() {
   // Monitor active downloads/generations and show toasts for them
   const activeDownloads = useRestoreActiveTasks();
 
+  // Subscribe to SSE for pending generations — handles completion, auto-play, and history refresh
+  useGenerationProgress();
+
   return (
     <AppFrame>
       <div className="flex flex-1 min-h-0 overflow-hidden">

app/src/stores/generationStore.ts

@@ -1,15 +1,58 @@
 import { create } from 'zustand';
 
 interface GenerationState {
+  /** IDs of generations currently in progress */
+  pendingGenerationIds: Set<string>;
+  /** Whether any generation is in progress (derived from pendingGenerationIds) */
   isGenerating: boolean;
-  activeGenerationId: string | null;
-  setIsGenerating: (generating: boolean) => void;
+  /** Map of generationId → storyId for deferred story additions */
+  pendingStoryAdds: Map<string, string>;
+  addPendingGeneration: (id: string) => void;
+  removePendingGeneration: (id: string) => void;
+  addPendingStoryAdd: (generationId: string, storyId: string) => void;
+  removePendingStoryAdd: (generationId: string) => string | undefined;
   setActiveGenerationId: (id: string | null) => void;
+  activeGenerationId: string | null;
 }
 
-export const useGenerationStore = create<GenerationState>((set) => ({
+export const useGenerationStore = create<GenerationState>((set, get) => ({
+  pendingGenerationIds: new Set(),
   isGenerating: false,
   activeGenerationId: null,
-  setIsGenerating: (generating) => set({ isGenerating: generating }),
+  pendingStoryAdds: new Map(),
+
+  addPendingGeneration: (id) =>
+    set((state) => {
+      const next = new Set(state.pendingGenerationIds);
+      next.add(id);
+      return { pendingGenerationIds: next, isGenerating: true };
+    }),
+
+  removePendingGeneration: (id) =>
+    set((state) => {
+      const next = new Set(state.pendingGenerationIds);
+      next.delete(id);
+      return { pendingGenerationIds: next, isGenerating: next.size > 0 };
+    }),
+
+  addPendingStoryAdd: (generationId, storyId) =>
+    set((state) => {
+      const next = new Map(state.pendingStoryAdds);
+      next.set(generationId, storyId);
+      return { pendingStoryAdds: next };
+    }),
+
+  removePendingStoryAdd: (generationId) => {
+    const storyId = get().pendingStoryAdds.get(generationId);
+    if (storyId) {
+      set((state) => {
+        const next = new Map(state.pendingStoryAdds);
+        next.delete(generationId);
+        return { pendingStoryAdds: next };
+      });
+    }
+    return storyId;
+  },
+
   setActiveGenerationId: (id) => set({ activeGenerationId: id }),
 }));

app/src/stores/serverStore.ts

@@ -19,6 +19,15 @@ interface ServerStore {
 
   crossfadeMs: number;
   setCrossfadeMs: (value: number) => void;
+
+  normalizeAudio: boolean;
+  setNormalizeAudio: (value: boolean) => void;
+
+  autoplayOnGenerate: boolean;
+  setAutoplayOnGenerate: (value: boolean) => void;
+
+  customModelsDir: string | null;
+  setCustomModelsDir: (dir: string | null) => void;
 }
 
 export const useServerStore = create<ServerStore>()(
@@ -41,6 +50,15 @@ export const useServerStore = create<ServerStore>()(
 
       crossfadeMs: 50,
       setCrossfadeMs: (value) => set({ crossfadeMs: value }),
+
+      normalizeAudio: true,
+      setNormalizeAudio: (value) => set({ normalizeAudio: value }),
+
+      autoplayOnGenerate: true,
+      setAutoplayOnGenerate: (value) => set({ autoplayOnGenerate: value }),
+
+      customModelsDir: null,
+      setCustomModelsDir: (dir) => set({ customModelsDir: dir }),
     }),
     {
       name: 'voicebox-server',

backend/database.py

@@ -45,10 +45,14 @@ class Generation(Base):
     profile_id = Column(String, ForeignKey("profiles.id"), nullable=False)
     text = Column(Text, nullable=False)
     language = Column(String, default="en")
-    audio_path = Column(String, nullable=False)
-    duration = Column(Float, nullable=False)
+    audio_path = Column(String, nullable=True)
+    duration = Column(Float, nullable=True)
     seed = Column(Integer)
     instruct = Column(Text)
+    engine = Column(String, default="qwen")
+    model_size = Column(String, nullable=True)
+    status = Column(String, default="completed")  # generating, completed, failed
+    error = Column(Text, nullable=True)
     created_at = Column(DateTime, default=datetime.utcnow)
 
 
@@ -288,6 +292,36 @@ def _run_migrations(engine):
                 conn.commit()
                 print("Added avatar_path column to profiles")
 
+    # Migration: Add status and error columns to generations table
+    if 'generations' in inspector.get_table_names():
+        columns = {col['name'] for col in inspector.get_columns('generations')}
+        if 'status' not in columns:
+            print("Migrating generations: adding status column")
+            with engine.connect() as conn:
+                conn.execute(text("ALTER TABLE generations ADD COLUMN status VARCHAR DEFAULT 'completed'"))
+                conn.commit()
+                print("Added status column to generations")
+        if 'error' not in columns:
+            print("Migrating generations: adding error column")
+            with engine.connect() as conn:
+                conn.execute(text("ALTER TABLE generations ADD COLUMN error TEXT"))
+                conn.commit()
+                print("Added error column to generations")
+        if 'engine' not in columns:
+            print("Migrating generations: adding engine column")
+            with engine.connect() as conn:
+                conn.execute(text("ALTER TABLE generations ADD COLUMN engine VARCHAR DEFAULT 'qwen'"))
+                conn.commit()
+                print("Added engine column to generations")
+        # Re-read columns after engine migration (variable name shadows outer `engine`)
+        columns = {col['name'] for col in inspector.get_columns('generations')}
+        if 'model_size' not in columns:
+            print("Migrating generations: adding model_size column")
+            with engine.connect() as conn:
+                conn.execute(text("ALTER TABLE generations ADD COLUMN model_size VARCHAR"))
+                conn.commit()
+                print("Added model_size column to generations")
+
 
 def get_db():
     """Get database session (generator for dependency injection)."""

backend/history.py

@@ -29,6 +29,10 @@ async def create_generation(
     seed: Optional[int],
     db: Session,
     instruct: Optional[str] = None,
+    generation_id: Optional[str] = None,
+    status: str = "completed",
+    engine: Optional[str] = "qwen",
+    model_size: Optional[str] = None,
 ) -> GenerationResponse:
     """
     Create a new generation history entry.
@@ -42,12 +46,16 @@ async def create_generation(
         seed: Random seed used (if any)
         db: Database session
         instruct: Natural language instruction used (if any)
+        generation_id: Pre-assigned ID (for async generation flow)
+        status: Generation status (generating, completed, failed)
+        engine: TTS engine used (qwen, luxtts, chatterbox, chatterbox_turbo)
+        model_size: Model size variant (1.7B, 0.6B) — only relevant for qwen
 
     Returns:
         Created generation entry
     """
     db_generation = DBGeneration(
-        id=str(uuid.uuid4()),
+        id=generation_id or str(uuid.uuid4()),
         profile_id=profile_id,
         text=text,
         language=language,
@@ -55,6 +63,9 @@ async def create_generation(
         duration=duration,
         seed=seed,
         instruct=instruct,
+        engine=engine,
+        model_size=model_size,
+        status=status,
         created_at=datetime.utcnow(),
     )
 
@@ -65,6 +76,32 @@ async def create_generation(
     return GenerationResponse.model_validate(db_generation)
 
 
+async def update_generation_status(
+    generation_id: str,
+    status: str,
+    db: Session,
+    audio_path: Optional[str] = None,
+    duration: Optional[float] = None,
+    error: Optional[str] = None,
+) -> Optional[GenerationResponse]:
+    """Update the status of a generation (used by async generation flow)."""
+    generation = db.query(DBGeneration).filter_by(id=generation_id).first()
+    if not generation:
+        return None
+
+    generation.status = status
+    if audio_path is not None:
+        generation.audio_path = audio_path
+    if duration is not None:
+        generation.duration = duration
+    if error is not None:
+        generation.error = error
+
+    db.commit()
+    db.refresh(generation)
+    return GenerationResponse.model_validate(generation)
+
+
 async def get_generation(
     generation_id: str,
     db: Session,
@@ -143,6 +180,10 @@ async def list_generations(
             duration=generation.duration,
             seed=generation.seed,
             instruct=generation.instruct,
+            engine=generation.engine or "qwen",
+            model_size=generation.model_size,
+            status=generation.status or "completed",
+            error=generation.error,
             created_at=generation.created_at,
         ))
     

backend/main.py

@@ -62,6 +62,9 @@ from .platform_detect import get_backend_type
 # Keep references to fire-and-forget background tasks to prevent GC
 _background_tasks: set = set()
 
+# Generation queue — serializes TTS inference to avoid GPU contention
+_generation_queue: asyncio.Queue = None  # type: ignore  # initialized at startup
+
 
 def _create_background_task(coro) -> asyncio.Task:
     """Create a background task and prevent it from being garbage collected."""
@@ -71,6 +74,24 @@ def _create_background_task(coro) -> asyncio.Task:
     return task
 
 
+async def _generation_worker():
+    """Worker that processes generation tasks one at a time."""
+    while True:
+        coro = await _generation_queue.get()
+        try:
+            await coro
+        except Exception:
+            import traceback
+            traceback.print_exc()
+        finally:
+            _generation_queue.task_done()
+
+
+def _enqueue_generation(coro):
+    """Add a generation coroutine to the serial queue."""
+    _generation_queue.put_nowait(coro)
+
+
 app = FastAPI(
     title="voicebox API",
     description="Production-quality Qwen3-TTS voice cloning API",
@@ -695,138 +716,67 @@ async def generate_speech(
     data: models.GenerationRequest,
     db: Session = Depends(get_db),
 ):
-    """Generate speech from text using a voice profile."""
+    """Generate speech from text using a voice profile.
+    
+    Creates a history entry immediately with status='generating' and kicks off
+    TTS in the background. The frontend can poll or use SSE to detect completion.
+    """
     task_manager = get_task_manager()
     generation_id = str(uuid.uuid4())
 
-    try:
-        # Start tracking generation
+    # Validate profile exists before creating the record
+    profile = await profiles.get_profile(data.profile_id, db)
+    if not profile:
+        raise HTTPException(status_code=404, detail="Profile not found")
+
+    from .backends import get_tts_backend_for_engine
+    engine = data.engine or "qwen"
+    tts_model = get_tts_backend_for_engine(engine)
+    model_size = data.model_size or "1.7B"
+
+    # Create the history entry immediately with status="generating"
+    generation = await history.create_generation(
+        profile_id=data.profile_id,
+        text=data.text,
+        language=data.language,
+        audio_path="",
+        duration=0,
+        seed=data.seed,
+        db=db,
+        instruct=data.instruct,
+        generation_id=generation_id,
+        status="generating",
+        engine=engine,
+        model_size=model_size if engine == "qwen" else None,
+    )
+
+    # Track in task manager
     task_manager.start_generation(
         task_id=generation_id,
         profile_id=data.profile_id,
         text=data.text,
     )
 
-        # Get profile
-        profile = await profiles.get_profile(data.profile_id, db)
-        if not profile:
-            raise HTTPException(status_code=404, detail="Profile not found")
-        
-        # Generate audio
-        from .backends import get_tts_backend_for_engine
-
-        engine = data.engine or "qwen"
-        tts_model = get_tts_backend_for_engine(engine)
-
-        # Resolve model size (only relevant for Qwen engine)
-        model_size = data.model_size or "1.7B"
-
-        # Check if model needs to be downloaded first
+    # Kick off TTS in background
+    async def _run_generation():
+        bg_db = next(get_db())
+        try:
+            # Load model
             if engine == "qwen":
-            if not tts_model._is_model_cached(model_size):
-                model_name = f"qwen-tts-{model_size}"
-
-                async def download_model_background():
-                    try:
                 await tts_model.load_model_async(model_size)
-                    except Exception as e:
-                        task_manager.error_download(model_name, str(e))
-
-                task_manager.start_download(model_name)
-                _create_background_task(download_model_background())
-
-                raise HTTPException(
-                    status_code=202,
-                    detail={
-                        "message": f"Model {model_size} is being downloaded. Please wait and try again.",
-                        "model_name": model_name,
-                        "downloading": True,
-                    },
-                )
-
-            # Load (or switch to) the requested model
-            await tts_model.load_model_async(model_size)
-        elif engine == "luxtts":
-            if not tts_model._is_model_cached():
-                model_name = "luxtts"
-
-                async def download_luxtts_background():
-                    try:
-                        await tts_model.load_model()
-                    except Exception as e:
-                        task_manager.error_download(model_name, str(e))
-
-                task_manager.start_download(model_name)
-                _create_background_task(download_luxtts_background())
-
-                raise HTTPException(
-                    status_code=202,
-                    detail={
-                        "message": "LuxTTS model is being downloaded. Please wait and try again.",
-                        "model_name": model_name,
-                        "downloading": True,
-                    },
-                )
-
-            await tts_model.load_model()
-        elif engine == "chatterbox":
-            if not tts_model._is_model_cached():
-                model_name = "chatterbox-tts"
-
-                async def download_chatterbox_background():
-                    try:
-                        await tts_model.load_model()
-                    except Exception as e:
-                        task_manager.error_download(model_name, str(e))
-
-                task_manager.start_download(model_name)
-                asyncio.create_task(download_chatterbox_background())
-
-                raise HTTPException(
-                    status_code=202,
-                    detail={
-                        "message": "Chatterbox model is being downloaded. Please wait and try again.",
-                        "model_name": model_name,
-                        "downloading": True,
-                    },
-                )
-
-            await tts_model.load_model()
-        elif engine == "chatterbox_turbo":
-            if not tts_model._is_model_cached():
-                model_name = "chatterbox-turbo"
-
-                async def download_chatterbox_turbo_background():
-                    try:
-                        await tts_model.load_model()
-                    except Exception as e:
-                        task_manager.error_download(model_name, str(e))
-
-                task_manager.start_download(model_name)
-                asyncio.create_task(download_chatterbox_turbo_background())
-
-                raise HTTPException(
-                    status_code=202,
-                    detail={
-                        "message": "Chatterbox Turbo model is being downloaded. Please wait and try again.",
-                        "model_name": model_name,
-                        "downloading": True,
-                    },
-                )
-
+            else:
                 await tts_model.load_model()
 
-        # Create voice prompt from profile
+            # Create voice prompt
             voice_prompt = await profiles.create_voice_prompt_for_profile(
                 data.profile_id,
-            db,
+                bg_db,
                 use_cache=True,
                 engine=engine,
             )
 
             from .utils.chunked_tts import generate_chunked
 
-        # Resolve per-chunk trim function for engines that need it
             trim_fn = None
             if engine in ("chatterbox", "chatterbox_turbo"):
                 from .utils.audio import trim_tts_output
@@ -844,61 +794,177 @@ async def generate_speech(
                 trim_fn=trim_fn,
             )
 
-        # Calculate duration
-        duration = len(audio) / sample_rate
+            if data.normalize:
+                from .utils.audio import normalize_audio
+                audio = normalize_audio(audio)
 
-        # Save audio
+            duration = len(audio) / sample_rate
             audio_path = config.get_generations_dir() / f"{generation_id}.wav"
 
             from .utils.audio import save_audio
-        import errno
-
-        try:
             save_audio(audio, str(audio_path), sample_rate)
-        except BrokenPipeError:
-            raise HTTPException(
-                status_code=500,
-                detail="Audio save failed: broken pipe (the output stream was closed unexpectedly)",
-            )
-        except OSError as save_err:
-            err_no = getattr(save_err, "errno", None) or (
-                getattr(save_err.__cause__, "errno", None)
-                if save_err.__cause__
-                else None
-            )
-            if err_no == errno.ENOENT:
-                msg = f"Audio save failed: directory not found — {audio_path.parent}"
-            elif err_no == errno.EACCES:
-                msg = f"Audio save failed: permission denied — {audio_path.parent}"
-            elif err_no == errno.ENOSPC:
-                msg = "Audio save failed: no disk space remaining"
-            else:
-                msg = f"Audio save failed: {save_err}"
-            raise HTTPException(status_code=500, detail=msg)
 
-        # Create history entry
-        generation = await history.create_generation(
-            profile_id=data.profile_id,
-            text=data.text,
-            language=data.language,
+            # Update the record to completed
+            await history.update_generation_status(
+                generation_id=generation_id,
+                status="completed",
+                db=bg_db,
                 audio_path=str(audio_path),
                 duration=duration,
-            seed=data.seed,
-            db=db,
-            instruct=data.instruct,
             )
 
-        # Mark generation as complete
+        except Exception as e:
+            import traceback
+            traceback.print_exc()
+            await history.update_generation_status(
+                generation_id=generation_id,
+                status="failed",
+                db=bg_db,
+                error=str(e),
+            )
+        finally:
             task_manager.complete_generation(generation_id)
+            bg_db.close()
+
+    _enqueue_generation(_run_generation())
 
     return generation
 
-    except ValueError as e:
-        task_manager.complete_generation(generation_id)
-        raise HTTPException(status_code=400, detail=str(e))
+
+@app.post("/generate/{generation_id}/retry", response_model=models.GenerationResponse)
+async def retry_generation(generation_id: str, db: Session = Depends(get_db)):
+    """Retry a failed generation using the same parameters."""
+    gen = db.query(DBGeneration).filter_by(id=generation_id).first()
+    if not gen:
+        raise HTTPException(status_code=404, detail="Generation not found")
+
+    if (gen.status or "completed") != "failed":
+        raise HTTPException(status_code=400, detail="Only failed generations can be retried")
+
+    # Reset the record to generating
+    gen.status = "generating"
+    gen.error = None
+    gen.audio_path = ""
+    gen.duration = 0
+    db.commit()
+    db.refresh(gen)
+
+    task_manager = get_task_manager()
+    task_manager.start_generation(
+        task_id=generation_id,
+        profile_id=gen.profile_id,
+        text=gen.text,
+    )
+
+    # Resolve engine/model from stored values
+    retry_engine = gen.engine or "qwen"
+    retry_model_size = gen.model_size or "1.7B"
+
+    from .backends import get_tts_backend_for_engine
+    tts_model = get_tts_backend_for_engine(retry_engine)
+
+    async def _run_retry():
+        bg_db = next(get_db())
+        try:
+            if retry_engine == "qwen":
+                await tts_model.load_model_async(retry_model_size)
+            else:
+                await tts_model.load_model()
+
+            voice_prompt = await profiles.create_voice_prompt_for_profile(
+                gen.profile_id,
+                bg_db,
+                use_cache=True,
+                engine=retry_engine,
+            )
+
+            from .utils.chunked_tts import generate_chunked
+
+            trim_fn = None
+            if retry_engine in ("chatterbox", "chatterbox_turbo"):
+                from .utils.audio import trim_tts_output
+                trim_fn = trim_tts_output
+
+            audio, sample_rate = await generate_chunked(
+                tts_model,
+                gen.text,
+                voice_prompt,
+                language=gen.language,
+                seed=gen.seed,
+                instruct=gen.instruct,
+                trim_fn=trim_fn,
+            )
+
+            duration = len(audio) / sample_rate
+            audio_path = config.get_generations_dir() / f"{generation_id}.wav"
+
+            from .utils.audio import save_audio
+            save_audio(audio, str(audio_path), sample_rate)
+
+            await history.update_generation_status(
+                generation_id=generation_id,
+                status="completed",
+                db=bg_db,
+                audio_path=str(audio_path),
+                duration=duration,
+            )
         except Exception as e:
+            import traceback
+            traceback.print_exc()
+            await history.update_generation_status(
+                generation_id=generation_id,
+                status="failed",
+                db=bg_db,
+                error=str(e),
+            )
+        finally:
             task_manager.complete_generation(generation_id)
-        raise HTTPException(status_code=500, detail=str(e))
+            bg_db.close()
+
+    _enqueue_generation(_run_retry())
+
+    return models.GenerationResponse.model_validate(gen)
+
+
+@app.get("/generate/{generation_id}/status")
+async def get_generation_status(generation_id: str, db: Session = Depends(get_db)):
+    """SSE endpoint that streams generation status updates.
+    
+    Polls the DB every second and yields the current status. Closes when
+    the generation reaches 'completed' or 'failed'.
+    """
+    import json
+
+    async def event_stream():
+        while True:
+            db.expire_all()
+            gen = db.query(DBGeneration).filter_by(id=generation_id).first()
+            if not gen:
+                yield f"data: {json.dumps({'status': 'not_found', 'id': generation_id})}\n\n"
+                return
+
+            payload = {
+                "id": gen.id,
+                "status": gen.status or "completed",
+                "duration": gen.duration,
+                "error": gen.error,
+            }
+            yield f"data: {json.dumps(payload)}\n\n"
+
+            if (gen.status or "completed") in ("completed", "failed"):
+                return
+
+            await asyncio.sleep(1)
+
+    return StreamingResponse(
+        event_stream(),
+        media_type="text/event-stream",
+        headers={
+            "Cache-Control": "no-cache",
+            "Connection": "keep-alive",
+            "X-Accel-Buffering": "no",
+        },
+    )
 
 
 @app.post("/generate/stream")
@@ -975,6 +1041,10 @@ async def stream_speech(
         trim_fn=trim_fn,
     )
 
+    if data.normalize:
+        from .utils.audio import normalize_audio
+        audio = normalize_audio(audio)
+
     wav_bytes = tts.audio_to_wav_bytes(audio, sample_rate)
 
     async def _wav_stream():
@@ -1588,6 +1658,141 @@ async def get_model_progress(model_name: str):
     )
 
 
+@app.get("/models/cache-dir")
+async def get_models_cache_dir():
+    """Get the path to the HuggingFace model cache directory."""
+    from huggingface_hub import constants as hf_constants
+    return {"path": str(Path(hf_constants.HF_HUB_CACHE))}
+
+
+def _get_dir_size(path: Path) -> int:
+    """Get total size of a directory in bytes."""
+    total = 0
+    for f in path.rglob("*"):
+        if f.is_file():
+            total += f.stat().st_size
+    return total
+
+
+def _copy_with_progress(src: Path, dst: Path, progress_manager, copied_so_far: int, total_bytes: int) -> int:
+    """Copy a directory tree with byte-level progress tracking."""
+    import shutil
+    dst.mkdir(parents=True, exist_ok=True)
+    for item in src.iterdir():
+        dest_item = dst / item.name
+        if item.is_dir():
+            copied_so_far = _copy_with_progress(item, dest_item, progress_manager, copied_so_far, total_bytes)
+        else:
+            size = item.stat().st_size
+            shutil.copy2(str(item), str(dest_item))
+            copied_so_far += size
+            progress_manager.update_progress(
+                "migration", copied_so_far, total_bytes,
+                filename=item.name, status="downloading",
+            )
+    return copied_so_far
+
+
+@app.post("/models/migrate")
+async def migrate_models(request: models.ModelMigrateRequest):
+    """Move all downloaded models to a new directory with byte-level progress via SSE."""
+    import shutil
+    from huggingface_hub import constants as hf_constants
+
+    source = Path(hf_constants.HF_HUB_CACHE)
+    destination = Path(request.destination)
+
+    if not source.exists():
+        raise HTTPException(status_code=404, detail="Current model cache directory not found")
+
+    model_dirs = [d for d in source.iterdir() if d.name.startswith("models--") and d.is_dir()]
+    if not model_dirs:
+        return {"moved": 0, "errors": [], "source": str(source), "destination": str(destination)}
+
+    destination.mkdir(parents=True, exist_ok=True)
+
+    progress_manager = get_progress_manager()
+
+    # Check if source and destination are on the same filesystem (rename is instant)
+    same_fs = False
+    try:
+        same_fs = source.stat().st_dev == destination.stat().st_dev
+    except OSError:
+        pass
+
+    async def migrate_background():
+        moved = 0
+        errors = []
+        try:
+            if same_fs:
+                # Same filesystem: rename is instant, just track model count
+                total = len(model_dirs)
+                for i, item in enumerate(model_dirs):
+                    dest_item = destination / item.name
+                    try:
+                        if dest_item.exists():
+                            shutil.rmtree(dest_item)
+                        shutil.move(str(item), str(dest_item))
+                        moved += 1
+                        progress_manager.update_progress(
+                            "migration", i + 1, total,
+                            filename=item.name, status="downloading",
+                        )
+                    except Exception as e:
+                        errors.append(f"{item.name}: {str(e)}")
+            else:
+                # Cross-filesystem: copy with byte-level progress, then delete source
+                total_bytes = sum(_get_dir_size(d) for d in model_dirs)
+                progress_manager.update_progress("migration", 0, total_bytes, filename="Calculating...", status="downloading")
+
+                copied = 0
+                for item in model_dirs:
+                    dest_item = destination / item.name
+                    try:
+                        if dest_item.exists():
+                            shutil.rmtree(dest_item)
+                        copied = await asyncio.to_thread(
+                            _copy_with_progress, item, dest_item, progress_manager, copied, total_bytes
+                        )
+                        # Remove source after successful copy
+                        await asyncio.to_thread(shutil.rmtree, str(item))
+                        moved += 1
+                    except Exception as e:
+                        errors.append(f"{item.name}: {str(e)}")
+
+            progress_manager.update_progress("migration", 1, 1, status="complete")
+            progress_manager.mark_complete("migration")
+        except Exception as e:
+            progress_manager.update_progress("migration", 0, 0, status="error")
+            progress_manager.mark_error("migration", str(e))
+
+    _create_background_task(migrate_background())
+
+    return {"source": str(source), "destination": str(destination)}
+
+
+@app.get("/models/migrate/progress")
+async def get_migration_progress():
+    """Get model migration progress via Server-Sent Events."""
+    from fastapi.responses import StreamingResponse
+
+    progress_manager = get_progress_manager()
+
+    async def event_generator():
+        async for event in progress_manager.subscribe("migration"):
+            yield event
+
+    return StreamingResponse(
+        event_generator(),
+        media_type="text/event-stream",
+        headers={
+            "Cache-Control": "no-cache",
+            "Connection": "keep-alive",
+            "X-Accel-Buffering": "no",
+        },
+    )
+
+
 @app.get("/models/status", response_model=models.ModelStatusListResponse)
 async def get_model_status():
     """Get status of all available models."""
@@ -2337,9 +2542,29 @@ def _get_gpu_status() -> str:
 @app.on_event("startup")
 async def startup_event():
     """Run on application startup."""
+    global _generation_queue
     print("voicebox API starting up...")
     database.init_db()
     print(f"Database initialized at {database._db_path}")
+
+    # Start the serial generation worker
+    _generation_queue = asyncio.Queue()
+    _create_background_task(_generation_worker())
+
+    # Mark any stale "generating" records as failed — these are leftovers
+    # from a previous process that was killed mid-generation
+    try:
+        from sqlalchemy import text as sa_text
+        db = next(get_db())
+        result = db.execute(
+            sa_text("UPDATE generations SET status = 'failed', error = 'Server was shut down during generation' WHERE status = 'generating'")
+        )
+        if result.rowcount > 0:
+            print(f"Marked {result.rowcount} stale generation(s) as failed")
+        db.commit()
+        db.close()
+    except Exception as e:
+        print(f"Warning: Could not clean up stale generations: {e}")
     backend_type = get_backend_type()
     print(f"Backend: {backend_type.upper()}")
     print(f"GPU available: {_get_gpu_status()}")

backend/models.py

@@ -60,6 +60,7 @@ class GenerationRequest(BaseModel):
     engine: Optional[str] = Field(default="qwen", pattern="^(qwen|luxtts|chatterbox|chatterbox_turbo)$")
     max_chunk_chars: int = Field(default=800, ge=100, le=5000, description="Max characters per chunk for long text splitting")
     crossfade_ms: int = Field(default=50, ge=0, le=500, description="Crossfade duration in ms between chunks (0 for hard cut)")
+    normalize: bool = Field(default=True, description="Normalize output audio volume")
 
 
 class GenerationResponse(BaseModel):
@@ -68,10 +69,14 @@ class GenerationResponse(BaseModel):
     profile_id: str
     text: str
     language: str
-    audio_path: str
-    duration: float
-    seed: Optional[int]
-    instruct: Optional[str]
+    audio_path: Optional[str] = None
+    duration: Optional[float] = None
+    seed: Optional[int] = None
+    instruct: Optional[str] = None
+    engine: Optional[str] = "qwen"
+    model_size: Optional[str] = None
+    status: str = "completed"
+    error: Optional[str] = None
     created_at: datetime
 
     class Config:
@@ -93,10 +98,14 @@ class HistoryResponse(BaseModel):
     profile_name: str
     text: str
     language: str
-    audio_path: str
-    duration: float
-    seed: Optional[int]
-    instruct: Optional[str]
+    audio_path: Optional[str] = None
+    duration: Optional[float] = None
+    seed: Optional[int] = None
+    instruct: Optional[str] = None
+    engine: Optional[str] = "qwen"
+    model_size: Optional[str] = None
+    status: str = "completed"
+    error: Optional[str] = None
     created_at: datetime
 
     class Config:
@@ -170,6 +179,11 @@ class ModelDownloadRequest(BaseModel):
     model_name: str
 
 
+class ModelMigrateRequest(BaseModel):
+    """Request model for migrating models to a new directory."""
+    destination: str
+
+
 class ActiveDownloadTask(BaseModel):
     """Response model for active download task."""
     model_name: str

backend/stories.py

@@ -270,11 +270,14 @@ async def add_item_to_story(
             generation_created_at=generation.created_at,
         )
 
+    # Get track from data or default to 0
+    track = data.track if data.track is not None else 0
+
     # Calculate start_time_ms if not provided
     if data.start_time_ms is not None:
         start_time_ms = data.start_time_ms
     else:
-        # Find the maximum end time (start_time_ms + duration_ms) of existing items
+        # Find the maximum end time on the target track only
         existing_items = db.query(
             DBStoryItem,
             DBGeneration
@@ -282,11 +285,11 @@ async def add_item_to_story(
             DBGeneration,
             DBStoryItem.generation_id == DBGeneration.id
         ).filter(
-            DBStoryItem.story_id == story_id
+            DBStoryItem.story_id == story_id,
+            DBStoryItem.track == track,
         ).all()
         
         if not existing_items:
-            # First item starts at 0
             start_time_ms = 0
         else:
             max_end_time_ms = 0
@@ -297,9 +300,6 @@ async def add_item_to_story(
             # Add 200ms gap after the last item
             start_time_ms = max_end_time_ms + 200
 
-    # Get track from data or default to 0
-    track = data.track if data.track is not None else 0
-
     # Create item
     item = DBStoryItem(
         id=str(uuid.uuid4()),

bun.lock

@@ -4,6 +4,10 @@
   "workspaces": {
     "": {
       "name": "voicebox",
+      "dependencies": {
+        "loaders.css": "^0.1.2",
+        "react-loaders": "^3.0.1",
+      },
       "devDependencies": {
         "@biomejs/biome": "2.3.12",
         "@types/node": "^20.0.0",
@@ -13,7 +17,7 @@
     },
     "app": {
       "name": "@voicebox/app",
-      "version": "0.1.11",
+      "version": "0.1.13",
       "dependencies": {
         "@dnd-kit/core": "^6.3.1",
         "@dnd-kit/sortable": "^10.0.0",
@@ -68,7 +72,7 @@
     },
     "landing": {
       "name": "@voicebox/landing",
-      "version": "0.1.11",
+      "version": "0.1.13",
       "dependencies": {
         "@radix-ui/react-separator": "^1.1.8",
         "@radix-ui/react-slot": "^1.2.4",
@@ -93,7 +97,7 @@
     },
     "tauri": {
       "name": "@voicebox/tauri",
-      "version": "0.1.11",
+      "version": "0.1.13",
       "dependencies": {
         "@tauri-apps/api": "^2.0.0",
         "@tauri-apps/plugin-dialog": "^2.0.0",
@@ -116,7 +120,7 @@
     },
     "web": {
       "name": "@voicebox/web",
-      "version": "0.1.11",
+      "version": "0.1.13",
       "dependencies": {
         "@tanstack/react-query": "^5.0.0",
         "react": "^18.3.0",
@@ -125,6 +129,7 @@
         "zustand": "^4.5.0",
       },
       "devDependencies": {
+        "@tailwindcss/vite": "^4.0.0",
         "@types/react": "^18.3.0",
         "@types/react-dom": "^18.3.0",
         "@typescript-eslint/eslint-plugin": "^7.0.0",
@@ -677,6 +682,8 @@
 
     "class-variance-authority": ["[email protected]", "", { "dependencies": { "clsx": "^2.1.1" } }, "sha512-Ka+9Trutv7G8M6WT6SeiRWz792K5qEqIGEGzXKhAE6xOWAY6pPH8U+9IY3oCMv6kqTmLsv7Xh/2w2RigkePMsg=="],
 
+    "classnames": ["[email protected]", "", {}, "sha512-saHYOzhIQs6wy2sVxTM6bUDsQO4F50V9RQ22qBpEdCW+I+/Wmke2HOl6lS6dTpdxVhb88/I6+Hs+438c3lfUow=="],
+
     "client-only": ["[email protected]", "", {}, "sha512-IV3Ou0jSMzZrd3pZ48nLkT9DA7Ag1pnPzaiQhpW7c3RbcqqzvzzVu+L8gfqMp/8IM2MQtSiqaCxrrcfu8I8rMA=="],
 
     "clsx": ["[email protected]", "", {}, "sha512-eYm0QWBtUrBWZWG0d386OGAw16Z995PiOVo2B7bjWSbHedGl5e0ZWaq65kOGgUSNesEIDkB9ISbTg/JK9dhCZA=="],
@@ -873,6 +880,8 @@
 
     "lines-and-columns": ["[email protected]", "", {}, "sha512-7ylylesZQ/PV29jhEDl3Ufjo6ZX7gCqJr5F7PKrqc93v7fzSymt1BpwEU8nAUXs8qzzvqhbjhK5QZg6Mt/HkBg=="],
 
+    "loaders.css": ["[email protected]", "", {}, "sha512-Rhowlq24ey1VOeor+3wYOt9+MjaxBOJm1u4KlQgNC3+0xJ0LS4wq4iG57D/BPzvuD/7HHDGQOWJ+81oR2EI9bQ=="],
+
     "locate-path": ["[email protected]", "", { "dependencies": { "p-locate": "^5.0.0" } }, "sha512-iPZK6eYjbxRu3uB4/WZ3EsEIMJFMqAoopl3R+zuq0UjcAm/MO6KCweDgPfP3elTztoKP3KtnVHxTn2NHBSDVUw=="],
 
     "lodash.merge": ["[email protected]", "", {}, "sha512-0KpjqXRVvrYyCsX1swR/XTK0va6VQkQM6MNo7PqW77ByjAhoARA8EfrP1N4+KlKj8YS0ZUCtRT/YUuhyYDujIQ=="],
@@ -959,6 +968,8 @@
 
     "prelude-ls": ["[email protected]", "", {}, "sha512-vkcDPrRZo1QZLbn5RLGPpg/WmIQ65qoWWhcGKf/b5eplkkarX0m9z8ppCat4mlOqUsWpyNuYgO3VRyrYHSzX5g=="],
 
+    "prop-types": ["[email protected]", "", { "dependencies": { "loose-envify": "^1.4.0", "object-assign": "^4.1.1", "react-is": "^16.13.1" } }, "sha512-oj87CgZICdulUohogVAR7AjlC0327U4el4L6eAvOqCeudMDVU0NThNaV+b9Df4dXgSP1gXMTnPdhfe/2qDH5cg=="],
+
     "punycode": ["[email protected]", "", {}, "sha512-vYt7UD1U9Wg6138shLtLOvdAu+8DsC/ilFtEVHcH+wydcSpNE20AfSOduf6MkRFahL5FY7X1oU7nKVZFtfq8Fg=="],
 
     "queue-microtask": ["[email protected]", "", {}, "sha512-NuaNSa6flKT5JaSYQzJok04JzTL1CA6aGhv5rfLW3PgqA+M2ChpZQnAC8h8i4ZFkBS8X5RqkDBHA7r4hej3K9A=="],
@@ -969,6 +980,10 @@
 
     "react-hook-form": ["[email protected]", "", { "peerDependencies": { "react": "^16.8.0 || ^17 || ^18 || ^19" } }, "sha512-9SUJKCGKo8HUSsCO+y0CtqkqI5nNuaDqTxyqPsZPqIwudpj4rCrAz/jZV+jn57bx5gtZKOh3neQu94DXMc+w5w=="],
 
+    "react-is": ["[email protected]", "", {}, "sha512-24e6ynE2H+OKt4kqsOvNd8kBpV65zoxbA4BVsEOB3ARVWQki/DHzaUoC5KuON/BiccDaCCTZBuOcfZs70kR8bQ=="],
+
+    "react-loaders": ["[email protected]", "", { "dependencies": { "classnames": "^2.2.3" }, "peerDependencies": { "prop-types": ">=15.6.0", "react": ">=15" } }, "sha512-4igMNqs9Fb3d4Z+0UHIGQNJsw/37gX0nUO8QxupnEKRn1dtyYC1LGwk5GuaoDciMQCQc/MmPwb4Fn6ZfdoX1FQ=="],
+
     "react-refresh": ["[email protected]", "", {}, "sha512-z6F7K9bV85EfseRCp2bzrpyQ0Gkw1uLoCel9XBVWPg/TjRj94SkJzUTGfOa4bs7iJvBWtQG0Wq7wnI0syw3EBQ=="],
 
     "react-remove-scroll": ["[email protected]", "", { "dependencies": { "react-remove-scroll-bar": "^2.3.7", "react-style-singleton": "^2.2.3", "tslib": "^2.1.0", "use-callback-ref": "^1.3.3", "use-sidecar": "^1.1.3" }, "peerDependencies": { "@types/react": "*", "react": "^16.8.0 || ^17.0.0 || ^18.0.0 || ^19.0.0 || ^19.0.0-rc" }, "optionalPeers": ["@types/react"] }, "sha512-Iqb9NjCCTt6Hf+vOdNIZGdTiH1QSqr27H/Ek9sv/a97gfueI/5h1s3yRi1nngzMUaOOToin5dI1dXKdXiF+u0Q=="],

docs/issue-pain-points.md

@@ -0,0 +1,67 @@
+# Voicebox Issue Pain Points (Snapshot)
+
+## Scope
+
+- Dataset: **128 total issues** (**107 open**, **21 closed**)
+- Source: GitHub issues in `jamiepine/voicebox`
+- Classification: keyword/theme clustering
+- Note: counts below are **non-exclusive** (one issue can belong to multiple pain points)
+
+## Most Common Pain Points (Open Issues)
+
+| Rank | Pain Point | Open Issues | What users are reporting |
+|---|---|---:|---|
+| 1 | Model download & offline reliability | **32** | Downloads failing/stalling, cache/offline behavior inconsistent, wrong model size selected, Errno issues |
+| 2 | GPU/backend compatibility | **22** | GPU not detected, backend fallback surprises, platform-specific runtime failures (Windows/Mac) |
+| 3 | Export/save/file persistence | **15** | Export fails, "failed to fetch/download audio", samples/profiles not saving |
+| 4 | Language/accent quality & coverage | **14** | Missing language support, accent mismatch, robotic outputs |
+| 5 | Update/restart safety + long-op controls | **4** | Auto-restart without warning, update confusion, lack of cancel/pause controls |
+
+## Representative Issues by Pain Point
+
+### 1) Model download & offline reliability (32)
+
+- [#159](https://github.com/jamiepine/voicebox/issues/159) - Qwen download fails with Errno 22
+- [#151](https://github.com/jamiepine/voicebox/issues/151) - Model loading hangs / server crashes
+- [#150](https://github.com/jamiepine/voicebox/issues/150) - Internet required despite downloaded models
+- [#149](https://github.com/jamiepine/voicebox/issues/149) - Cancel/pause controls for large downloads
+- [#96](https://github.com/jamiepine/voicebox/issues/96) - 0.6B selection still uses/downloads 1.7B
+
+### 2) GPU/backend compatibility (22)
+
+- [#164](https://github.com/jamiepine/voicebox/issues/164) - Windows: no GPU usage + multiple breakages
+- [#141](https://github.com/jamiepine/voicebox/issues/141) - Using CPU only, GPU not used
+- [#131](https://github.com/jamiepine/voicebox/issues/131) - Numpy ABI mismatch in bundled app
+- [#130](https://github.com/jamiepine/voicebox/issues/130) - Intel Mac tensor/padding generation error
+- [#127](https://github.com/jamiepine/voicebox/issues/127) - GPU not found
+
+### 3) Export/save/file persistence (15)
+
+- [#148](https://github.com/jamiepine/voicebox/issues/148) - Japanese export fails on 0.1.12
+- [#143](https://github.com/jamiepine/voicebox/issues/143) - Samples not saving
+- [#134](https://github.com/jamiepine/voicebox/issues/134) - Can't save profile
+- [#105](https://github.com/jamiepine/voicebox/issues/105) - Export audio fails (failed to fetch)
+- [#49](https://github.com/jamiepine/voicebox/issues/49) - Export filename/location ignored on Windows
+
+### 4) Language/accent quality & coverage (14)
+
+- [#162](https://github.com/jamiepine/voicebox/issues/162) - Persian audio request/problem
+- [#117](https://github.com/jamiepine/voicebox/issues/117) - Arabic language support
+- [#113](https://github.com/jamiepine/voicebox/issues/113) - Polish language support
+- [#109](https://github.com/jamiepine/voicebox/issues/109) - Ukrainian support
+- [#100](https://github.com/jamiepine/voicebox/issues/100) - Non-US accent quality issues
+
+### 5) Update/restart safety + controls (4)
+
+- [#164](https://github.com/jamiepine/voicebox/issues/164) - Update behavior + usability failures
+- [#136](https://github.com/jamiepine/voicebox/issues/136) - Auto-restart without warning
+- [#86](https://github.com/jamiepine/voicebox/issues/86) - Unexpected restart with no confirmation
+- [#149](https://github.com/jamiepine/voicebox/issues/149) - Need pause/cancel and pre-download confirmation
+
+## Additional Signal
+
+- There is also a large **feature-request/misc** bucket (**36 open**) that is competing with stability triage (audiobook, Linux build, additional ASR/TTS models, integrations).
+
+## Takeaway
+
+Most user pain is concentrated in four stability areas: **download/offline path**, **GPU/backend detection**, **save/export reliability**, and **language/accent correctness**. Addressing those first should reduce the majority of current support friction.

docs/plans/PROJECT_STATUS.md

@@ -321,7 +321,7 @@ Notable requests:
 
 ## New Model Integration — Landscape
 
-### Models Worth Supporting (2026 SOTA)
+### Models Worth Supporting (2026 SOTA — updated March 13)
 
 | Model | Cloning | Speed | Sample Rate | Languages | VRAM | Integration Ease | Status |
 |-------|---------|-------|-------------|-----------|------|-----------------|--------|
@@ -329,10 +329,23 @@ Notable requests:
 | **LuxTTS** | 3s zero-shot | 150x RT, CPU ok | 48 kHz | English | <1 GB | **Shipped** | PR #254 |
 | **Chatterbox MTL** | 5s zero-shot | Medium | 24 kHz | 23 | Medium | **Shipped** | PR #257 |
 | **Chatterbox Turbo** | 5s zero-shot | Fast | 24 kHz | English | Low | **PR #258** | In review |
+| **HumeAI TADA 1B/3B** | Zero-shot | 5× faster than LLM-TTS | — | EN (1B), Multilingual (3B) | Medium | Needs vetting | MIT, 700s+ coherent, synced transcript output |
+| **MOSS-TTS Family** | Zero-shot | — | — | Multilingual | Medium | Needs vetting | Apache 2.0, multi-speaker dialogue, text-to-voice design (no ref audio) |
+| **VoxCPM 1.5** | Zero-shot (seconds) | ~0.15 RTF streaming | — | Bilingual (EN/ZH) | Medium | Needs vetting | Apache 2.0, tokenizer-free continuous diffusion, LoRA-friendly |
+| **Pocket TTS** | Zero-shot + streaming | >1× RT on CPU | — | English | ~100M params, CPU-first | Needs vetting | MIT, Kyutai Labs, no GPU required |
+| **Kokoro-82M** | 3s instant | CPU realtime | 24 kHz | English | Tiny (82M) | Ready | Apache 2.0, multi-engine arch in place |
 | **XTTS-v2** | 6s zero-shot | Mid-GPU | 24 kHz | 17+ | Medium | Ready | Multi-engine arch in place |
 | **Fish Speech** | 10-30s few-shot | Real-time | 24-44 kHz | 50+ | Medium | Ready | Multi-engine arch in place |
 | **CosyVoice2-0.5B** | 3-10s zero-shot | Very fast | 24 kHz | Multilingual | Low | Ready | Multi-engine arch in place |
-| **Kokoro-82M** | 3s instant | CPU realtime | 24 kHz | English | Tiny | Ready | Multi-engine arch in place |
+
+#### Notes on New Candidates (March 2026)
+
+- **HumeAI TADA** — Text-Audio Dual Alignment arch. Near-zero hallucinations/drift, free synced transcript. 700+ seconds coherent audio. Best candidate for Stories long-form reliability. [HF: HumeAI/tada-1b](https://huggingface.co/HumeAI/tada-1b) | [GitHub: HumeAI/tada](https://github.com/HumeAI/tada)
+- **MOSS-TTS** — Modular suite: flagship cloning, MOSS-TTSD (multi-speaker dialogue), MOSS-VoiceGenerator (create voices from text descriptions, no ref audio). Unique UX for Stories voice design. [GitHub: OpenMOSS/MOSS-TTS](https://github.com/OpenMOSS/MOSS-TTS)
+- **VoxCPM 1.5** — Tokenizer-free continuous diffusion + autoregressive. No discrete token artifacts. Context-aware prosody/emotion, real-time streaming, LoRA fine-tuning. Trained on 1.8M+ hours. [GitHub: OpenBMB/VoxCPM](https://github.com/OpenBMB/VoxCPM)
+- **Pocket TTS** — 100M param CPU-first model from Kyutai Labs (Moshi team). Runs >1× realtime without GPU. Broadens hardware support significantly. [GitHub: kyutai-labs/pocket-tts](https://github.com/kyutai-labs/pocket-tts)
+- **Watch list:** MioTTS-2.6B (fast LLM-based EN/JP, vLLM compatible), Oolel-Voices (Soynade Research, expressive modular control)
+- **Skipped:** Fish Audio S2 — restrictive research license (commercial use requires approval), despite strong features
 
 ### Adding a New Engine (Now Straightforward)
 
@@ -402,16 +415,21 @@ The generation form now uses a flat model dropdown with engine-based routing. Pe
 
 ### Tier 3 — Future (v0.3.0+)
 
-| Item | Notes |
-|------|-------|
-| XTTS-v2 / Fish Speech / CosyVoice | Multi-engine arch is ready; just needs backend implementation |
-| OpenAI-compatible API (plan doc exists) | Low effort once API is stable |
-| LoRA fine-tuning (PR #195) | Complex, needs rework for multi-engine |
-| External/remote providers | Depends on use case demand |
-| GGUF support (#226) | Depends on model ecosystem maturity |
-| Queue system (#234) | Batch generation |
-| Streaming for non-MLX engines | Currently MLX-only |
-| Kokoro-82M | Tiny model, great for CPU-only machines |
+| Priority | Item | Notes |
+|----------|------|-------|
+| 1 | **HumeAI TADA** | Long-form reliability for Stories, synced transcripts. Addresses #234, #203, #191, #111, #69. Needs API vetting. |
+| 2 | **Pocket TTS** (Kyutai) | CPU-first 100M model, broadens hardware support. Kyutai ships clean code. Needs API vetting. |
+| 3 | **MOSS-TTS** | Text-to-voice design (no ref audio) is unique. Multi-speaker dialogue for Stories. Needs thorough API vetting. |
+| 4 | **Kokoro-82M** | 82M params, CPU realtime, Apache 2.0. Easy win. |
+| 5 | **Model config registry refactor** | Reduce 5-dispatch-point duplication in main.py — do before adding 3+ more engines |
+| 6 | XTTS-v2 / Fish Speech / CosyVoice | Multi-engine arch is ready; just needs backend implementation |
+| 7 | **VoxCPM 1.5** | Tokenizer-free streaming, interesting but uncertain integration surface |
+| 8 | OpenAI-compatible API (plan doc exists) | Low effort once API is stable |
+| 9 | LoRA fine-tuning (PR #195) | Complex, needs rework for multi-engine |
+| 10 | External/remote providers | Depends on use case demand |
+| 11 | GGUF support (#226) | Depends on model ecosystem maturity |
+| 12 | Queue system (#234) | Batch generation |
+| 13 | Streaming for non-MLX engines | Currently MLX-only |
 
 ---
 

package.json

@@ -40,5 +40,9 @@
   "engines": {
     "bun": ">=1.0.0"
   },
-  "packageManager": "[email protected]"
+  "packageManager": "[email protected]",
+  "dependencies": {
+    "loaders.css": "^0.1.2",
+    "react-loaders": "^3.0.1"
+  }
 }

tauri/src-tauri/src/main.rs

@@ -16,6 +16,7 @@ struct ServerState {
     child: Mutex<Option<tauri_plugin_shell::process::CommandChild>>,
     server_pid: Mutex<Option<u32>>,
     keep_running_on_close: Mutex<bool>,
+    models_dir: Mutex<Option<String>>,
 }
 
 #[command]
@@ -23,7 +24,16 @@ async fn start_server(
     app: tauri::AppHandle,
     state: State<'_, ServerState>,
     remote: Option<bool>,
+    models_dir: Option<String>,
 ) -> Result<String, String> {
+    // Store models_dir for use on restart (empty string means reset to default)
+    if let Some(ref dir) = models_dir {
+        if dir.is_empty() {
+            *state.models_dir.lock().unwrap() = None;
+        } else {
+            *state.models_dir.lock().unwrap() = Some(dir.clone());
+        }
+    }
     // Check if server is already running (managed by this app instance)
     if state.child.lock().unwrap().is_some() {
         return Ok(format!("http://127.0.0.1:{}", SERVER_PORT));
@@ -274,6 +284,12 @@ async fn start_server(
     let port_str = SERVER_PORT.to_string();
     let is_remote = remote.unwrap_or(false);
 
+    // Resolve the custom models directory from the parameter or stored state
+    let effective_models_dir = models_dir.or_else(|| state.models_dir.lock().unwrap().clone());
+    if let Some(ref dir) = effective_models_dir {
+        println!("Custom models directory: {}", dir);
+    }
+
     // If CUDA binary exists, launch it directly instead of the bundled sidecar
     let spawn_result = if let Some(ref cuda_path) = cuda_binary {
         println!("Launching CUDA backend: {:?}", cuda_path);
@@ -282,6 +298,9 @@ async fn start_server(
         if is_remote {
             cmd = cmd.args(["--host", "0.0.0.0"]);
         }
+        if let Some(ref dir) = effective_models_dir {
+            cmd = cmd.env("VOICEBOX_MODELS_DIR", dir);
+        }
         cmd.spawn()
     } else {
         // Use the bundled CPU sidecar
@@ -289,6 +308,9 @@ async fn start_server(
         if is_remote {
             sidecar = sidecar.args(["--host", "0.0.0.0"]);
         }
+        if let Some(ref dir) = effective_models_dir {
+            sidecar = sidecar.env("VOICEBOX_MODELS_DIR", dir);
+        }
         println!("Spawning server process...");
         sidecar.spawn()
     };
@@ -613,9 +635,19 @@ async fn stop_server(state: State<'_, ServerState>) -> Result<(), String> {
 async fn restart_server(
     app: tauri::AppHandle,
     state: State<'_, ServerState>,
+    models_dir: Option<String>,
 ) -> Result<String, String> {
     println!("restart_server: stopping current server...");
 
+    // Update stored models_dir: empty string means reset to default, non-empty means set
+    if let Some(ref dir) = models_dir {
+        if dir.is_empty() {
+            *state.models_dir.lock().unwrap() = None;
+        } else {
+            *state.models_dir.lock().unwrap() = Some(dir.clone());
+        }
+    }
+
     // Stop the current server
     stop_server(state.clone()).await?;
 
@@ -623,9 +655,9 @@ async fn restart_server(
     println!("restart_server: waiting for port release...");
     tokio::time::sleep(tokio::time::Duration::from_millis(1000)).await;
 
-    // Start server again (will auto-detect CUDA binary)
+    // Start server again (will auto-detect CUDA binary and use stored models_dir)
     println!("restart_server: starting server...");
-    start_server(app, state, None).await
+    start_server(app, state, None, None).await
 }
 
 #[command]
@@ -686,6 +718,7 @@ pub fn run() {
             child: Mutex::new(None),
             server_pid: Mutex::new(None),
             keep_running_on_close: Mutex::new(false),
+            models_dir: Mutex::new(None),
         })
         .manage(audio_capture::AudioCaptureState::new())
         .manage(audio_output::AudioOutputState::new())

tauri/src-tauri/tauri.conf.json

@@ -56,7 +56,7 @@
   },
   "plugins": {
     "shell": {
-      "open": true
+      "open": ".*"
     },
     "updater": {
       "pubkey": "dW50cnVzdGVkIGNvbW1lbnQ6IG1pbmlzaWduIHB1YmxpYyBrZXk6IEUxRENBQkRBQjdBNTM1OTIKUldTU05hVzMycXZjNGJGcUxmcVVocll2QjdSaTJNdlFxR2M3VDJsMnVvbDdyZGRPMmRlOW9aWTcK",

tauri/src/platform/lifecycle.ts

@@ -5,9 +5,12 @@ import type { PlatformLifecycle } from '@/platform/types';
 class TauriLifecycle implements PlatformLifecycle {
   onServerReady?: () => void;
 
-  async startServer(remote = false): Promise<string> {
+  async startServer(remote = false, modelsDir?: string | null): Promise<string> {
     try {
-      const result = await invoke<string>('start_server', { remote });
+      const result = await invoke<string>('start_server', {
+        remote,
+        modelsDir: modelsDir ?? undefined,
+      });
       console.log('Server started:', result);
       this.onServerReady?.();
       return result;
@@ -27,9 +30,11 @@ class TauriLifecycle implements PlatformLifecycle {
     }
   }
 
-  async restartServer(): Promise<string> {
+  async restartServer(modelsDir?: string | null): Promise<string> {
     try {
-      const result = await invoke<string>('restart_server');
+      const result = await invoke<string>('restart_server', {
+        modelsDir: modelsDir ?? undefined,
+      });
       console.log('Server restarted:', result);
       this.onServerReady?.();
       return result;

web/src/platform/lifecycle.ts

@@ -3,7 +3,7 @@ import type { PlatformLifecycle } from '@/platform/types';
 class WebLifecycle implements PlatformLifecycle {
   onServerReady?: () => void;
 
-  async startServer(_remote = false): Promise<string> {
+  async startServer(_remote = false, _modelsDir?: string | null): Promise<string> {
     // Web assumes server is running externally
     // Return a default URL - this should be configured via env vars
     const serverUrl = import.meta.env.VITE_SERVER_URL || 'http://localhost:17493';
@@ -15,7 +15,7 @@ class WebLifecycle implements PlatformLifecycle {
     // No-op for web - server is managed externally
   }
 
-  async restartServer(): Promise<string> {
+  async restartServer(_modelsDir?: string | null): Promise<string> {
     // No-op for web - server is managed externally
     return import.meta.env.VITE_SERVER_URL || 'http://localhost:17493';
   }