Add GPU type information to health check response

- Updated the health check endpoint to include the type of GPU available (CUDA or MPS).
- Modified the HealthResponse model to accommodate the new gpu_type field, enhancing the response with detailed GPU information.
- This change improves the clarity of system capabilities for users and developers.

.bumpversion.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 0.1.6
+current_version = 0.1.8
 commit = True
 tag = True
 tag_name = v{new_version}
@@ -34,6 +34,6 @@ replace = "version": "{new_version}"
 search = "version": "{current_version}"
 replace = "version": "{new_version}"
 
-[bumpversion:file:backend/main.py]
+[bumpversion:file:backend/__init__.py]
-search = "version": "{current_version}"
+search = __version__ = "{current_version}"
-replace = "version": "{new_version}"
+replace = __version__ = "{new_version}"

README.md

@@ -22,7 +22,7 @@
 
 <p align="center">
   <a href="https://voicebox.sh">
-    <img src=".github/assets/screenshot.webp" alt="Voicebox App Screenshot" width="800" />
+    <img src="landing/public/assets/app-screenshot-1.webp" alt="Voicebox App Screenshot" width="800" />
   </a>
 </p>
 
@@ -32,17 +32,29 @@
 
 <br/>
 
-## Why Voicebox?
+<p align="center">
+  <img src="landing/public/assets/app-screenshot-2.webp" alt="Voicebox Screenshot 2" width="800" />
+</p>
 
-Voice AI is exploding, but most tools are either cloud-locked, expensive, or a nightmare to set up. Voicebox is different:
+<p align="center">
+  <img src="landing/public/assets/app-screenshot-3.webp" alt="Voicebox Screenshot 3" width="800" />
+</p>
 
-- **100% Local** — Your voice data never leaves your machine
+<br/>
-- **Lightweight** — No bloated Electron, native Tauri performance
-- **Fast** — Near-instant on CUDA, optimized for Apple Silicon
-- **Flexible** — Use the app, integrate the API, or both
-- **Open Source** — No subscriptions, no limits, no lock-in
 
-Built with **Tauri** (Rust), **TypeScript**, **React**, and **Python**. Native performance meets modern DX.
+## What is Voicebox?
+
+Voicebox is a **local-first voice cloning studio** with DAW-like features for professional voice synthesis. Think of it as the **Ollama for voice** — download models, clone voices, and generate speech entirely on your machine.
+
+Unlike cloud services that lock your voice data behind subscriptions, Voicebox gives you:
+
+- **Complete privacy** — models and voice data stay on your machine
+- **Professional tools** — multi-track timeline editor, audio trimming, conversation mixing
+- **Model flexibility** — currently powered by Qwen3-TTS, with support for XTTS, Bark, and other models coming soon
+- **API-first** — use the desktop app or integrate voice synthesis into your own projects
+- **Native performance** — built with Tauri (Rust), not Electron
+
+Download a voice model, clone any voice from a few seconds of audio, and compose multi-voice projects with studio-grade editing tools. No Python install required, no cloud dependency, no limits.
 
 ---
 
@@ -75,6 +87,7 @@ Powered by Alibaba's **Qwen3-TTS** — a breakthrough model that achieves near-p
 
 - **Create profiles** from audio files or record directly in-app
 - **Import/Export** profiles to share or backup
+- **Multi-sample support** — combine multiple samples for higher quality cloning
 - **Organize** with descriptions and language tags
 
 ### Speech Generation
@@ -83,9 +96,19 @@ Powered by Alibaba's **Qwen3-TTS** — a breakthrough model that achieves near-p
 - **Batch generation** for long-form content
 - **Smart caching** — regenerate instantly with voice prompt caching
 
+### Stories Editor
+
+Create multi-voice narratives, podcasts, and conversations with a timeline-based editor.
+
+- **Multi-track composition** — arrange multiple voice tracks in a single project
+- **Inline audio editing** — trim and split clips directly in the timeline
+- **Auto-playback** — preview stories with synchronized playhead
+- **Voice mixing** — build conversations with multiple participants
+
 ### Recording & Transcription
 
 - **In-app recording** with waveform visualization
+- **System audio capture** — record desktop audio on macOS and Windows
 - **Automatic transcription** powered by Whisper
 - **Export recordings** in multiple formats
 

app/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@voicebox/app",
-  "version": "0.1.6",
+  "version": "0.1.8",
   "private": true,
   "type": "module",
   "scripts": {

app/src/components/AudioPlayer/AudioPlayer.tsx

@@ -357,14 +357,22 @@ export function AudioPlayer() {
           }
         }
 
-        // Standard WaveSurfer auto-play
+        // Only auto-play if shouldAutoPlay flag is set (user explicitly clicked to play)
-        // Use a small delay to ensure audio element is fully ready
+        const shouldAutoPlayNow = usePlayerStore.getState().shouldAutoPlay;
-        setTimeout(() => {
+        if (shouldAutoPlayNow) {
-          wavesurfer.play().catch((error) => {
+          // Clear the flag first
-            debug.error('Failed to autoplay:', error);
+          usePlayerStore.getState().clearAutoPlayFlag();
-            // Don't show error for autoplay failures (browser restrictions)
+          
-          });
+          // Use a small delay to ensure audio element is fully ready
-        }, 100);
+          setTimeout(() => {
+            wavesurfer.play().catch((error) => {
+              debug.error('Failed to autoplay:', error);
+              // Don't show error for autoplay failures (browser restrictions)
+            });
+          }, 100);
+        } else {
+          debug.log('Skipping auto-play - shouldAutoPlay is false');
+        }
       });
 
       // Handle play/pause

app/src/components/History/HistoryTable.tsx

@@ -53,7 +53,7 @@ export function HistoryTable() {
   const exportGeneration = useExportGeneration();
   const exportGenerationAudio = useExportGenerationAudio();
   const importGeneration = useImportGeneration();
-  const setAudio = usePlayerStore((state) => state.setAudio);
+  const setAudioWithAutoPlay = usePlayerStore((state) => state.setAudioWithAutoPlay);
   const restartCurrentAudio = usePlayerStore((state) => state.restartCurrentAudio);
   const currentAudioId = usePlayerStore((state) => state.audioId);
   const isPlaying = usePlayerStore((state) => state.isPlaying);
@@ -77,9 +77,9 @@ export function HistoryTable() {
     if (currentAudioId === audioId) {
       restartCurrentAudio();
     } else {
-      // Otherwise, load the new audio
+      // Otherwise, load the new audio and auto-play it
       const audioUrl = apiClient.getAudioUrl(audioId);
-      setAudio(audioUrl, audioId, profileId, text.substring(0, 50));
+      setAudioWithAutoPlay(audioUrl, audioId, profileId, text.substring(0, 50));
     }
   };
 
@@ -233,7 +233,11 @@ export function HistoryTable() {
                   </div>
 
                   {/* Far right - Ellipsis actions */}
-                  <div className="w-10 shrink-0 flex justify-end">
+                  <div
+                    className="w-10 shrink-0 flex justify-end"
+                    onMouseDown={(e) => e.stopPropagation()}
+                    onClick={(e) => e.stopPropagation()}
+                  >
                     <DropdownMenu>
                       <DropdownMenuTrigger asChild>
                         <Button
@@ -241,7 +245,6 @@ export function HistoryTable() {
                           size="icon"
                           className="h-8 w-8"
                           aria-label="Actions"
-                          onClick={(e) => e.stopPropagation()}
                         >
                           <MoreHorizontal className="h-4 w-4" />
                         </Button>

app/src/components/StoriesTab/StoryContent.tsx

@@ -131,13 +131,13 @@ export function StoryContent() {
     }
   }, [isPlaying]);
 
-  const handleRemoveItem = (generationId: string) => {
+  const handleRemoveItem = (itemId: string) => {
     if (!story) return;
 
     removeItem.mutate(
       {
         storyId: story.id,
-        generationId,
+        itemId,
       },
       {
         onError: (error) => {
@@ -360,7 +360,7 @@ export function StoryContent() {
                       item={item}
                       storyId={story.id}
                       index={index}
-                      onRemove={() => handleRemoveItem(item.generation_id)}
+                      onRemove={() => handleRemoveItem(item.id)}
                       currentTimeMs={currentTimeMs}
                       isPlaying={isPlaying && playbackStoryId === story.id}
                     />

app/src/components/StoriesTab/StoryTrackEditor.tsx

@@ -1,21 +1,59 @@
-import { GripHorizontal, Minus, Pause, Play, Plus, Square } from 'lucide-react';
+import {
+  Copy,
+  GripHorizontal,
+  Minus,
+  Pause,
+  Play,
+  Plus,
+  Scissors,
+  Square,
+  Trash2,
+} from 'lucide-react';
 import { useCallback, useEffect, useMemo, useRef, useState } from 'react';
 import WaveSurfer from 'wavesurfer.js';
 import { Button } from '@/components/ui/button';
 import { useToast } from '@/components/ui/use-toast';
 import { apiClient } from '@/lib/api/client';
-import { useMoveStoryItem } from '@/lib/hooks/useStories';
-import { useStoryStore } from '@/stores/storyStore';
 import type { StoryItemDetail } from '@/lib/api/types';
+import {
+  useDuplicateStoryItem,
+  useMoveStoryItem,
+  useRemoveStoryItem,
+  useSplitStoryItem,
+  useTrimStoryItem,
+} from '@/lib/hooks/useStories';
 import { cn } from '@/lib/utils/cn';
+import { useStoryStore } from '@/stores/storyStore';
 
-// Clip waveform component
+// Clip waveform component with trim support
-function ClipWaveform({ generationId, width }: { generationId: string; width: number }) {
+function ClipWaveform({
-  const containerRef = useRef<HTMLDivElement>(null);
+  generationId,
+  width,
+  trimStartMs,
+  trimEndMs,
+  duration,
+}: {
+  generationId: string;
+  width: number;
+  trimStartMs: number;
+  trimEndMs: number;
+  duration: number;
+}) {
+  const waveformRef = useRef<HTMLDivElement>(null);
   const wavesurferRef = useRef<WaveSurfer | null>(null);
 
+  // Calculate the full waveform width based on the original duration
+  // The visible portion (width) represents the effective duration after trimming
+  const effectiveDurationMs = duration * 1000 - trimStartMs - trimEndMs;
+  const fullWaveformWidth =
+    effectiveDurationMs > 0 ? (width / effectiveDurationMs) * (duration * 1000) : width;
+
+  // Calculate how much to offset the waveform to hide the trimmed start
+  const offsetX =
+    effectiveDurationMs > 0 ? (trimStartMs / (duration * 1000)) * fullWaveformWidth : 0;
+
   useEffect(() => {
-    if (!containerRef.current || width < 20) return;
+    if (!waveformRef.current || fullWaveformWidth < 20) return;
 
     // Get CSS colors
     const root = document.documentElement;
@@ -27,7 +65,7 @@ function ClipWaveform({ generationId, width }: { generationId: string; width: nu
     const waveColor = getCSSVar('--accent-foreground');
 
     const wavesurfer = WaveSurfer.create({
-      container: containerRef.current,
+      container: waveformRef.current,
       waveColor,
       progressColor: waveColor,
       cursorWidth: 0,
@@ -50,9 +88,21 @@ function ClipWaveform({ generationId, width }: { generationId: string; width: nu
       wavesurfer.destroy();
       wavesurferRef.current = null;
     };
-  }, [generationId, width]);
+  }, [generationId, fullWaveformWidth]);
 
-  return <div ref={containerRef} className="w-full h-full opacity-60" />;
+  return (
+    <div className="w-full h-full opacity-60 overflow-hidden">
+      {/* Inner container that holds the full waveform, offset to show only visible portion */}
+      <div
+        ref={waveformRef}
+        style={{
+          width: `${fullWaveformWidth}px`,
+          transform: `translateX(-${offsetX}px)`,
+        }}
+        className="h-full"
+      />
+    </div>
+  );
 }
 
 interface StoryTrackEditorProps {
@@ -80,8 +130,25 @@ export function StoryTrackEditor({ storyId, items }: StoryTrackEditorProps) {
   const resizeStartY = useRef(0);
   const resizeStartHeight = useRef(0);
   const moveItem = useMoveStoryItem();
+  const trimItem = useTrimStoryItem();
+  const splitItem = useSplitStoryItem();
+  const duplicateItem = useDuplicateStoryItem();
+  const removeItem = useRemoveStoryItem();
   const { toast } = useToast();
 
+  // Selection state
+  const selectedClipId = useStoryStore((state) => state.selectedClipId);
+  const setSelectedClipId = useStoryStore((state) => state.setSelectedClipId);
+
+  // Trim state
+  const [trimmingItem, setTrimmingItem] = useState<string | null>(null);
+  const [trimSide, setTrimSide] = useState<'start' | 'end' | null>(null);
+  const [trimStartX, setTrimStartX] = useState(0);
+  const [tempTrimValues, setTempTrimValues] = useState<{
+    trim_start_ms: number;
+    trim_end_ms: number;
+  } | null>(null);
+
   // Track editor height from store (shared with FloatingGenerateBox)
   const editorHeight = useStoryStore((state) => state.trackEditorHeight);
   const setEditorHeight = useStoryStore((state) => state.setTrackEditorHeight);
@@ -89,16 +156,32 @@ export function StoryTrackEditor({ storyId, items }: StoryTrackEditorProps) {
   // Playback state
   const isPlaying = useStoryStore((state) => state.isPlaying);
   const currentTimeMs = useStoryStore((state) => state.currentTimeMs);
-  const storeTotalDurationMs = useStoryStore((state) => state.totalDurationMs);
   const playbackStoryId = useStoryStore((state) => state.playbackStoryId);
   const play = useStoryStore((state) => state.play);
   const pause = useStoryStore((state) => state.pause);
   const stop = useStoryStore((state) => state.stop);
   const seek = useStoryStore((state) => state.seek);
+  const setActiveStory = useStoryStore((state) => state.setActiveStory);
 
   const isActiveStory = playbackStoryId === storyId;
   const isCurrentlyPlaying = isPlaying && isActiveStory;
 
+  // Auto-activate this story when the editor is shown so playhead is visible
+  useEffect(() => {
+    if (items.length > 0 && !isActiveStory) {
+      const totalDuration = Math.max(
+        ...items.map((item) => {
+          const trimStart = item.trim_start_ms || 0;
+          const trimEnd = item.trim_end_ms || 0;
+          const effectiveDuration = item.duration * 1000 - trimStart - trimEnd;
+          return item.start_time_ms + effectiveDuration;
+        }),
+        0,
+      );
+      setActiveStory(storyId, items, totalDuration);
+    }
+  }, [storyId, items, isActiveStory, setActiveStory]);
+
   // Sort items by start time for play
   const sortedItems = useMemo(() => {
     return [...items].sort((a, b) => a.start_time_ms - b.start_time_ms);
@@ -140,14 +223,16 @@ export function StoryTrackEditor({ storyId, items }: StoryTrackEditorProps) {
     return () => observer.disconnect();
   }, []);
 
-  // Calculate total duration
+  // Calculate effective duration (accounting for trims)
+  const getEffectiveDuration = (item: StoryItemDetail) => {
+    return item.duration * 1000 - (item.trim_start_ms || 0) - (item.trim_end_ms || 0);
+  };
+
+  // Calculate total duration (using effective durations)
   const totalDurationMs = useMemo(() => {
     if (items.length === 0) return 10000; // Default 10 seconds
-    return Math.max(
+    return Math.max(...items.map((item) => item.start_time_ms + getEffectiveDuration(item)), 10000);
-      ...items.map((item) => item.start_time_ms + item.duration * 1000),
+  }, [items, getEffectiveDuration]);
-      10000
-    );
-  }, [items]);
 
   // Calculate timeline width - at least full container width
   const contentWidth = (totalDurationMs / 1000) * pixelsPerSecond + 200; // Content width with padding
@@ -175,15 +260,9 @@ export function StoryTrackEditor({ storyId, items }: StoryTrackEditorProps) {
     return `${minutes}:${seconds.toString().padStart(2, '0')}`;
   };
 
-  const msToPixels = useCallback(
+  const msToPixels = useCallback((ms: number) => (ms / 1000) * pixelsPerSecond, [pixelsPerSecond]);
-    (ms: number) => (ms / 1000) * pixelsPerSecond,
-    [pixelsPerSecond]
-  );
 
-  const pixelsToMs = useCallback(
+  const pixelsToMs = useCallback((px: number) => (px / pixelsPerSecond) * 1000, [pixelsPerSecond]);
-    (px: number) => (px / pixelsPerSecond) * 1000,
-    [pixelsPerSecond]
-  );
 
   const handleZoomIn = () => {
     setPixelsPerSecond((prev) => Math.min(prev * 1.5, MAX_PIXELS_PER_SECOND));
@@ -194,22 +273,28 @@ export function StoryTrackEditor({ storyId, items }: StoryTrackEditorProps) {
   };
 
   // Resize handlers
-  const handleResizeStart = useCallback((e: React.MouseEvent) => {
+  const handleResizeStart = useCallback(
-    e.preventDefault();
+    (e: React.MouseEvent) => {
-    setIsResizing(true);
+      e.preventDefault();
-    resizeStartY.current = e.clientY;
+      setIsResizing(true);
-    resizeStartHeight.current = editorHeight;
+      resizeStartY.current = e.clientY;
-  }, [editorHeight]);
+      resizeStartHeight.current = editorHeight;
+    },
+    [editorHeight],
+  );
 
-  const handleResizeMove = useCallback((e: MouseEvent) => {
+  const handleResizeMove = useCallback(
-    if (!isResizing) return;
+    (e: MouseEvent) => {
-    const deltaY = resizeStartY.current - e.clientY;
+      if (!isResizing) return;
-    const newHeight = Math.min(
+      const deltaY = resizeStartY.current - e.clientY;
-      MAX_EDITOR_HEIGHT,
+      const newHeight = Math.min(
-      Math.max(MIN_EDITOR_HEIGHT, resizeStartHeight.current + deltaY)
+        MAX_EDITOR_HEIGHT,
-    );
+        Math.max(MIN_EDITOR_HEIGHT, resizeStartHeight.current + deltaY),
-    setEditorHeight(newHeight);
+      );
-  }, [isResizing, setEditorHeight]);
+      setEditorHeight(newHeight);
+    },
+    [isResizing, setEditorHeight],
+  );
 
   const handleResizeEnd = useCallback(() => {
     setIsResizing(false);
@@ -228,17 +313,269 @@ export function StoryTrackEditor({ storyId, items }: StoryTrackEditorProps) {
   }, [isResizing, handleResizeMove, handleResizeEnd]);
 
   const handleTimelineClick = (e: React.MouseEvent<HTMLDivElement>) => {
-    if (!tracksRef.current || draggingItem) return;
+    if (!tracksRef.current || draggingItem || trimmingItem) return;
     const rect = tracksRef.current.getBoundingClientRect();
     const x = e.clientX - rect.left + tracksRef.current.scrollLeft;
     const timeMs = Math.max(0, pixelsToMs(x));
     seek(timeMs);
+    // Deselect clip when clicking on timeline
+    setSelectedClipId(null);
   };
 
-  const handleDragStart = (
+  const handleClipClick = (e: React.MouseEvent, item: StoryItemDetail) => {
-    e: React.MouseEvent,
+    e.stopPropagation();
-    item: StoryItemDetail
+    if (draggingItem || trimmingItem) return;
-  ) => {
+    setSelectedClipId(item.id);
+  };
+
+  const handleTrimStart = (e: React.MouseEvent, item: StoryItemDetail, side: 'start' | 'end') => {
+    e.stopPropagation();
+    if (!tracksRef.current) return;
+    setTrimmingItem(item.id);
+    setTrimSide(side);
+    setSelectedClipId(item.id);
+    setTrimStartX(e.clientX);
+    trimStartItemRef.current = {
+      item,
+      initialTrimStart: item.trim_start_ms || 0,
+      initialTrimEnd: item.trim_end_ms || 0,
+    };
+  };
+
+  const trimStartItemRef = useRef<{
+    item: StoryItemDetail;
+    initialTrimStart: number;
+    initialTrimEnd: number;
+  } | null>(null);
+
+  const handleTrimMove = useCallback(
+    (e: MouseEvent) => {
+      if (!trimmingItem || !trimSide || !trimStartItemRef.current) return;
+
+      const deltaX = e.clientX - trimStartX;
+      const deltaMs = pixelsToMs(deltaX); // Signed delta in milliseconds
+
+      const { item, initialTrimStart, initialTrimEnd } = trimStartItemRef.current;
+      const originalDurationMs = item.duration * 1000;
+
+      let newTrimStart = initialTrimStart;
+      let newTrimEnd = initialTrimEnd;
+
+      if (trimSide === 'start') {
+        // Moving right increases trim_start (trims more from start)
+        // Moving left decreases trim_start (restores from start)
+        newTrimStart = Math.round(
+          Math.max(
+            0,
+            Math.min(initialTrimStart + deltaMs, originalDurationMs - initialTrimEnd - 100),
+          ),
+        );
+      } else {
+        // Moving right decreases trim_end (restores from end)
+        // Moving left increases trim_end (trims more from end)
+        newTrimEnd = Math.round(
+          Math.max(
+            0,
+            Math.min(initialTrimEnd - deltaMs, originalDurationMs - initialTrimStart - 100),
+          ),
+        );
+      }
+
+      // Validate that we don't exceed duration
+      if (newTrimStart + newTrimEnd >= originalDurationMs - 100) {
+        return; // Don't allow trimming to less than 100ms
+      }
+
+      // Update temporary trim values for visual feedback
+      setTempTrimValues({
+        trim_start_ms: newTrimStart,
+        trim_end_ms: newTrimEnd,
+      });
+    },
+    [trimmingItem, trimSide, trimStartX, pixelsToMs],
+  );
+
+  const handleTrimEnd = useCallback(() => {
+    if (!trimmingItem || !trimSide || !trimStartItemRef.current) {
+      setTrimmingItem(null);
+      setTrimSide(null);
+      setTempTrimValues(null);
+      trimStartItemRef.current = null;
+      return;
+    }
+
+    const { initialTrimStart, initialTrimEnd } = trimStartItemRef.current;
+
+    // Use temporary trim values if available, otherwise use initial values
+    // Ensure values are integers for the backend
+    const finalTrimStart = Math.round(tempTrimValues?.trim_start_ms ?? initialTrimStart);
+    const finalTrimEnd = Math.round(tempTrimValues?.trim_end_ms ?? initialTrimEnd);
+
+    // Only update if values changed
+    if (finalTrimStart !== initialTrimStart || finalTrimEnd !== initialTrimEnd) {
+      trimItem.mutate(
+        {
+          storyId,
+          itemId: trimmingItem,
+          data: {
+            trim_start_ms: finalTrimStart,
+            trim_end_ms: finalTrimEnd,
+          },
+        },
+        {
+          onError: (error) => {
+            toast({
+              title: 'Failed to trim clip',
+              description: error instanceof Error ? error.message : String(error),
+              variant: 'destructive',
+            });
+          },
+        },
+      );
+    }
+
+    setTrimmingItem(null);
+    setTrimSide(null);
+    setTempTrimValues(null);
+    trimStartItemRef.current = null;
+  }, [trimmingItem, trimSide, tempTrimValues, storyId, trimItem, toast]);
+
+  const handleSplit = useCallback(() => {
+    if (!selectedClipId) return;
+
+    const item = items.find((i) => i.id === selectedClipId);
+    if (!item) return;
+
+    const splitTimeMs = currentTimeMs - item.start_time_ms;
+    const effectiveDuration = getEffectiveDuration(item);
+
+    if (splitTimeMs <= 0 || splitTimeMs >= effectiveDuration) {
+      toast({
+        title: 'Invalid split point',
+        description: 'Playhead must be within the selected clip',
+        variant: 'destructive',
+      });
+      return;
+    }
+
+    splitItem.mutate(
+      {
+        storyId,
+        itemId: selectedClipId,
+        data: { split_time_ms: splitTimeMs },
+      },
+      {
+        onSuccess: () => {
+          setSelectedClipId(null);
+        },
+        onError: (error) => {
+          toast({
+            title: 'Failed to split clip',
+            description: error instanceof Error ? error.message : String(error),
+            variant: 'destructive',
+          });
+        },
+      },
+    );
+  }, [
+    selectedClipId,
+    items,
+    currentTimeMs,
+    getEffectiveDuration,
+    storyId,
+    splitItem,
+    toast,
+    setSelectedClipId,
+  ]);
+
+  const handleDuplicate = useCallback(() => {
+    if (!selectedClipId) return;
+
+    duplicateItem.mutate(
+      {
+        storyId,
+        itemId: selectedClipId,
+      },
+      {
+        onError: (error) => {
+          toast({
+            title: 'Failed to duplicate clip',
+            description: error instanceof Error ? error.message : String(error),
+            variant: 'destructive',
+          });
+        },
+      },
+    );
+  }, [selectedClipId, storyId, duplicateItem, toast]);
+
+  const handleDelete = useCallback(() => {
+    if (!selectedClipId) return;
+
+    removeItem.mutate(
+      {
+        storyId,
+        itemId: selectedClipId,
+      },
+      {
+        onSuccess: () => {
+          setSelectedClipId(null);
+        },
+        onError: (error) => {
+          toast({
+            title: 'Failed to delete clip',
+            description: error instanceof Error ? error.message : String(error),
+            variant: 'destructive',
+          });
+        },
+      },
+    );
+  }, [selectedClipId, storyId, removeItem, toast, setSelectedClipId]);
+
+  // Keyboard shortcuts
+  useEffect(() => {
+    const handleKeyDown = (e: KeyboardEvent) => {
+      // Only handle shortcuts when editor is focused or no input is focused
+      if (e.target instanceof HTMLInputElement || e.target instanceof HTMLTextAreaElement) {
+        return;
+      }
+
+      if (e.key === 'Escape') {
+        setSelectedClipId(null);
+      } else if (e.key === 's' || e.key === 'S') {
+        if (selectedClipId) {
+          e.preventDefault();
+          handleSplit();
+        }
+      } else if (e.key === 'd' || e.key === 'D') {
+        if (selectedClipId && (e.metaKey || e.ctrlKey)) {
+          e.preventDefault();
+          handleDuplicate();
+        }
+      } else if (e.key === 'Delete' || e.key === 'Backspace') {
+        if (selectedClipId) {
+          e.preventDefault();
+          handleDelete();
+        }
+      }
+    };
+
+    window.addEventListener('keydown', handleKeyDown);
+    return () => window.removeEventListener('keydown', handleKeyDown);
+  }, [selectedClipId, handleSplit, handleDuplicate, handleDelete, setSelectedClipId]);
+
+  // Add global mouse listeners for trimming
+  useEffect(() => {
+    if (trimmingItem) {
+      window.addEventListener('mousemove', handleTrimMove);
+      window.addEventListener('mouseup', handleTrimEnd);
+      return () => {
+        window.removeEventListener('mousemove', handleTrimMove);
+        window.removeEventListener('mouseup', handleTrimEnd);
+      };
+    }
+  }, [trimmingItem, handleTrimMove, handleTrimEnd]);
+
+  const handleDragStart = (e: React.MouseEvent, item: StoryItemDetail) => {
     e.stopPropagation();
     if (!tracksRef.current) return;
 
@@ -251,7 +588,7 @@ export function StoryTrackEditor({ storyId, items }: StoryTrackEditorProps) {
       x: rect.left - tracksRef.current.getBoundingClientRect().left + tracksRef.current.scrollLeft,
       y: rect.top - tracksRef.current.getBoundingClientRect().top,
     });
-    setDraggingItem(item.generation_id);
+    setDraggingItem(item.id);
   };
 
   const handleDragMove = useCallback(
@@ -264,7 +601,7 @@ export function StoryTrackEditor({ storyId, items }: StoryTrackEditorProps) {
 
       setDragPosition({ x: Math.max(0, x), y });
     },
-    [draggingItem, dragOffset]
+    [draggingItem, dragOffset],
   );
 
   const handleDragEnd = useCallback(() => {
@@ -273,7 +610,7 @@ export function StoryTrackEditor({ storyId, items }: StoryTrackEditorProps) {
       return;
     }
 
-    const item = items.find((i) => i.generation_id === draggingItem);
+    const item = items.find((i) => i.id === draggingItem);
     if (!item) {
       setDraggingItem(null);
       return;
@@ -292,7 +629,7 @@ export function StoryTrackEditor({ storyId, items }: StoryTrackEditorProps) {
       moveItem.mutate(
         {
           storyId,
-          generationId: item.generation_id,
+          itemId: item.id,
           data: {
             start_time_ms: newTimeMs,
             track: newTrack,
@@ -302,11 +639,11 @@ export function StoryTrackEditor({ storyId, items }: StoryTrackEditorProps) {
           onError: (error) => {
             toast({
               title: 'Failed to move item',
-              description: error.message,
+              description: error instanceof Error ? error.message : String(error),
               variant: 'destructive',
             });
           },
-        }
+        },
       );
     }
 
@@ -318,9 +655,10 @@ export function StoryTrackEditor({ storyId, items }: StoryTrackEditorProps) {
 
   // Calculate clip position and dimensions
   const getClipStyle = (item: StoryItemDetail) => {
-    const isDragging = draggingItem === item.generation_id;
+    const isDragging = draggingItem === item.id;
     const trackIndex = getTrackIndex(item.track);
-    const width = msToPixels(item.duration * 1000);
+    const effectiveDuration = getEffectiveDuration(item);
+    const width = msToPixels(effectiveDuration);
     const left = isDragging ? dragPosition.x : msToPixels(item.start_time_ms);
     const top = isDragging ? dragPosition.y : trackIndex * TRACK_HEIGHT;
 
@@ -345,176 +683,271 @@ export function StoryTrackEditor({ storyId, items }: StoryTrackEditorProps) {
 
   return (
     <div className="fixed bottom-0 left-0 right-0 border-t bg-background/95 backdrop-blur supports-backdrop-filter:bg-background/60 z-50">
-      <div className="border-t bg-background/30 backdrop-blur-2xl overflow-hidden relative" ref={containerRef}>
+      <div
-      {/* Resize handle at top */}
+        className="border-t bg-background/30 backdrop-blur-2xl overflow-hidden relative"
-      <button
+        ref={containerRef}
-        type="button"
-        className="absolute top-0 left-0 right-0 h-2 cursor-ns-resize flex items-center justify-center hover:bg-muted/50 transition-colors z-20 group"
-        onMouseDown={handleResizeStart}
-        aria-label="Resize track editor"
       >
-        <GripHorizontal className="h-3 w-3 text-muted-foreground/50 group-hover:text-muted-foreground" />
+        {/* Resize handle at top */}
-      </button>
+        <button
-
+          type="button"
-      {/* Toolbar */}
+          className="absolute top-0 left-0 right-0 h-2 cursor-ns-resize flex items-center justify-center hover:bg-muted/50 transition-colors z-20 group"
-      <div className="flex items-center justify-between px-3 py-2 border-b bg-muted/30 mt-2">
+          onMouseDown={handleResizeStart}
-        {/* Play controls - left side */}
+          aria-label="Resize track editor"
-        <div className="flex items-center gap-2">
-          <Button variant="ghost" size="icon" className="h-7 w-7" onClick={handlePlayPause}>
-            {isCurrentlyPlaying ? (
-              <Pause className="h-4 w-4" />
-            ) : (
-              <Play className="h-4 w-4" />
-            )}
-          </Button>
-          <Button variant="ghost" size="icon" className="h-7 w-7" onClick={handleStop} disabled={!isActiveStory}>
-            <Square className="h-3 w-3" />
-          </Button>
-          <span className="text-xs text-muted-foreground tabular-nums ml-2">
-            {formatTime(isActiveStory ? currentTimeMs : 0)} / {formatTime(isActiveStory ? storeTotalDurationMs : 0)}
-          </span>
-        </div>
-
-        {/* Zoom controls - right side */}
-        <div className="flex items-center gap-2">
-          <span className="text-xs text-muted-foreground">Zoom:</span>
-          <Button variant="ghost" size="icon" className="h-6 w-6" onClick={handleZoomOut}>
-            <Minus className="h-3 w-3" />
-          </Button>
-          <Button variant="ghost" size="icon" className="h-6 w-6" onClick={handleZoomIn}>
-            <Plus className="h-3 w-3" />
-          </Button>
-        </div>
-      </div>
-
-      {/* Timeline container with track labels sidebar */}
-      <div className="flex" style={{ height: `${timelineContainerHeight}px` }}>
-        {/* Track labels sidebar - fixed width */}
-        <div className="w-16 shrink-0 border-r bg-muted/20 overflow-hidden">
-          {/* Spacer for time ruler */}
-          <div className="h-6 border-b bg-muted/30" />
-          {/* Track labels */}
-          <div style={{ height: `${tracksAreaHeight}px` }}>
-            {tracks.map((trackNumber, index) => (
-              <div
-                key={trackNumber}
-                className={cn(
-                  'border-b flex items-center justify-center',
-                  index % 2 === 0 ? 'bg-background' : 'bg-muted/10'
-                )}
-                style={{ height: `${TRACK_HEIGHT}px` }}
-              >
-                <span className="text-[10px] text-muted-foreground select-none">
-                  {trackNumber}
-                </span>
-              </div>
-            ))}
-          </div>
-        </div>
-
-        {/* Scrollable timeline area */}
-        {/* biome-ignore lint/a11y/noStaticElementInteractions: Container handles drag events for child clips */}
-        <div
-          ref={tracksRef}
-          className="overflow-auto relative flex-1"
-          onMouseMove={draggingItem ? handleDragMove : undefined}
-          onMouseUp={draggingItem ? handleDragEnd : undefined}
-          onMouseLeave={draggingItem ? handleDragEnd : undefined}
         >
-          {/* Time ruler */}
+          <GripHorizontal className="h-3 w-3 text-muted-foreground/50 group-hover:text-muted-foreground" />
-          <div
+        </button>
-            className="h-6 border-b bg-muted/20 sticky top-0 z-10"
+
-            style={{ width: `${timelineWidth}px` }}
+        {/* Toolbar */}
-          >
+        <div className="flex items-center justify-between px-3 py-2 border-b bg-muted/30 mt-2">
-            {timeMarkers.map((ms) => (
+          {/* Play controls - left side */}
-              <div
+          <div className="flex items-center gap-2">
-                key={ms}
+            <Button variant="ghost" size="icon" className="h-7 w-7" onClick={handlePlayPause}>
-                className="absolute top-0 h-full flex flex-col justify-end"
+              {isCurrentlyPlaying ? <Pause className="h-4 w-4" /> : <Play className="h-4 w-4" />}
-                style={{ left: `${msToPixels(ms)}px` }}
+            </Button>
-              >
+            <Button
-                <div className="h-2 w-px bg-border" />
+              variant="ghost"
-                <span className="text-[10px] text-muted-foreground ml-1 select-none">
+              size="icon"
-                  {formatTime(ms)}
+              className="h-7 w-7"
-                </span>
+              onClick={handleStop}
-              </div>
+              disabled={!isCurrentlyPlaying}
-            ))}
+            >
+              <Square className="h-3 w-3" />
+            </Button>
+            <span className="text-xs text-muted-foreground tabular-nums ml-2">
+              {formatTime(currentTimeMs)} / {formatTime(totalDurationMs)}
+            </span>
           </div>
 
-          {/* Tracks area */}
+          {/* Clip editing controls - center */}
-          <div
+          {selectedClipId && (
-            className="relative"
+            <div className="flex items-center gap-1">
-            style={{ width: `${timelineWidth}px`, height: `${tracksAreaHeight}px` }}
+              <Button
-          >
+                variant="ghost"
-            {/* Track backgrounds */}
+                size="icon"
-            {tracks.map((trackNumber, index) => (
+                className="h-7 w-7"
-              <div
+                onClick={handleSplit}
-                key={trackNumber}
+                title="Split at playhead (S)"
-                className={cn(
-                  'absolute left-0 right-0 border-b',
-                  index % 2 === 0 ? 'bg-background' : 'bg-muted/10'
-                )}
-                style={{
-                  top: `${index * TRACK_HEIGHT}px`,
-                  height: `${TRACK_HEIGHT}px`,
-                }}
-              />
-            ))}
-
-            {/* Click area for seeking - z-index lower than clips */}
-            <button
-              type="button"
-              className="absolute inset-0 z-0 cursor-pointer"
-              onClick={handleTimelineClick}
-              aria-label="Seek timeline"
-            />
-
-          {/* Audio clips */}
-          {items.map((item) => {
-            const isDragging = draggingItem === item.generation_id;
-            const style = getClipStyle(item);
-            const clipWidth = msToPixels(item.duration * 1000);
-
-            return (
-              <button
-                type="button"
-                key={item.generation_id}
-                className={cn(
-                  'absolute rounded cursor-move select-none overflow-hidden z-10',
-                  'bg-accent/80 hover:bg-accent border border-accent-foreground/20',
-                  'flex flex-col justify-center',
-                  isDragging && 'opacity-80 shadow-lg z-20',
-                  !isDragging && 'transition-all duration-100'
-                )}
-                style={style}
-                onMouseDown={(e) => handleDragStart(e, item)}
               >
-                {/* Clip label */}
+                <Scissors className="h-4 w-4" />
-                <div className="absolute top-0 left-1 right-1 z-10">
+              </Button>
-                  <p className="text-[9px] font-medium text-accent-foreground truncate">
+              <Button
-                    {item.profile_name}
+                variant="ghost"
-                  </p>
+                size="icon"
-                </div>
+                className="h-7 w-7"
-                {/* Waveform */}
+                onClick={handleDuplicate}
-                <div className="absolute inset-0 top-3">
+                title="Duplicate (Cmd/Ctrl+D)"
-                  <ClipWaveform generationId={item.generation_id} width={clipWidth} />
+              >
-                </div>
+                <Copy className="h-4 w-4" />
-              </button>
+              </Button>
-            );
+              <Button
-          })}
+                variant="ghost"
-
+                size="icon"
-          {/* Playhead */}
+                className="h-7 w-7"
-          {isActiveStory && (
+                onClick={handleDelete}
-            <div
+                title="Delete (Delete/Backspace)"
-              className="absolute top-0 bottom-0 w-1 bg-accent z-30 pointer-events-none rounded-full"
+              >
-              style={{ left: `${playheadLeft}px` }}
+                <Trash2 className="h-4 w-4" />
-            >
+              </Button>
-              <div className="absolute -top-1 left-1/2 -translate-x-1/2 w-3 h-3 bg-accent rounded-full" />
             </div>
           )}
+
+          {/* Zoom controls - right side */}
+          <div className="flex items-center gap-2">
+            <span className="text-xs text-muted-foreground">Zoom:</span>
+            <Button variant="ghost" size="icon" className="h-6 w-6" onClick={handleZoomOut}>
+              <Minus className="h-3 w-3" />
+            </Button>
+            <Button variant="ghost" size="icon" className="h-6 w-6" onClick={handleZoomIn}>
+              <Plus className="h-3 w-3" />
+            </Button>
+          </div>
         </div>
+
+        {/* Timeline container with track labels sidebar */}
+        <div className="flex" style={{ height: `${timelineContainerHeight}px` }}>
+          {/* Track labels sidebar - fixed width */}
+          <div className="w-16 shrink-0 border-r bg-muted/20 overflow-hidden">
+            {/* Spacer for time ruler */}
+            <div className="h-6 border-b bg-muted/30" />
+            {/* Track labels */}
+            <div style={{ height: `${tracksAreaHeight}px` }}>
+              {tracks.map((trackNumber, index) => (
+                <div
+                  key={trackNumber}
+                  className={cn(
+                    'border-b flex items-center justify-center',
+                    index % 2 === 0 ? 'bg-background' : 'bg-muted/10',
+                  )}
+                  style={{ height: `${TRACK_HEIGHT}px` }}
+                >
+                  <span className="text-[10px] text-muted-foreground select-none">
+                    {trackNumber}
+                  </span>
+                </div>
+              ))}
+            </div>
+          </div>
+
+          {/* Scrollable timeline area */}
+          {/* biome-ignore lint/a11y/noStaticElementInteractions: Container handles drag events for child clips */}
+          <div
+            ref={tracksRef}
+            className="overflow-auto relative flex-1"
+            onMouseMove={draggingItem ? handleDragMove : undefined}
+            onMouseUp={draggingItem ? handleDragEnd : undefined}
+            onMouseLeave={draggingItem ? handleDragEnd : undefined}
+          >
+            {/* Time ruler - clickable to seek */}
+            <button
+              type="button"
+              className="h-6 border-b bg-muted/20 sticky top-0 z-10 cursor-pointer text-left"
+              style={{ width: `${timelineWidth}px` }}
+              onClick={handleTimelineClick}
+              aria-label="Seek timeline"
+            >
+              {timeMarkers.map((ms) => (
+                <div
+                  key={ms}
+                  className="absolute top-0 h-full flex flex-col justify-end pointer-events-none"
+                  style={{ left: `${msToPixels(ms)}px` }}
+                >
+                  <div className="h-2 w-px bg-border" />
+                  <span className="text-[10px] text-muted-foreground ml-1 select-none">
+                    {formatTime(ms)}
+                  </span>
+                </div>
+              ))}
+            </button>
+
+            {/* Tracks area */}
+            <div
+              className="relative"
+              style={{ width: `${timelineWidth}px`, height: `${tracksAreaHeight}px` }}
+            >
+              {/* Track backgrounds - pointer-events-none to allow clicks to pass through */}
+              {tracks.map((trackNumber, index) => (
+                <div
+                  key={trackNumber}
+                  className={cn(
+                    'absolute left-0 right-0 border-b pointer-events-none',
+                    index % 2 === 0 ? 'bg-background' : 'bg-muted/10',
+                  )}
+                  style={{
+                    top: `${index * TRACK_HEIGHT}px`,
+                    height: `${TRACK_HEIGHT}px`,
+                  }}
+                />
+              ))}
+
+              {/* Click area for seeking - z-index lower than clips */}
+              <button
+                type="button"
+                className="absolute inset-0 z-0 cursor-pointer"
+                onClick={handleTimelineClick}
+                aria-label="Seek timeline"
+              />
+
+              {/* Audio clips */}
+              {items.map((item) => {
+                const isDragging = draggingItem === item.id;
+                const isSelected = selectedClipId === item.id;
+                const isTrimming = trimmingItem === item.id;
+
+                // Use temporary trim values during trimming for visual feedback
+                const displayTrimStart =
+                  isTrimming && tempTrimValues
+                    ? tempTrimValues.trim_start_ms
+                    : item.trim_start_ms || 0;
+                const displayTrimEnd =
+                  isTrimming && tempTrimValues ? tempTrimValues.trim_end_ms : item.trim_end_ms || 0;
+                const effectiveDuration = item.duration * 1000 - displayTrimStart - displayTrimEnd;
+
+                const style = getClipStyle({
+                  ...item,
+                  trim_start_ms: displayTrimStart,
+                  trim_end_ms: displayTrimEnd,
+                });
+                const clipWidth = msToPixels(effectiveDuration);
+
+                return (
+                  <div
+                    key={item.id}
+                    className={cn(
+                      'absolute rounded select-none overflow-visible z-10',
+                      isSelected && 'ring-2 ring-primary ring-offset-1',
+                      isTrimming && 'ring-2 ring-accent',
+                    )}
+                    style={style}
+                  >
+                    <button
+                      type="button"
+                      className={cn(
+                        'w-full h-full rounded cursor-move overflow-hidden',
+                        'bg-accent/80 hover:bg-accent border border-accent-foreground/20',
+                        'flex flex-col justify-center',
+                        isDragging && 'opacity-80 shadow-lg z-20',
+                        !isDragging && 'transition-all duration-100',
+                      )}
+                      onClick={(e) => handleClipClick(e, item)}
+                      onMouseDown={(e) => {
+                        // Only start drag if not clicking on trim handles
+                        if (!(e.target as HTMLElement).closest('.trim-handle')) {
+                          handleDragStart(e, item);
+                        }
+                      }}
+                    >
+                      {/* Clip label */}
+                      <div className="absolute top-0 left-1 right-1 z-10">
+                        <p className="text-[9px] font-medium text-accent-foreground truncate">
+                          {item.profile_name}
+                        </p>
+                      </div>
+                      {/* Waveform */}
+                      <div className="absolute inset-0 top-3">
+                        <ClipWaveform
+                          generationId={item.generation_id}
+                          width={clipWidth}
+                          trimStartMs={displayTrimStart}
+                          trimEndMs={displayTrimEnd}
+                          duration={item.duration}
+                        />
+                      </div>
+                    </button>
+
+                    {/* Trim handles */}
+                    {isSelected && (
+                      <>
+                        {/* Left trim handle */}
+                        <button
+                          type="button"
+                          className="trim-handle absolute left-0 top-0 bottom-0 w-2 cursor-ew-resize hover:bg-primary/30 bg-primary/20 z-30 rounded-l"
+                          onMouseDown={(e) => handleTrimStart(e, item, 'start')}
+                          aria-label="Trim start"
+                        />
+                        {/* Right trim handle */}
+                        <button
+                          type="button"
+                          className="trim-handle absolute right-0 top-0 bottom-0 w-2 cursor-ew-resize hover:bg-primary/30 bg-primary/20 z-30 rounded-r"
+                          onMouseDown={(e) => handleTrimStart(e, item, 'end')}
+                          aria-label="Trim end"
+                        />
+                      </>
+                    )}
+                  </div>
+                );
+              })}
+
+              {/* Playhead - always visible */}
+              <div
+                className="absolute top-0 bottom-0 w-1 bg-accent z-30 pointer-events-none rounded-full"
+                style={{ left: `${playheadLeft}px` }}
+              >
+                <div className="absolute -top-1 left-1/2 -translate-x-1/2 w-3 h-3 bg-accent rounded-full" />
+              </div>
+            </div>
+          </div>
         </div>
       </div>
-      </div>
     </div>
   );
 }

app/src/components/VoiceProfiles/SampleList.tsx

@@ -1,9 +1,9 @@
 import { Plus, Trash2, Play } from 'lucide-react';
 import { useState } from 'react';
 import { Button } from '@/components/ui/button';
+import { apiClient } from '@/lib/api/client';
 import { useDeleteSample, useProfileSamples } from '@/lib/hooks/useProfiles';
 import { usePlayerStore } from '@/stores/playerStore';
-import { apiClient } from '@/lib/api/client';
 import { SampleUpload } from './SampleUpload';
 
 interface SampleListProps {
@@ -14,7 +14,7 @@ export function SampleList({ profileId }: SampleListProps) {
   const { data: samples, isLoading } = useProfileSamples(profileId);
   const deleteSample = useDeleteSample();
   const [uploadOpen, setUploadOpen] = useState(false);
-  const setAudio = usePlayerStore((state) => state.setAudio);
+  const setAudioWithAutoPlay = usePlayerStore((state) => state.setAudioWithAutoPlay);
   const currentAudioId = usePlayerStore((state) => state.audioId);
   const isPlaying = usePlayerStore((state) => state.isPlaying);
 
@@ -26,7 +26,7 @@ export function SampleList({ profileId }: SampleListProps) {
 
   const handlePlay = (referenceText: string, sampleId: string) => {
     const audioUrl = apiClient.getSampleUrl(sampleId);
-    setAudio(audioUrl, sampleId, referenceText.substring(0, 50));
+    setAudioWithAutoPlay(audioUrl, sampleId, null, referenceText.substring(0, 50));
   };
 
   if (isLoading) {

app/src/lib/api/client.ts

@@ -21,6 +21,8 @@ import type {
   StoryItemBatchUpdate,
   StoryItemReorder,
   StoryItemMove,
+  StoryItemTrim,
+  StoryItemSplit,
 } from './types';
 
 class ApiClient {
@@ -406,8 +408,8 @@ class ApiClient {
     });
   }
 
-  async removeStoryItem(storyId: string, generationId: string): Promise<void> {
+  async removeStoryItem(storyId: string, itemId: string): Promise<void> {
-    await this.request<void>(`/stories/${storyId}/items/${generationId}`, {
+    await this.request<void>(`/stories/${storyId}/items/${itemId}`, {
       method: 'DELETE',
     });
   }
@@ -426,13 +428,33 @@ class ApiClient {
     });
   }
 
-  async moveStoryItem(storyId: string, generationId: string, data: StoryItemMove): Promise<StoryItemDetail> {
+  async moveStoryItem(storyId: string, itemId: string, data: StoryItemMove): Promise<StoryItemDetail> {
-    return this.request<StoryItemDetail>(`/stories/${storyId}/items/${generationId}/move`, {
+    return this.request<StoryItemDetail>(`/stories/${storyId}/items/${itemId}/move`, {
       method: 'PUT',
       body: JSON.stringify(data),
     });
   }
 
+  async trimStoryItem(storyId: string, itemId: string, data: StoryItemTrim): Promise<StoryItemDetail> {
+    return this.request<StoryItemDetail>(`/stories/${storyId}/items/${itemId}/trim`, {
+      method: 'PUT',
+      body: JSON.stringify(data),
+    });
+  }
+
+  async splitStoryItem(storyId: string, itemId: string, data: StoryItemSplit): Promise<StoryItemDetail[]> {
+    return this.request<StoryItemDetail[]>(`/stories/${storyId}/items/${itemId}/split`, {
+      method: 'POST',
+      body: JSON.stringify(data),
+    });
+  }
+
+  async duplicateStoryItem(storyId: string, itemId: string): Promise<StoryItemDetail> {
+    return this.request<StoryItemDetail>(`/stories/${storyId}/items/${itemId}/duplicate`, {
+      method: 'POST',
+    });
+  }
+
   async exportStoryAudio(storyId: string): Promise<Blob> {
     const url = `${this.getBaseUrl()}/stories/${storyId}/export-audio`;
     const response = await fetch(url);

app/src/lib/api/types.ts

@@ -144,6 +144,8 @@ export interface StoryItemDetail {
   generation_id: string;
   start_time_ms: number;
   track: number;
+  trim_start_ms: number;
+  trim_end_ms: number;
   created_at: string;
   profile_id: string;
   profile_name: string;
@@ -188,3 +190,12 @@ export interface StoryItemMove {
   start_time_ms: number;
   track: number;
 }
+
+export interface StoryItemTrim {
+  trim_start_ms: number;
+  trim_end_ms: number;
+}
+
+export interface StoryItemSplit {
+  split_time_ms: number;
+}

app/src/lib/hooks/useGenerationForm.ts

@@ -28,7 +28,7 @@ interface UseGenerationFormOptions {
 export function useGenerationForm(options: UseGenerationFormOptions = {}) {
   const { toast } = useToast();
   const generation = useGeneration();
-  const setAudio = usePlayerStore((state) => state.setAudio);
+  const setAudioWithAutoPlay = usePlayerStore((state) => state.setAudioWithAutoPlay);
   const setIsGenerating = useGenerationStore((state) => state.setIsGenerating);
   const [downloadingModelName, setDownloadingModelName] = useState<string | null>(null);
   const [downloadingDisplayName, setDownloadingDisplayName] = useState<string | null>(null);
@@ -97,7 +97,7 @@ export function useGenerationForm(options: UseGenerationFormOptions = {}) {
       });
 
       const audioUrl = apiClient.getAudioUrl(result.id);
-      setAudio(audioUrl, result.id, selectedProfileId, data.text.substring(0, 50));
+      setAudioWithAutoPlay(audioUrl, result.id, selectedProfileId, data.text.substring(0, 50));
 
       form.reset();
       options.onSuccess?.(result.id);

app/src/lib/hooks/useStories.ts

@@ -1,6 +1,6 @@
 import { useMutation, useQuery, useQueryClient } from '@tanstack/react-query';
 import { apiClient } from '@/lib/api/client';
-import type { StoryCreate, StoryItemCreate, StoryItemBatchUpdate, StoryItemReorder, StoryItemMove } from '@/lib/api/types';
+import type { StoryCreate, StoryItemCreate, StoryItemBatchUpdate, StoryItemReorder, StoryItemMove, StoryItemTrim, StoryItemSplit } from '@/lib/api/types';
 import { isTauri } from '@/lib/tauri';
 
 export function useStories() {
@@ -70,8 +70,8 @@ export function useRemoveStoryItem() {
   const queryClient = useQueryClient();
 
   return useMutation({
-    mutationFn: ({ storyId, generationId }: { storyId: string; generationId: string }) =>
+    mutationFn: ({ storyId, itemId }: { storyId: string; itemId: string }) =>
-      apiClient.removeStoryItem(storyId, generationId),
+      apiClient.removeStoryItem(storyId, itemId),
     onSuccess: (_, variables) => {
       queryClient.invalidateQueries({ queryKey: ['stories'] });
       queryClient.invalidateQueries({ queryKey: ['stories', variables.storyId] });
@@ -109,8 +109,47 @@ export function useMoveStoryItem() {
   const queryClient = useQueryClient();
 
   return useMutation({
-    mutationFn: ({ storyId, generationId, data }: { storyId: string; generationId: string; data: StoryItemMove }) =>
+    mutationFn: ({ storyId, itemId, data }: { storyId: string; itemId: string; data: StoryItemMove }) =>
-      apiClient.moveStoryItem(storyId, generationId, data),
+      apiClient.moveStoryItem(storyId, itemId, data),
+    onSuccess: (_, variables) => {
+      queryClient.invalidateQueries({ queryKey: ['stories'] });
+      queryClient.invalidateQueries({ queryKey: ['stories', variables.storyId] });
+    },
+  });
+}
+
+export function useTrimStoryItem() {
+  const queryClient = useQueryClient();
+
+  return useMutation({
+    mutationFn: ({ storyId, itemId, data }: { storyId: string; itemId: string; data: StoryItemTrim }) =>
+      apiClient.trimStoryItem(storyId, itemId, data),
+    onSuccess: (_, variables) => {
+      queryClient.invalidateQueries({ queryKey: ['stories'] });
+      queryClient.invalidateQueries({ queryKey: ['stories', variables.storyId] });
+    },
+  });
+}
+
+export function useSplitStoryItem() {
+  const queryClient = useQueryClient();
+
+  return useMutation({
+    mutationFn: ({ storyId, itemId, data }: { storyId: string; itemId: string; data: StoryItemSplit }) =>
+      apiClient.splitStoryItem(storyId, itemId, data),
+    onSuccess: (_, variables) => {
+      queryClient.invalidateQueries({ queryKey: ['stories'] });
+      queryClient.invalidateQueries({ queryKey: ['stories', variables.storyId] });
+    },
+  });
+}
+
+export function useDuplicateStoryItem() {
+  const queryClient = useQueryClient();
+
+  return useMutation({
+    mutationFn: ({ storyId, itemId }: { storyId: string; itemId: string }) =>
+      apiClient.duplicateStoryItem(storyId, itemId),
     onSuccess: (_, variables) => {
       queryClient.invalidateQueries({ queryKey: ['stories'] });
       queryClient.invalidateQueries({ queryKey: ['stories', variables.storyId] });

app/src/lib/hooks/useStoryPlayback.ts

@@ -5,6 +5,7 @@ import { useStoryStore } from '@/stores/storyStore';
 
 interface ActiveSource {
   source: AudioBufferSourceNode;
+  itemId: string;
   generationId: string;
   startTimeMs: number;
   endTimeMs: number;
@@ -26,9 +27,9 @@ export function useStoryPlayback(items: StoryItemDetail[] | undefined) {
   const audioContextRef = useRef<AudioContext | null>(null);
   // Master gain for volume control
   const masterGainRef = useRef<GainNode | null>(null);
-  // Preloaded AudioBuffers by generation_id
+  // Preloaded AudioBuffers by generation_id (audio file is shared between split clips)
   const audioBuffersRef = useRef<Map<string, AudioBuffer>>(new Map());
-  // Currently playing AudioBufferSourceNodes by generation_id
+  // Currently playing AudioBufferSourceNodes by item.id (unique per clip)
   const activeSourcesRef = useRef<Map<string, ActiveSource>>(new Map());
   // Animation frame for syncing visual playhead
   const animationFrameRef = useRef<number | null>(null);
@@ -56,16 +57,16 @@ export function useStoryPlayback(items: StoryItemDetail[] | undefined) {
     return audioContextRef.current;
   }, []);
 
-  // Stop a source
+  // Stop a source by item id
-  const stopSource = useCallback((generationId: string) => {
+  const stopSource = useCallback((itemId: string) => {
-    const activeSource = activeSourcesRef.current.get(generationId);
+    const activeSource = activeSourcesRef.current.get(itemId);
     if (activeSource) {
       try {
         activeSource.source.stop();
       } catch {
         // Source may have already stopped
       }
-      activeSourcesRef.current.delete(generationId);
+      activeSourcesRef.current.delete(itemId);
     }
   }, []);
 
@@ -123,8 +124,8 @@ export function useStoryPlayback(items: StoryItemDetail[] | undefined) {
   useEffect(() => {
     return () => {
       // Stop all sources
-      for (const [generationId] of activeSourcesRef.current) {
+      for (const [itemId] of activeSourcesRef.current) {
-        stopSource(generationId);
+        stopSource(itemId);
       }
       activeSourcesRef.current.clear();
 
@@ -151,7 +152,11 @@ export function useStoryPlayback(items: StoryItemDetail[] | undefined) {
     (storyTimeMs: number, itemList: StoryItemDetail[]): StoryItemDetail[] => {
       return itemList.filter((item) => {
         const itemStart = item.start_time_ms;
-        const itemEnd = item.start_time_ms + item.duration * 1000;
+        // Use effective duration (accounting for trims)
+        const trimStartMs = item.trim_start_ms || 0;
+        const trimEndMs = item.trim_end_ms || 0;
+        const effectiveDurationMs = item.duration * 1000 - trimStartMs - trimEndMs;
+        const itemEnd = item.start_time_ms + effectiveDurationMs;
         return storyTimeMs >= itemStart && storyTimeMs < itemEnd;
       });
     },
@@ -185,8 +190,8 @@ export function useStoryPlayback(items: StoryItemDetail[] | undefined) {
   // Stop all sources
   const stopAllSources = useCallback(() => {
     console.log('[StoryPlayback] Stopping all sources');
-    for (const [generationId] of activeSourcesRef.current) {
+    for (const [itemId] of activeSourcesRef.current) {
-      stopSource(generationId);
+      stopSource(itemId);
     }
     activeSourcesRef.current.clear();
   }, [stopSource]);
@@ -199,18 +204,18 @@ export function useStoryPlayback(items: StoryItemDetail[] | undefined) {
 
       // Find all items that should be playing
       const shouldBePlaying = findActiveItems(storyTimeMs, itemList);
-      const shouldBePlayingIds = new Set(shouldBePlaying.map((item) => item.generation_id));
+      const shouldBePlayingIds = new Set(shouldBePlaying.map((item) => item.id));
 
       // Stop sources that shouldn't be playing anymore
-      for (const [generationId] of activeSourcesRef.current) {
+      for (const [itemId] of activeSourcesRef.current) {
-        if (!shouldBePlayingIds.has(generationId)) {
+        if (!shouldBePlayingIds.has(itemId)) {
-          stopSource(generationId);
+          stopSource(itemId);
         }
       }
 
       // Schedule new sources for items that should be playing
       for (const item of shouldBePlaying) {
-        if (!activeSourcesRef.current.has(item.generation_id)) {
+        if (!activeSourcesRef.current.has(item.id)) {
           const buffer = audioBuffersRef.current.get(item.generation_id);
           if (!buffer) {
             console.warn('[StoryPlayback] Buffer not loaded for:', item.generation_id);
@@ -219,16 +224,24 @@ export function useStoryPlayback(items: StoryItemDetail[] | undefined) {
 
           // Calculate when this item should start in AudioContext time
           const itemStartContextTime = storyTimeToContextTime(item.start_time_ms);
-          const itemEndStoryTime = item.start_time_ms + item.duration * 1000;
+          
+          // Calculate effective duration and trim offsets
+          const trimStartSec = (item.trim_start_ms || 0) / 1000;
+          const trimEndSec = (item.trim_end_ms || 0) / 1000;
+          const effectiveDuration = item.duration - trimStartSec - trimEndSec;
+          const itemEndStoryTime = item.start_time_ms + effectiveDuration * 1000;
 
           // Calculate offset into the buffer (if seeking mid-way)
-          const offsetIntoBuffer = Math.max(0, (storyTimeMs - item.start_time_ms) / 1000);
+          // Offset is relative to the trimmed start of the clip
-          const duration = item.duration - offsetIntoBuffer;
+          const offsetIntoEffectiveClip = Math.max(0, (storyTimeMs - item.start_time_ms) / 1000);
+          const offsetIntoBuffer = trimStartSec + offsetIntoEffectiveClip;
+          const duration = effectiveDuration - offsetIntoEffectiveClip;
 
           // If the item should have already started, schedule it to start immediately
           const startAtContextTime = Math.max(currentContextTime, itemStartContextTime);
 
           console.log('[StoryPlayback] Scheduling source:', {
+            itemId: item.id,
             generationId: item.generation_id,
             storyTimeMs,
             itemStart: item.start_time_ms,
@@ -243,20 +256,21 @@ export function useStoryPlayback(items: StoryItemDetail[] | undefined) {
 
           const activeSource: ActiveSource = {
             source,
+            itemId: item.id,
             generationId: item.generation_id,
             startTimeMs: item.start_time_ms,
             endTimeMs: itemEndStoryTime,
           };
 
-          activeSourcesRef.current.set(item.generation_id, activeSource);
+          activeSourcesRef.current.set(item.id, activeSource);
 
           // Schedule playback
           source.start(startAtContextTime, offsetIntoBuffer, duration);
 
           // Clean up when source ends
           source.onended = () => {
-            console.log('[StoryPlayback] Source ended:', item.generation_id);
+            console.log('[StoryPlayback] Source ended:', item.id);
-            activeSourcesRef.current.delete(item.generation_id);
+            activeSourcesRef.current.delete(item.id);
           };
         }
       }

app/src/stores/storyStore.ts

@@ -5,6 +5,8 @@ interface StoryPlaybackState {
   // Selection
   selectedStoryId: string | null;
   setSelectedStoryId: (id: string | null) => void;
+  selectedClipId: string | null;
+  setSelectedClipId: (id: string | null) => void;
 
   // Track editor UI state
   trackEditorHeight: number;
@@ -26,6 +28,7 @@ interface StoryPlaybackState {
   stop: () => void;
   seek: (timeMs: number) => void;
   setPlaybackTiming: (contextTime: number, storyTime: number) => void; // Set timing anchors for Web Audio API
+  setActiveStory: (storyId: string, items: StoryItemDetail[], totalDurationMs: number) => void; // Activate story for seeking without playing
 }
 
 const DEFAULT_TRACK_EDITOR_HEIGHT = 250;
@@ -34,6 +37,8 @@ export const useStoryStore = create<StoryPlaybackState>((set, get) => ({
   // Selection
   selectedStoryId: null,
   setSelectedStoryId: (id) => set({ selectedStoryId: id }),
+  selectedClipId: null,
+  setSelectedClipId: (id) => set({ selectedClipId: id }),
 
   // Track editor UI state
   trackEditorHeight: DEFAULT_TRACK_EDITOR_HEIGHT,
@@ -53,14 +58,11 @@ export const useStoryStore = create<StoryPlaybackState>((set, get) => ({
     // Calculate total duration from items
     const maxEndTimeMs = Math.max(
       ...items.map((item) => item.start_time_ms + item.duration * 1000),
-      0
+      0,
     );
 
     // Find the minimum start time (first item)
-    const minStartTimeMs = Math.min(
+    const minStartTimeMs = Math.min(...items.map((item) => item.start_time_ms), 0);
-      ...items.map((item) => item.start_time_ms),
-      0
-    );
 
     // If resuming the same story, keep position; otherwise start at first item
     const currentState = get();
@@ -70,7 +72,11 @@ export const useStoryStore = create<StoryPlaybackState>((set, get) => ({
     console.log('[StoryStore] Play called:', {
       storyId,
       itemCount: items.length,
-      items: items.map(i => ({ id: i.generation_id, start: i.start_time_ms, duration: i.duration })),
+      items: items.map((i) => ({
+        id: i.generation_id,
+        start: i.start_time_ms,
+        duration: i.duration,
+      })),
       maxEndTimeMs,
       minStartTimeMs,
       startTimeMs,
@@ -122,4 +128,18 @@ export const useStoryStore = create<StoryPlaybackState>((set, get) => ({
       playbackStartStoryTime: storyTime,
     });
   },
+
+  setActiveStory: (storyId, items, totalDurationMs) => {
+    const currentState = get();
+    // Only update if switching to a different story
+    if (currentState.playbackStoryId !== storyId) {
+      set({
+        playbackStoryId: storyId,
+        playbackItems: items,
+        totalDurationMs,
+        currentTimeMs: 0,
+        isPlaying: false,
+      });
+    }
+  },
 }));

backend/__init__.py

@@ -1 +1,3 @@
 # Backend package
+
+__version__ = "0.1.8"

backend/database.py

@@ -71,6 +71,8 @@ class StoryItem(Base):
     generation_id = Column(String, ForeignKey("generations.id"), nullable=False)
     start_time_ms = Column(Integer, nullable=False, default=0)  # Milliseconds from story start
     track = Column(Integer, nullable=False, default=0)  # Track number (0 = main track)
+    trim_start_ms = Column(Integer, nullable=False, default=0)  # Milliseconds trimmed from start
+    trim_end_ms = Column(Integer, nullable=False, default=0)  # Milliseconds trimmed from end
     created_at = Column(DateTime, default=datetime.utcnow)
 
 
@@ -257,6 +259,24 @@ def _run_migrations(engine):
             conn.commit()
             print("Added track column to story_items")
     
+    # Migration: Add trim columns if they don't exist
+    # Re-check columns after potential track migration
+    columns = {col['name'] for col in inspector.get_columns('story_items')}
+    if 'trim_start_ms' not in columns:
+        print("Migrating story_items: adding trim_start_ms column")
+        with engine.connect() as conn:
+            conn.execute(text("ALTER TABLE story_items ADD COLUMN trim_start_ms INTEGER NOT NULL DEFAULT 0"))
+            conn.commit()
+            print("Added trim_start_ms column to story_items")
+    
+    columns = {col['name'] for col in inspector.get_columns('story_items')}
+    if 'trim_end_ms' not in columns:
+        print("Migrating story_items: adding trim_end_ms column")
+        with engine.connect() as conn:
+            conn.execute(text("ALTER TABLE story_items ADD COLUMN trim_end_ms INTEGER NOT NULL DEFAULT 0"))
+            conn.commit()
+            print("Added trim_end_ms column to story_items")
+
 
 def get_db():
     """Get database session (generator for dependency injection)."""

backend/main.py

@@ -11,6 +11,7 @@ from fastapi.staticfiles import StaticFiles
 from sqlalchemy.orm import Session
 from typing import List, Optional
 from datetime import datetime
+import asyncio
 import uvicorn
 import argparse
 import torch
@@ -18,8 +19,11 @@ import tempfile
 import io
 from pathlib import Path
 import uuid
+import asyncio
+import signal
+import os
 
-from . import database, models, profiles, history, tts, transcribe, config, export_import, channels, stories
+from . import database, models, profiles, history, tts, transcribe, config, export_import, channels, stories, __version__
 from .database import get_db, Generation as DBGeneration, VoiceProfile as DBVoiceProfile
 from .utils.progress import get_progress_manager
 from .utils.tasks import get_task_manager
@@ -27,7 +31,7 @@ from .utils.tasks import get_task_manager
 app = FastAPI(
     title="voicebox API",
     description="Production-quality Qwen3-TTS voice cloning API",
-    version="0.1.0",
+    version=__version__,
 )
 
 # CORS middleware
@@ -47,13 +51,24 @@ app.add_middleware(
 @app.get("/")
 async def root():
     """Root endpoint."""
-    return {"message": "voicebox API", "version": "0.1.6"}
+    return {"message": "voicebox API", "version": __version__}
+
+
+@app.post("/shutdown")
+async def shutdown():
+    """Gracefully shutdown the server."""
+    async def shutdown_async():
+        await asyncio.sleep(0.1)  # Give response time to send
+        os.kill(os.getpid(), signal.SIGTERM)
+
+    asyncio.create_task(shutdown_async())
+    return {"message": "Shutting down..."}
 
 
 @app.get("/health", response_model=models.HealthResponse)
 async def health():
     """Health check endpoint."""
-    from huggingface_hub import hf_hub_download
+    from huggingface_hub import hf_hub_download, constants as hf_constants
     from pathlib import Path
     import os
 
@@ -64,6 +79,12 @@ async def health():
     has_mps = hasattr(torch.backends, 'mps') and torch.backends.mps.is_available()
     gpu_available = has_cuda or has_mps
 
+    gpu_type = None
+    if has_cuda:
+        gpu_type = f"CUDA ({torch.cuda.get_device_name(0)})"
+    elif has_mps:
+        gpu_type = "MPS (Apple Silicon)"
+
     vram_used = None
     if has_cuda:
         vram_used = torch.cuda.memory_allocated() / 1024 / 1024  # MB
@@ -101,8 +122,8 @@ async def health():
                     model_downloaded = True
                     break
         except (ImportError, Exception):
-            # Method 2: Check cache directory
+            # Method 2: Check cache directory (using HuggingFace's OS-specific cache location)
-            cache_dir = os.path.expanduser("~/.cache/huggingface/hub")
+            cache_dir = hf_constants.HF_HUB_CACHE
             repo_cache = Path(cache_dir) / "models--" + default_model_id.replace("/", "--")
             if repo_cache.exists():
                 has_model_files = (
@@ -121,6 +142,7 @@ async def health():
         model_downloaded=model_downloaded,
         model_size=model_size,
         gpu_available=gpu_available,
+        gpu_type=gpu_type,
         vram_used_mb=vram_used,
     )
 
@@ -451,6 +473,36 @@ async def generate_speech(
         tts_model = tts.get_tts_model()
         # Load the requested model size if different from current (async to not block)
         model_size = data.model_size or "1.7B"
+
+        # Check if model needs to be downloaded first
+        model_path = tts_model._get_model_path(model_size)
+        if model_path.startswith("Qwen/"):
+            # Model not cached - check if it exists remotely or needs download
+            from huggingface_hub import constants as hf_constants
+            repo_cache = Path(hf_constants.HF_HUB_CACHE) / ("models--" + model_path.replace("/", "--"))
+            if not repo_cache.exists():
+                # Start download in background
+                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)
+                asyncio.create_task(download_model_background())
+
+                # Return 202 Accepted with download info
+                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
+                    }
+                )
+
         await tts_model.load_model_async(model_size)
         audio, sample_rate = await tts_model.generate(
             data.text,
@@ -684,6 +736,37 @@ async def transcribe_audio(
         
         # Transcribe
         whisper_model = transcribe.get_whisper_model()
+
+        # Check if Whisper model is downloaded (uses default size "base")
+        model_size = whisper_model.model_size
+        model_name = f"openai/whisper-{model_size}"
+
+        # Check if model is cached
+        from huggingface_hub import constants as hf_constants
+        repo_cache = Path(hf_constants.HF_HUB_CACHE) / ("models--" + model_name.replace("/", "--"))
+        if not repo_cache.exists():
+            # Start download in background
+            progress_model_name = f"whisper-{model_size}"
+
+            async def download_whisper_background():
+                try:
+                    await whisper_model.load_model_async(model_size)
+                except Exception as e:
+                    get_task_manager().error_download(progress_model_name, str(e))
+
+            get_task_manager().start_download(progress_model_name)
+            asyncio.create_task(download_whisper_background())
+
+            # Return 202 Accepted
+            raise HTTPException(
+                status_code=202,
+                detail={
+                    "message": f"Whisper model {model_size} is being downloaded. Please wait and try again.",
+                    "model_name": progress_model_name,
+                    "downloading": True
+                }
+            )
+
         text = await whisper_model.transcribe(tmp_path, language)
         
         return models.TranscriptionResponse(
@@ -770,14 +853,14 @@ async def add_story_item(
     return item
 
 
-@app.delete("/stories/{story_id}/items/{generation_id}")
+@app.delete("/stories/{story_id}/items/{item_id}")
 async def remove_story_item(
     story_id: str,
-    generation_id: str,
+    item_id: str,
     db: Session = Depends(get_db),
 ):
-    """Remove a generation from a story."""
+    """Remove a story item from a story."""
-    success = await stories.remove_item_from_story(story_id, generation_id, db)
+    success = await stories.remove_item_from_story(story_id, item_id, db)
     if not success:
         raise HTTPException(status_code=404, detail="Story item not found")
     return {"message": "Item removed successfully"}
@@ -809,15 +892,56 @@ async def reorder_story_items(
     return items
 
 
-@app.put("/stories/{story_id}/items/{generation_id}/move", response_model=models.StoryItemDetail)
+@app.put("/stories/{story_id}/items/{item_id}/move", response_model=models.StoryItemDetail)
 async def move_story_item(
     story_id: str,
-    generation_id: str,
+    item_id: str,
     data: models.StoryItemMove,
     db: Session = Depends(get_db),
 ):
     """Move a story item (update position and/or track)."""
-    item = await stories.move_story_item(story_id, generation_id, data, db)
+    item = await stories.move_story_item(story_id, item_id, data, db)
+    if item is None:
+        raise HTTPException(status_code=404, detail="Story item not found")
+    return item
+
+
+@app.put("/stories/{story_id}/items/{item_id}/trim", response_model=models.StoryItemDetail)
+async def trim_story_item(
+    story_id: str,
+    item_id: str,
+    data: models.StoryItemTrim,
+    db: Session = Depends(get_db),
+):
+    """Trim a story item (update trim_start_ms and trim_end_ms)."""
+    item = await stories.trim_story_item(story_id, item_id, data, db)
+    if item is None:
+        raise HTTPException(status_code=404, detail="Story item not found or invalid trim values")
+    return item
+
+
+@app.post("/stories/{story_id}/items/{item_id}/split", response_model=List[models.StoryItemDetail])
+async def split_story_item(
+    story_id: str,
+    item_id: str,
+    data: models.StoryItemSplit,
+    db: Session = Depends(get_db),
+):
+    """Split a story item at a given time, creating two clips."""
+    items = await stories.split_story_item(story_id, item_id, data, db)
+    if items is None:
+        raise HTTPException(status_code=404, detail="Story item not found or invalid split point")
+    return items
+
+
+@app.post("/stories/{story_id}/items/{item_id}/duplicate", response_model=models.StoryItemDetail)
+async def duplicate_story_item(
+    story_id: str,
+    item_id: str,
+    db: Session = Depends(get_db),
+):
+    """Duplicate a story item, creating a copy with all properties."""
+    item = await stories.duplicate_story_item(story_id, item_id, db)
     if item is None:
         raise HTTPException(status_code=404, detail="Story item not found")
     return item
@@ -953,7 +1077,7 @@ async def get_model_progress(model_name: str):
 @app.get("/models/status", response_model=models.ModelStatusListResponse)
 async def get_model_status():
     """Get status of all available models."""
-    from huggingface_hub import hf_hub_download
+    from huggingface_hub import hf_hub_download, constants as hf_constants
     from pathlib import Path
     import os
     
@@ -1056,10 +1180,10 @@ async def get_model_status():
                             pass
                         break
             
-            # Method 2: Fallback to checking cache directory directly
+            # Method 2: Fallback to checking cache directory directly (using HuggingFace's OS-specific cache location)
             if not downloaded:
                 try:
-                    cache_dir = os.path.expanduser("~/.cache/huggingface/hub")
+                    cache_dir = hf_constants.HF_HUB_CACHE
                     repo_cache = Path(cache_dir) / "models--" + config["hf_repo_id"].replace("/", "--")
                     
                     if repo_cache.exists():
@@ -1168,22 +1292,22 @@ async def trigger_model_download(request: models.ModelDownloadRequest):
     
     config = model_configs[request.model_name]
     
-    try:
+    async def download_in_background():
-        # Start tracking download
+        """Download model in background without blocking the HTTP request."""
-        task_manager.start_download(request.model_name)
+        try:
+            await asyncio.to_thread(config["load_func"])
+            task_manager.complete_download(request.model_name)
+        except Exception as e:
+            task_manager.error_download(request.model_name, str(e))
 
-        # Trigger download by loading the model (which will download if not cached)
+    # Start tracking download
-        # Run in background to avoid blocking
+    task_manager.start_download(request.model_name)
-        await asyncio.to_thread(config["load_func"])
 
-        # Mark download as complete
+    # Start download in background task (don't await)
-        task_manager.complete_download(request.model_name)
+    asyncio.create_task(download_in_background())
 
-        return {"message": f"Model {request.model_name} download started"}
+    # Return immediately - frontend should poll progress endpoint
-    except Exception as e:
+    return {"message": f"Model {request.model_name} download started"}
-        # Mark download as failed
-        task_manager.error_download(request.model_name, str(e))
-        raise HTTPException(status_code=500, detail=str(e))
 
 
 @app.delete("/models/{model_name}")
@@ -1191,6 +1315,7 @@ async def delete_model(model_name: str):
     """Delete a downloaded model from the HuggingFace cache."""
     import shutil
     import os
+    from huggingface_hub import constants as hf_constants
     
     # Map model names to HuggingFace repo IDs
     model_configs = {
@@ -1243,8 +1368,8 @@ async def delete_model(model_name: str):
             if whisper_model.is_loaded() and whisper_model.model_size == config["model_size"]:
                 transcribe.unload_whisper_model()
         
-        # Find and delete the cache directory
+        # Find and delete the cache directory (using HuggingFace's OS-specific cache location)
-        cache_dir = os.path.expanduser("~/.cache/huggingface/hub")
+        cache_dir = hf_constants.HF_HUB_CACHE
         repo_cache_dir = Path(cache_dir) / ("models--" + hf_repo_id.replace("/", "--"))
         
         # Check if the cache directory exists
@@ -1355,6 +1480,16 @@ async def startup_event():
     print(f"Database initialized at {database._db_path}")
     print(f"GPU available: {_get_gpu_status()}")
 
+    # Ensure HuggingFace cache directory exists
+    try:
+        from huggingface_hub import constants as hf_constants
+        cache_dir = Path(hf_constants.HF_HUB_CACHE)
+        cache_dir.mkdir(parents=True, exist_ok=True)
+        print(f"HuggingFace cache directory: {cache_dir}")
+    except Exception as e:
+        print(f"Warning: Could not create HuggingFace cache directory: {e}")
+        print("Model downloads may fail. Please ensure the directory exists and has write permissions.")
+
 
 @app.on_event("shutdown")
 async def shutdown_event():

backend/models.py

@@ -118,6 +118,7 @@ class HealthResponse(BaseModel):
     model_downloaded: Optional[bool] = None  # Whether model is cached/downloaded
     model_size: Optional[str] = None  # Current model size if loaded
     gpu_available: bool
+    gpu_type: Optional[str] = None  # GPU type (CUDA, MPS, or None)
     vram_used_mb: Optional[float] = None
 
 
@@ -221,6 +222,8 @@ class StoryItemDetail(BaseModel):
     generation_id: str
     start_time_ms: int
     track: int = 0
+    trim_start_ms: int = 0
+    trim_end_ms: int = 0
     created_at: datetime
     # Generation details
     profile_id: str
@@ -277,3 +280,14 @@ class StoryItemMove(BaseModel):
     """Request model for moving a story item (position and/or track)."""
     start_time_ms: int = Field(..., ge=0)
     track: int = 0
+
+
+class StoryItemTrim(BaseModel):
+    """Request model for trimming a story item."""
+    trim_start_ms: int = Field(..., ge=0)
+    trim_end_ms: int = Field(..., ge=0)
+
+
+class StoryItemSplit(BaseModel):
+    """Request model for splitting a story item."""
+    split_time_ms: int = Field(..., ge=0)  # Time within the clip to split at (relative to clip start)

backend/stories.py

@@ -18,6 +18,8 @@ from .models import (
     StoryItemCreate,
     StoryItemBatchUpdate,
     StoryItemMove,
+    StoryItemTrim,
+    StoryItemSplit,
 )
 from .database import Story as DBStory, StoryItem as DBStoryItem, Generation as DBGeneration, VoiceProfile as DBVoiceProfile
 from .utils.audio import load_audio, save_audio
@@ -129,6 +131,8 @@ async def get_story(
             generation_id=item.generation_id,
             start_time_ms=item.start_time_ms,
             track=item.track,
+            trim_start_ms=getattr(item, 'trim_start_ms', 0),
+            trim_end_ms=getattr(item, 'trim_end_ms', 0),
             created_at=item.created_at,
             profile_id=generation.profile_id,
             profile_name=profile_name,
@@ -252,6 +256,8 @@ async def add_item_to_story(
             generation_id=existing.generation_id,
             start_time_ms=existing.start_time_ms,
             track=existing.track,
+            trim_start_ms=getattr(existing, 'trim_start_ms', 0),
+            trim_end_ms=getattr(existing, 'trim_end_ms', 0),
             created_at=existing.created_at,
             profile_id=generation.profile_id,
             profile_name=profile.name if profile else "Unknown",
@@ -321,6 +327,8 @@ async def add_item_to_story(
         generation_id=item.generation_id,
         start_time_ms=item.start_time_ms,
         track=item.track,
+        trim_start_ms=getattr(item, 'trim_start_ms', 0),
+        trim_end_ms=getattr(item, 'trim_end_ms', 0),
         created_at=item.created_at,
         profile_id=generation.profile_id,
         profile_name=profile.name if profile else "Unknown",
@@ -336,7 +344,7 @@ async def add_item_to_story(
 
 async def move_story_item(
     story_id: str,
-    generation_id: str,
+    item_id: str,
     data: StoryItemMove,
     db: Session,
 ) -> Optional[StoryItemDetail]:
@@ -345,7 +353,7 @@ async def move_story_item(
 
     Args:
         story_id: Story ID
-        generation_id: Generation ID of the item to move
+        item_id: Story item ID
         data: New position and track data
         db: Database session
 
@@ -354,14 +362,14 @@ async def move_story_item(
     """
     # Get the item
     item = db.query(DBStoryItem).filter_by(
+        id=item_id,
         story_id=story_id,
-        generation_id=generation_id
     ).first()
     if not item:
         return None
 
     # Get the generation
-    generation = db.query(DBGeneration).filter_by(id=generation_id).first()
+    generation = db.query(DBGeneration).filter_by(id=item.generation_id).first()
     if not generation:
         return None
 
@@ -386,6 +394,8 @@ async def move_story_item(
         generation_id=item.generation_id,
         start_time_ms=item.start_time_ms,
         track=item.track,
+        trim_start_ms=getattr(item, 'trim_start_ms', 0),
+        trim_end_ms=getattr(item, 'trim_end_ms', 0),
         created_at=item.created_at,
         profile_id=generation.profile_id,
         profile_name=profile.name if profile else "Unknown",
@@ -401,23 +411,23 @@ async def move_story_item(
 
 async def remove_item_from_story(
     story_id: str,
-    generation_id: str,
+    item_id: str,
     db: Session,
 ) -> bool:
     """
-    Remove a generation from a story.
+    Remove a story item from a story.
 
     Args:
         story_id: Story ID
-        generation_id: Generation ID to remove
+        item_id: Story item ID to remove
         db: Database session
 
     Returns:
         True if removed, False if not found
     """
     item = db.query(DBStoryItem).filter_by(
+        id=item_id,
         story_id=story_id,
-        generation_id=generation_id
     ).first()
     if not item:
         return False
@@ -434,6 +444,277 @@ async def remove_item_from_story(
     return True
 
 
+async def trim_story_item(
+    story_id: str,
+    item_id: str,
+    data: StoryItemTrim,
+    db: Session,
+) -> Optional[StoryItemDetail]:
+    """
+    Trim a story item (update trim_start_ms and trim_end_ms).
+
+    Args:
+        story_id: Story ID
+        item_id: Story item ID
+        data: Trim data (trim_start_ms, trim_end_ms)
+        db: Database session
+
+    Returns:
+        Updated item detail or None if not found
+    """
+    # Get the item
+    item = db.query(DBStoryItem).filter_by(
+        id=item_id,
+        story_id=story_id,
+    ).first()
+    if not item:
+        return None
+
+    # Get the generation
+    generation = db.query(DBGeneration).filter_by(id=item.generation_id).first()
+    if not generation:
+        return None
+
+    # Validate trim values don't exceed duration
+    max_duration_ms = int(generation.duration * 1000)
+    if data.trim_start_ms + data.trim_end_ms >= max_duration_ms:
+        return None  # Invalid trim - would result in zero or negative duration
+
+    # Update trim values
+    item.trim_start_ms = data.trim_start_ms
+    item.trim_end_ms = data.trim_end_ms
+
+    # Update story updated_at
+    story = db.query(DBStory).filter_by(id=story_id).first()
+    if story:
+        story.updated_at = datetime.utcnow()
+
+    db.commit()
+    db.refresh(item)
+
+    # Get profile name
+    profile = db.query(DBVoiceProfile).filter_by(id=generation.profile_id).first()
+
+    return StoryItemDetail(
+        id=item.id,
+        story_id=item.story_id,
+        generation_id=item.generation_id,
+        start_time_ms=item.start_time_ms,
+        track=item.track,
+        trim_start_ms=item.trim_start_ms,
+        trim_end_ms=item.trim_end_ms,
+        created_at=item.created_at,
+        profile_id=generation.profile_id,
+        profile_name=profile.name if profile else "Unknown",
+        text=generation.text,
+        language=generation.language,
+        audio_path=generation.audio_path,
+        duration=generation.duration,
+        seed=generation.seed,
+        instruct=generation.instruct,
+        generation_created_at=generation.created_at,
+    )
+
+
+async def split_story_item(
+    story_id: str,
+    item_id: str,
+    data: StoryItemSplit,
+    db: Session,
+) -> Optional[List[StoryItemDetail]]:
+    """
+    Split a story item at a given time, creating two clips.
+
+    Args:
+        story_id: Story ID
+        item_id: Story item ID to split
+        data: Split data (split_time_ms - time within clip to split at)
+        db: Database session
+
+    Returns:
+        List of two updated item details (original and new) or None if not found/invalid
+    """
+    # Get the item
+    item = db.query(DBStoryItem).filter_by(
+        id=item_id,
+        story_id=story_id,
+    ).first()
+    if not item:
+        return None
+
+    # Get the generation
+    generation = db.query(DBGeneration).filter_by(id=item.generation_id).first()
+    if not generation:
+        return None
+
+    # Calculate effective duration and validate split point
+    current_trim_start = getattr(item, 'trim_start_ms', 0)
+    current_trim_end = getattr(item, 'trim_end_ms', 0)
+    original_duration_ms = int(generation.duration * 1000)
+    effective_duration_ms = original_duration_ms - current_trim_start - current_trim_end
+
+    # Validate split_time_ms is within the effective duration
+    if data.split_time_ms <= 0 or data.split_time_ms >= effective_duration_ms:
+        return None  # Invalid split point
+
+    # Calculate the absolute time in the original audio where we're splitting
+    absolute_split_ms = current_trim_start + data.split_time_ms
+
+    # Update original clip: trim from the end
+    item.trim_end_ms = original_duration_ms - absolute_split_ms
+
+    # Create new clip: starts after the split, trimmed from the start
+    new_item = DBStoryItem(
+        id=str(uuid.uuid4()),
+        story_id=story_id,
+        generation_id=item.generation_id,  # Same generation, different trim
+        start_time_ms=item.start_time_ms + data.split_time_ms,
+        track=item.track,
+        trim_start_ms=absolute_split_ms,
+        trim_end_ms=current_trim_end,
+        created_at=datetime.utcnow(),
+    )
+
+    db.add(new_item)
+
+    # Update story updated_at
+    story = db.query(DBStory).filter_by(id=story_id).first()
+    if story:
+        story.updated_at = datetime.utcnow()
+
+    db.commit()
+    db.refresh(item)
+    db.refresh(new_item)
+
+    # Get profile name
+    profile = db.query(DBVoiceProfile).filter_by(id=generation.profile_id).first()
+    profile_name = profile.name if profile else "Unknown"
+
+    # Build response items
+    original_item_detail = StoryItemDetail(
+        id=item.id,
+        story_id=item.story_id,
+        generation_id=item.generation_id,
+        start_time_ms=item.start_time_ms,
+        track=item.track,
+        trim_start_ms=item.trim_start_ms,
+        trim_end_ms=item.trim_end_ms,
+        created_at=item.created_at,
+        profile_id=generation.profile_id,
+        profile_name=profile_name,
+        text=generation.text,
+        language=generation.language,
+        audio_path=generation.audio_path,
+        duration=generation.duration,
+        seed=generation.seed,
+        instruct=generation.instruct,
+        generation_created_at=generation.created_at,
+    )
+
+    new_item_detail = StoryItemDetail(
+        id=new_item.id,
+        story_id=new_item.story_id,
+        generation_id=new_item.generation_id,
+        start_time_ms=new_item.start_time_ms,
+        track=new_item.track,
+        trim_start_ms=new_item.trim_start_ms,
+        trim_end_ms=new_item.trim_end_ms,
+        created_at=new_item.created_at,
+        profile_id=generation.profile_id,
+        profile_name=profile_name,
+        text=generation.text,
+        language=generation.language,
+        audio_path=generation.audio_path,
+        duration=generation.duration,
+        seed=generation.seed,
+        instruct=generation.instruct,
+        generation_created_at=generation.created_at,
+    )
+
+    return [original_item_detail, new_item_detail]
+
+
+async def duplicate_story_item(
+    story_id: str,
+    item_id: str,
+    db: Session,
+) -> Optional[StoryItemDetail]:
+    """
+    Duplicate a story item, creating a copy with all properties.
+
+    Args:
+        story_id: Story ID
+        item_id: Story item ID to duplicate
+        db: Database session
+
+    Returns:
+        New item detail or None if not found
+    """
+    # Get the original item
+    original_item = db.query(DBStoryItem).filter_by(
+        id=item_id,
+        story_id=story_id,
+    ).first()
+    if not original_item:
+        return None
+
+    # Get the generation
+    generation = db.query(DBGeneration).filter_by(id=original_item.generation_id).first()
+    if not generation:
+        return None
+
+    # Calculate effective duration
+    current_trim_start = getattr(original_item, 'trim_start_ms', 0)
+    current_trim_end = getattr(original_item, 'trim_end_ms', 0)
+    original_duration_ms = int(generation.duration * 1000)
+    effective_duration_ms = original_duration_ms - current_trim_start - current_trim_end
+
+    # Create duplicate item - place it right after the original
+    new_item = DBStoryItem(
+        id=str(uuid.uuid4()),
+        story_id=story_id,
+        generation_id=original_item.generation_id,  # Same generation as original
+        start_time_ms=original_item.start_time_ms + effective_duration_ms + 200,  # 200ms gap
+        track=original_item.track,
+        trim_start_ms=current_trim_start,
+        trim_end_ms=current_trim_end,
+        created_at=datetime.utcnow(),
+    )
+
+    db.add(new_item)
+
+    # Update story updated_at
+    story = db.query(DBStory).filter_by(id=story_id).first()
+    if story:
+        story.updated_at = datetime.utcnow()
+
+    db.commit()
+    db.refresh(new_item)
+
+    # Get profile name
+    profile = db.query(DBVoiceProfile).filter_by(id=generation.profile_id).first()
+
+    return StoryItemDetail(
+        id=new_item.id,
+        story_id=new_item.story_id,
+        generation_id=new_item.generation_id,
+        start_time_ms=new_item.start_time_ms,
+        track=new_item.track,
+        trim_start_ms=new_item.trim_start_ms,
+        trim_end_ms=new_item.trim_end_ms,
+        created_at=new_item.created_at,
+        profile_id=generation.profile_id,
+        profile_name=profile.name if profile else "Unknown",
+        text=generation.text,
+        language=generation.language,
+        audio_path=generation.audio_path,
+        duration=generation.duration,
+        seed=generation.seed,
+        instruct=generation.instruct,
+        generation_created_at=generation.created_at,
+    )
+
+
 async def update_story_item_times(
     story_id: str,
     data: StoryItemBatchUpdate,
@@ -538,6 +819,8 @@ async def reorder_story_items(
             generation_id=item.generation_id,
             start_time_ms=item.start_time_ms,
             track=item.track,
+            trim_start_ms=getattr(item, 'trim_start_ms', 0),
+            trim_end_ms=getattr(item, 'trim_end_ms', 0),
             created_at=item.created_at,
             profile_id=generation.profile_id,
             profile_name=profile_name,
@@ -602,14 +885,31 @@ async def export_story_audio(
             audio, sr = load_audio(str(audio_path), sample_rate=sample_rate)
             sample_rate = sr  # Use actual sample rate from first file
             
+            # Get trim values
+            trim_start_ms = getattr(item, 'trim_start_ms', 0)
+            trim_end_ms = getattr(item, 'trim_end_ms', 0)
+            
+            # Calculate effective duration
+            original_duration_ms = int(generation.duration * 1000)
+            effective_duration_ms = original_duration_ms - trim_start_ms - trim_end_ms
+            
+            # Slice audio based on trim values
+            trim_start_sample = int((trim_start_ms / 1000.0) * sample_rate)
+            trim_end_sample = int((trim_end_ms / 1000.0) * sample_rate)
+            
+            # Extract the trimmed portion
+            if trim_end_ms > 0:
+                trimmed_audio = audio[trim_start_sample:-trim_end_sample] if trim_end_sample > 0 else audio[trim_start_sample:]
+            else:
+                trimmed_audio = audio[trim_start_sample:]
+            
             # Store audio with its timecode info
             start_time_ms = item.start_time_ms
-            duration_ms = int(generation.duration * 1000)
             
             audio_data.append({
-                'audio': audio,
+                'audio': trimmed_audio,
                 'start_time_ms': start_time_ms,
-                'duration_ms': duration_ms,
+                'duration_ms': effective_duration_ms,
             })
         except Exception:
             # Skip files that can't be loaded

landing/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@voicebox/landing",
-  "version": "0.1.6",
+  "version": "0.1.8",
   "description": "Landing page for voicebox.sh",
   "scripts": {
     "dev": "bun --bun next dev --turbo",

landing/src/app/page.tsx

@@ -99,7 +99,7 @@ export default function Home() {
               <div className="flex justify-center lg:hidden my-8">
                 <div className="w-full max-w-2xl">
                   <Image
-                    src="/VoiceBoxAppScreenshot.webp"
+                    src="/assets/app-screenshot-1.webp"
                     alt="Voicebox Application Screenshot"
                     width={1920}
                     height={1080}
@@ -180,7 +180,7 @@ export default function Home() {
             <div className="hidden lg:block relative">
               <div className="absolute right-0 top-0 -mt-10 w-[200%] -mr-[100%]">
                 <Image
-                  src="/VoiceBoxAppScreenshot.webp"
+                  src="/assets/app-screenshot-1.webp"
                   alt="Voicebox Application Screenshot"
                   width={1920}
                   height={1080}
@@ -193,6 +193,41 @@ export default function Home() {
         </div>
       </section>
 
+      {/* Screenshots Section */}
+      <section className="py-12 sm:py-16 md:py-20">
+        <div className="w-full md:w-[150%] md:-ml-[25%]">
+          <div className="grid grid-cols-1 md:grid-cols-3 gap-6 px-8">
+            <div className="w-full">
+              <Image
+                src="/assets/app-screenshot-2.webp"
+                alt="Voicebox Screenshot 2"
+                width={1920}
+                height={1080}
+                className="w-full h-auto rounded-lg shadow-lg"
+              />
+            </div>
+            <div className="w-full">
+              <Image
+                src="/assets/app-screenshot-1.webp"
+                alt="Voicebox Screenshot 1"
+                width={1920}
+                height={1080}
+                className="w-full h-auto rounded-lg shadow-lg"
+              />
+            </div>
+            <div className="w-full">
+              <Image
+                src="/assets/app-screenshot-3.webp"
+                alt="Voicebox Screenshot 3"
+                width={1920}
+                height={1080}
+                className="w-full h-auto rounded-lg shadow-lg"
+              />
+            </div>
+          </div>
+        </div>
+      </section>
+
       {/* Demo Video Section */}
       <section className="py-12 sm:py-16 md:py-20">
         <div className="container mx-auto px-4 sm:px-6 lg:px-8 max-w-7xl">
@@ -207,7 +242,7 @@ export default function Home() {
                 controls
                 playsInline
                 preload="metadata"
-                poster="/VoiceBoxAppScreenshot.webp"
+                poster="/assets/app-screenshot-1.webp"
               >
                 <source
                   src="/voicebox-demo.webm"

package.json

@@ -1,6 +1,6 @@
 {
   "name": "voicebox",
-  "version": "0.1.6",
+  "version": "0.1.8",
   "private": true,
   "workspaces": [
     "app",

tauri/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@voicebox/tauri",
   "private": true,
-  "version": "0.1.6",
+  "version": "0.1.8",
   "type": "module",
   "scripts": {
     "dev": "vite",

tauri/src-tauri/Cargo.lock

@@ -406,6 +406,16 @@ dependencies = [
  "version_check",
 ]
 
+[[package]]
+name = "core-foundation"
+version = "0.9.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "91e195e091a93c46f7102ec7818a2aa394e1e1771c3ab4825963fa03e45afb8f"
+dependencies = [
+ "core-foundation-sys",
+ "libc",
+]
+
 [[package]]
 name = "core-foundation"
 version = "0.10.1"
@@ -429,9 +439,9 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "fa95a34622365fa5bbf40b20b75dba8dfa8c94c734aea8ac9a5ca38af14316f1"
 dependencies = [
  "bitflags 2.10.0",
- "core-foundation",
+ "core-foundation 0.10.1",
  "core-graphics-types",
- "foreign-types",
+ "foreign-types 0.5.0",
  "libc",
 ]
 
@@ -442,7 +452,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "3d44a101f213f6c4cdc1853d4b78aef6db6bdfa3468798cc1d9912f4735013eb"
 dependencies = [
  "bitflags 2.10.0",
- "core-foundation",
+ "core-foundation 0.10.1",
  "libc",
 ]
 
@@ -889,6 +899,15 @@ version = "1.0.7"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "3f9eec918d3f24069decb9af1554cad7c880e2da24a9afd88aca000531ab82c1"
 
+[[package]]
+name = "foreign-types"
+version = "0.3.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f6f339eb8adc052cd2ca78910fda869aefa38d22d5cb648e6485e4d3fc06f3b1"
+dependencies = [
+ "foreign-types-shared 0.1.1",
+]
+
 [[package]]
 name = "foreign-types"
 version = "0.5.0"
@@ -896,7 +915,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "d737d9aa519fb7b749cbc3b962edcf310a8dd1f4b67c91c4f83975dbdd17d965"
 dependencies = [
  "foreign-types-macros",
- "foreign-types-shared",
+ "foreign-types-shared 0.3.1",
 ]
 
 [[package]]
@@ -910,6 +929,12 @@ dependencies = [
  "syn 2.0.114",
 ]
 
+[[package]]
+name = "foreign-types-shared"
+version = "0.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "00b0228411908ca8685dba7fc2cdd70ec9990a6e753e89b6ac91a84c40fbaf4b"
+
 [[package]]
 name = "foreign-types-shared"
 version = "0.3.1"
@@ -942,6 +967,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "2dff15bf788c671c1934e366d07e30c1814a8ef514e1af724a602e8a2fbe1b10"
 dependencies = [
  "futures-core",
+ "futures-sink",
 ]
 
 [[package]]
@@ -1311,6 +1337,25 @@ dependencies = [
  "syn 2.0.114",
 ]
 
+[[package]]
+name = "h2"
+version = "0.4.12"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f3c0b69cfcb4e1b9f1bf2f53f95f766e4661169728ec61cd3fe5a0166f2d1386"
+dependencies = [
+ "atomic-waker",
+ "bytes",
+ "fnv",
+ "futures-core",
+ "futures-sink",
+ "http",
+ "indexmap 2.13.0",
+ "slab",
+ "tokio",
+ "tokio-util",
+ "tracing",
+]
+
 [[package]]
 name = "hashbrown"
 version = "0.12.3"
@@ -1408,6 +1453,7 @@ dependencies = [
  "bytes",
  "futures-channel",
  "futures-core",
+ "h2",
  "http",
  "http-body",
  "httparse",
@@ -1436,6 +1482,22 @@ dependencies = [
  "webpki-roots",
 ]
 
+[[package]]
+name = "hyper-tls"
+version = "0.6.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "70206fc6890eaca9fde8a0bf71caa2ddfc9fe045ac9e5c70df101a7dbde866e0"
+dependencies = [
+ "bytes",
+ "http-body-util",
+ "hyper",
+ "hyper-util",
+ "native-tls",
+ "tokio",
+ "tokio-native-tls",
+ "tower-service",
+]
+
 [[package]]
 name = "hyper-util"
 version = "0.1.19"
@@ -1455,9 +1517,11 @@ dependencies = [
  "percent-encoding",
  "pin-project-lite",
  "socket2",
+ "system-configuration",
  "tokio",
  "tower-service",
  "tracing",
+ "windows-registry",
 ]
 
 [[package]]
@@ -2024,6 +2088,23 @@ dependencies = [
  "windows-sys 0.60.2",
 ]
 
+[[package]]
+name = "native-tls"
+version = "0.2.14"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "87de3442987e9dbec73158d5c715e7ad9072fda936bb03d19d7fa10e00520f0e"
+dependencies = [
+ "libc",
+ "log",
+ "openssl",
+ "openssl-probe",
+ "openssl-sys",
+ "schannel",
+ "security-framework",
+ "security-framework-sys",
+ "tempfile",
+]
+
 [[package]]
 name = "ndk"
 version = "0.8.0"
@@ -2430,6 +2511,50 @@ dependencies = [
  "pathdiff",
 ]
 
+[[package]]
+name = "openssl"
+version = "0.10.75"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "08838db121398ad17ab8531ce9de97b244589089e290a384c900cb9ff7434328"
+dependencies = [
+ "bitflags 2.10.0",
+ "cfg-if",
+ "foreign-types 0.3.2",
+ "libc",
+ "once_cell",
+ "openssl-macros",
+ "openssl-sys",
+]
+
+[[package]]
+name = "openssl-macros"
+version = "0.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a948666b637a0f465e8564c73e89d4dde00d72d4d473cc972f390fc3dcee7d9c"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn 2.0.114",
+]
+
+[[package]]
+name = "openssl-probe"
+version = "0.1.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d05e27ee213611ffe7d6348b942e8f942b37114c00cc03cec254295a4a17852e"
+
+[[package]]
+name = "openssl-sys"
+version = "0.9.111"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "82cab2d520aa75e3c58898289429321eb788c3106963d0dc886ec7a5f4adc321"
+dependencies = [
+ "cc",
+ "libc",
+ "pkg-config",
+ "vcpkg",
+]
+
 [[package]]
 name = "option-ext"
 version = "0.2.0"
@@ -3077,16 +3202,22 @@ checksum = "eddd3ca559203180a307f12d114c268abf583f59b03cb906fd0b3ff8646c1147"
 dependencies = [
  "base64 0.22.1",
  "bytes",
+ "encoding_rs",
+ "futures-channel",
  "futures-core",
  "futures-util",
+ "h2",
  "http",
  "http-body",
  "http-body-util",
  "hyper",
  "hyper-rustls",
+ "hyper-tls",
  "hyper-util",
  "js-sys",
  "log",
+ "mime",
+ "native-tls",
  "percent-encoding",
  "pin-project-lite",
  "quinn",
@@ -3097,6 +3228,7 @@ dependencies = [
  "serde_urlencoded",
  "sync_wrapper",
  "tokio",
+ "tokio-native-tls",
  "tokio-rustls",
  "tokio-util",
  "tower",
@@ -3232,6 +3364,15 @@ dependencies = [
  "winapi-util",
 ]
 
+[[package]]
+name = "schannel"
+version = "0.1.28"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "891d81b926048e76efe18581bf793546b4c0eaf8448d72be8de2bbee5fd166e1"
+dependencies = [
+ "windows-sys 0.61.2",
+]
+
 [[package]]
 name = "schemars"
 version = "0.8.22"
@@ -3295,6 +3436,29 @@ version = "1.5.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "6ccf069cb109cf8e01ebdca0d55dfce45dbbf669e8c56ed5c62150b056d3ec9f"
 
+[[package]]
+name = "security-framework"
+version = "2.11.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "897b2245f0b511c87893af39b033e5ca9cce68824c4d7e7630b5a1d339658d02"
+dependencies = [
+ "bitflags 2.10.0",
+ "core-foundation 0.9.4",
+ "core-foundation-sys",
+ "libc",
+ "security-framework-sys",
+]
+
+[[package]]
+name = "security-framework-sys"
+version = "2.15.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cc1f0cbffaac4852523ce30d8bd3c5cdc873501d96ff467ca09b6767bb8cd5c0"
+dependencies = [
+ "core-foundation-sys",
+ "libc",
+]
+
 [[package]]
 name = "selectors"
 version = "0.24.0"
@@ -3931,6 +4095,27 @@ dependencies = [
  "syn 2.0.114",
 ]
 
+[[package]]
+name = "system-configuration"
+version = "0.6.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3c879d448e9d986b661742763247d3693ed13609438cf3d006f51f5368a5ba6b"
+dependencies = [
+ "bitflags 2.10.0",
+ "core-foundation 0.9.4",
+ "system-configuration-sys",
+]
+
+[[package]]
+name = "system-configuration-sys"
+version = "0.6.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8e1d1b10ced5ca923a1fcb8d03e96b8d3268065d724548c0211415ff6ac6bac4"
+dependencies = [
+ "core-foundation-sys",
+ "libc",
+]
+
 [[package]]
 name = "system-deps"
 version = "6.2.2"
@@ -3952,7 +4137,7 @@ checksum = "f3a753bdc39c07b192151523a3f77cd0394aa75413802c883a0f6f6a0e5ee2e7"
 dependencies = [
  "bitflags 2.10.0",
  "block2",
- "core-foundation",
+ "core-foundation 0.10.1",
  "core-graphics",
  "crossbeam-channel",
  "dispatch",
@@ -4495,6 +4680,16 @@ dependencies = [
  "syn 2.0.114",
 ]
 
+[[package]]
+name = "tokio-native-tls"
+version = "0.3.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "bbae76ab933c85776efabc971569dd6119c580d8f5d448769dec1764bf796ef2"
+dependencies = [
+ "native-tls",
+ "tokio",
+]
+
 [[package]]
 name = "tokio-rustls"
 version = "0.26.4"
@@ -4826,6 +5021,12 @@ dependencies = [
  "wasm-bindgen",
 ]
 
+[[package]]
+name = "vcpkg"
+version = "0.2.15"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "accd4ea62f7bb7a82fe23066fb0957d48ef677f6eeb8215f372f52e48bb32426"
+
 [[package]]
 name = "version-compare"
 version = "0.2.1"
@@ -4840,7 +5041,7 @@ checksum = "0b928f33d975fc6ad9f86c8f283853ad26bdd5b10b7f1542aa2fa15e2289105a"
 
 [[package]]
 name = "voicebox"
-version = "0.1.5"
+version = "0.1.8"
 dependencies = [
  "base64 0.22.1",
  "core-foundation-sys",
@@ -4848,6 +5049,7 @@ dependencies = [
  "cpal",
  "hound",
  "objc",
+ "reqwest",
  "scopeguard",
  "screencapturekit",
  "serde",
@@ -5330,6 +5532,17 @@ dependencies = [
  "windows-link 0.2.1",
 ]
 
+[[package]]
+name = "windows-registry"
+version = "0.6.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "02752bf7fbdcce7f2a27a742f798510f3e5ad88dbe84871e5168e2120c3d5720"
+dependencies = [
+ "windows-link 0.2.1",
+ "windows-result 0.4.1",
+ "windows-strings 0.5.1",
+]
+
 [[package]]
 name = "windows-result"
 version = "0.1.2"

tauri/src-tauri/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "voicebox"
-version = "0.1.6"
+version = "0.1.8"
 description = "A production-quality desktop app for Qwen3-TTS voice cloning and generation"
 authors = ["you"]
 license = ""
@@ -20,6 +20,7 @@ tauri-plugin-shell = "2.0"
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1.0"
 tokio = { version = "1", features = ["full"] }
+reqwest = { version = "0.12", features = ["blocking", "json"] }
 hound = "3.5"
 base64 = "0.22"
 cpal = "0.15"

tauri/src-tauri/gen/schemas/windows-schema.json

@@ -5948,6 +5948,36 @@
           "const": "fs:write-files",
           "markdownDescription": "This enables all file write related commands without any pre-configured accessible paths."
         },
+        {
+          "description": "This permission set configures which\nprocess features are by default exposed.\n\n#### Granted Permissions\n\nThis enables to quit via `allow-exit` and restart via `allow-restart`\nthe application.\n\n#### This default permission set includes:\n\n- `allow-exit`\n- `allow-restart`",
+          "type": "string",
+          "const": "process:default",
+          "markdownDescription": "This permission set configures which\nprocess features are by default exposed.\n\n#### Granted Permissions\n\nThis enables to quit via `allow-exit` and restart via `allow-restart`\nthe application.\n\n#### This default permission set includes:\n\n- `allow-exit`\n- `allow-restart`"
+        },
+        {
+          "description": "Enables the exit command without any pre-configured scope.",
+          "type": "string",
+          "const": "process:allow-exit",
+          "markdownDescription": "Enables the exit command without any pre-configured scope."
+        },
+        {
+          "description": "Enables the restart command without any pre-configured scope.",
+          "type": "string",
+          "const": "process:allow-restart",
+          "markdownDescription": "Enables the restart command without any pre-configured scope."
+        },
+        {
+          "description": "Denies the exit command without any pre-configured scope.",
+          "type": "string",
+          "const": "process:deny-exit",
+          "markdownDescription": "Denies the exit command without any pre-configured scope."
+        },
+        {
+          "description": "Denies the restart command without any pre-configured scope.",
+          "type": "string",
+          "const": "process:deny-restart",
+          "markdownDescription": "Denies the restart command without any pre-configured scope."
+        },
         {
           "description": "This permission set configures which\nshell functionality is exposed by default.\n\n#### Granted Permissions\n\nIt allows to use the `open` functionality with a reasonable\nscope pre-configured. It will allow opening `http(s)://`,\n`tel:` and `mailto:` links.\n\n#### This default permission set includes:\n\n- `allow-open`",
           "type": "string",

tauri/src-tauri/src/main.rs

@@ -303,6 +303,60 @@ async fn start_server(
     Ok(format!("http://127.0.0.1:{}", SERVER_PORT))
 }
 
+/// Check if a Windows process is still running
+#[cfg(windows)]
+fn is_process_running(pid: u32) -> bool {
+    use std::process::Command;
+    if let Ok(output) = Command::new("tasklist")
+        .args(["/FI", &format!("PID eq {}", pid), "/FO", "CSV", "/NH"])
+        .output()
+    {
+        // If process exists, tasklist returns it in output
+        let output_str = String::from_utf8_lossy(&output.stdout);
+        return !output_str.trim().is_empty() && output_str.contains(&pid.to_string());
+    }
+    false
+}
+
+/// Kill entire Windows process tree by enumerating children
+#[cfg(windows)]
+fn kill_windows_process_tree(parent_pid: u32) -> Result<(), String> {
+    use std::process::Command;
+
+    // Find all child processes using WMIC
+    let output = Command::new("wmic")
+        .args([
+            "process",
+            "where",
+            &format!("ParentProcessId={}", parent_pid),
+            "get",
+            "ProcessId"
+        ])
+        .output();
+
+    if let Ok(output) = output {
+        let output_str = String::from_utf8_lossy(&output.stdout);
+        for line in output_str.lines().skip(1) { // Skip header
+            if let Ok(child_pid) = line.trim().parse::<u32>() {
+                println!("Found child process: {}", child_pid);
+                // Recursively kill child's children
+                let _ = kill_windows_process_tree(child_pid);
+                // Kill the child
+                let _ = Command::new("taskkill")
+                    .args(["/PID", &child_pid.to_string(), "/F"])
+                    .output();
+            }
+        }
+    }
+
+    // Kill the parent process
+    let _ = Command::new("taskkill")
+        .args(["/PID", &parent_pid.to_string(), "/F"])
+        .output();
+
+    Ok(())
+}
+
 #[command]
 async fn stop_server(state: State<'_, ServerState>) -> Result<(), String> {
     let pid = state.server_pid.lock().unwrap().take();
@@ -332,13 +386,59 @@ async fn stop_server(state: State<'_, ServerState>) -> Result<(), String> {
         
         #[cfg(windows)]
         {
-            use std::process::Command;
+            // Layer 1: Try graceful HTTP shutdown first
-            let _ = Command::new("taskkill")
+            println!("Attempting graceful shutdown via HTTP...");
-                .args(["/PID", &pid.to_string(), "/T", "/F"])
+            let client = reqwest::blocking::Client::builder()
-                .output();
+                .timeout(std::time::Duration::from_secs(2))
+                .build()
+                .unwrap();
+
+            let shutdown_result = client
+                .post(&format!("http://127.0.0.1:{}/shutdown", SERVER_PORT))
+                .send();
+
+            if shutdown_result.is_ok() {
+                println!("HTTP shutdown sent, waiting for graceful exit...");
+                // Wait up to 3 seconds for graceful shutdown
+                for i in 0..30 {
+                    std::thread::sleep(std::time::Duration::from_millis(100));
+                    if !is_process_running(pid) {
+                        println!("Process exited gracefully after {}ms", i * 100);
+                        return Ok(());
+                    }
+                }
+                println!("Graceful shutdown timed out, forcing kill...");
+            } else {
+                println!("HTTP shutdown failed, forcing kill...");
+            }
+
+            // Layer 2: Kill process tree with enumeration
+            println!("Killing process tree for wrapper PID {}...", pid);
+            kill_windows_process_tree(pid)?;
+
+            // Layer 3: Verify and kill by name if still running
+            std::thread::sleep(std::time::Duration::from_millis(200));
+            if is_process_running(pid) {
+                println!("Process tree kill failed, killing by name...");
+                use std::process::Command;
+                let _ = Command::new("taskkill")
+                    .args(["/IM", "voicebox-server.exe", "/T", "/F"])
+                    .output();
+            }
+
+            // Layer 4: Final verification
+            std::thread::sleep(std::time::Duration::from_millis(200));
+            if is_process_running(pid) {
+                eprintln!("WARNING: Failed to kill server after all attempts");
+            } else {
+                println!("Server killed successfully");
+            }
         }
 
-        println!("stop_server: Process group kill completed");
+        #[cfg(unix)]
+        {
+            println!("stop_server: Process group kill completed");
+        }
     }
     
     Ok(())
@@ -560,11 +660,54 @@ pub fn run() {
                             
                             #[cfg(windows)]
                             {
-                                // On Windows, use taskkill with /T to kill child processes
+                                // Layer 1: Try graceful HTTP shutdown first
-                                use std::process::Command;
+                                println!("Attempting graceful shutdown via HTTP...");
-                                let _ = Command::new("taskkill")
+                                let client = reqwest::blocking::Client::builder()
-                                    .args(["/PID", &pid.to_string(), "/T", "/F"])
+                                    .timeout(std::time::Duration::from_secs(2))
-                                    .output();
+                                    .build()
+                                    .unwrap();
+
+                                let shutdown_result = client
+                                    .post(&format!("http://127.0.0.1:{}/shutdown", SERVER_PORT))
+                                    .send();
+
+                                if shutdown_result.is_ok() {
+                                    println!("HTTP shutdown sent, waiting for graceful exit...");
+                                    // Wait up to 3 seconds for graceful shutdown
+                                    for i in 0..30 {
+                                        std::thread::sleep(std::time::Duration::from_millis(100));
+                                        if !is_process_running(pid) {
+                                            println!("Process exited gracefully after {}ms", i * 100);
+                                            println!("Server process tree kill completed");
+                                            return;
+                                        }
+                                    }
+                                    println!("Graceful shutdown timed out, forcing kill...");
+                                } else {
+                                    println!("HTTP shutdown failed, forcing kill...");
+                                }
+
+                                // Layer 2: Kill process tree with enumeration
+                                println!("Killing process tree for wrapper PID {}...", pid);
+                                let _ = kill_windows_process_tree(pid);
+
+                                // Layer 3: Verify and kill by name if still running
+                                std::thread::sleep(std::time::Duration::from_millis(200));
+                                if is_process_running(pid) {
+                                    println!("Process tree kill failed, killing by name...");
+                                    use std::process::Command;
+                                    let _ = Command::new("taskkill")
+                                        .args(["/IM", "voicebox-server.exe", "/T", "/F"])
+                                        .output();
+                                }
+
+                                // Layer 4: Final verification
+                                std::thread::sleep(std::time::Duration::from_millis(200));
+                                if is_process_running(pid) {
+                                    eprintln!("WARNING: Failed to kill server after all attempts");
+                                } else {
+                                    println!("Server killed successfully");
+                                }
                                 println!("Server process tree kill completed");
                             }
                         } else {

tauri/src-tauri/tauri.conf.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://schema.tauri.app/config/2",
   "productName": "Voicebox",
-  "version": "0.1.6",
+  "version": "0.1.8",
   "identifier": "sh.voicebox.app",
   "build": {
     "beforeDevCommand": "bun run dev",

web/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@voicebox/web",
   "private": true,
-  "version": "0.1.6",
+  "version": "0.1.8",
   "type": "module",
   "scripts": {
     "dev": "vite",