Add ability to change transcription engine from Settings

New features:
- Settings > Transcription Engine > "Change Transcription Engine"
  button stops the sidecar, deletes downloaded files, and reloads
  the app to show the engine selection screen
- Improved SidecarSetup descriptions with detailed explanations
  of each variant and "Recommended" tag on Cloud (Deepgram)
- Cloud option listed first as the recommended choice
- New reset_sidecar Tauri command that cleans up sidecar files

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

src-tauri/src/lib.rs

@@ -68,6 +68,7 @@ pub fn run() {
             sidecar::get_sidecar_port,
             sidecar::start_sidecar,
             sidecar::stop_sidecar,
+            sidecar::reset_sidecar,
             write_log,
         ])
         .run(tauri::generate_context!())

src-tauri/src/sidecar/mod.rs

@@ -685,6 +685,42 @@ pub fn stop_sidecar(state: tauri::State<'_, ManagedSidecar>) -> Result<(), Strin
     Ok(())
 }
 
+/// Stop the running sidecar, delete its files and version marker.
+/// The next app launch will show the sidecar download prompt.
+#[tauri::command]
+pub fn reset_sidecar(state: tauri::State<'_, ManagedSidecar>) -> Result<(), String> {
+    // Stop the running sidecar first
+    {
+        let mut mgr = state
+            .0
+            .lock()
+            .map_err(|e| format!("Lock error: {e}"))?;
+        mgr.stop();
+    }
+
+    let data = data_dir();
+
+    // Delete the version file so check_sidecar returns false
+    let vf = version_file();
+    if vf.exists() {
+        std::fs::remove_file(&vf)
+            .map_err(|e| format!("Failed to delete version file: {e}"))?;
+    }
+
+    // Delete all sidecar directories
+    if let Ok(entries) = std::fs::read_dir(&data) {
+        for entry in entries.flatten() {
+            let name = entry.file_name().to_string_lossy().to_string();
+            if name.starts_with("sidecar-") && entry.path().is_dir() {
+                eprintln!("[sidecar] Removing {}", entry.path().display());
+                let _ = std::fs::remove_dir_all(entry.path());
+            }
+        }
+    }
+
+    Ok(())
+}
+
 // ---------------------------------------------------------------------------
 // Tests
 // ---------------------------------------------------------------------------

src/lib/components/Settings.svelte

@@ -232,6 +232,18 @@
     }
   }
 
+  async function handleChangeSidecar() {
+    try {
+      const { invoke } = await import("@tauri-apps/api/core");
+      await invoke("reset_sidecar");
+      // Force a page reload which will re-trigger the setup flow
+      window.location.reload();
+    } catch (err) {
+      console.error("Failed to reset sidecar:", err);
+      saveMessage = `Error: ${err}`;
+    }
+  }
+
   async function handleManagedLogin() {
     try {
       await backendStore.apiPost("/api/login", {
@@ -749,6 +761,17 @@
         </div>
         <button onclick={handleCheckUpdates}>Check Now</button>
       </section>
+
+      <!-- Transcription Engine -->
+      <section class="settings-section">
+        <h3>Transcription Engine</h3>
+        <p style="font-size: 12px; color: var(--text-secondary); margin-bottom: 12px;">
+          Switch between local (Whisper) and cloud (Deepgram) transcription engines.
+          This will stop the current engine, remove the downloaded files, and restart
+          with the new engine selection.
+        </p>
+        <button class="danger-btn" onclick={handleChangeSidecar}>Change Transcription Engine</button>
+      </section>
     </div>
 
     <div class="settings-footer">
@@ -919,4 +942,18 @@
   .save-message.error {
     color: #f44336;
   }
+
+  .danger-btn {
+    background: transparent;
+    border: 1px solid var(--accent-red, #f44336);
+    color: var(--accent-red, #f44336);
+    padding: 8px 16px;
+    border-radius: 6px;
+    cursor: pointer;
+    font-size: 13px;
+  }
+
+  .danger-btn:hover {
+    background: rgba(244, 67, 54, 0.1);
+  }
 </style>

src/lib/components/SidecarSetup.svelte

@@ -84,11 +84,29 @@
 
     {#if setupState === "choose"}
       <p class="setup-description">
-        The app needs to download its transcription engine before you can start.
+        Choose a transcription engine. You can change this later in Settings.
-        Choose the version that best fits your hardware.
       </p>
 
       <div class="variant-options">
+        <label class="variant-option" class:selected={variant === "cloud"}>
+          <input
+            type="radio"
+            name="variant"
+            value="cloud"
+            bind:group={variant}
+          />
+          <div class="variant-info">
+            <span class="variant-name">Cloud (Deepgram)</span>
+            <span class="variant-desc">~50 MB download</span>
+            <span class="variant-detail">
+              Fast, accurate streaming transcription via Deepgram's servers.
+              Requires internet and a Deepgram API key.
+              Best for most users — low resource usage, works on any hardware.
+            </span>
+            <span class="variant-tag recommended">Recommended</span>
+          </div>
+        </label>
+
         <label class="variant-option" class:selected={variant === "cpu"}>
           <input
             type="radio"
@@ -97,8 +115,13 @@
             bind:group={variant}
           />
           <div class="variant-info">
-            <span class="variant-name">Standard (CPU)</span>
+            <span class="variant-name">Local - CPU</span>
-            <span class="variant-desc">Works on all computers (~500 MB download)</span>
+            <span class="variant-desc">~500 MB download</span>
+            <span class="variant-detail">
+              Runs Whisper AI models locally on your CPU. No internet needed
+              after download. Good for privacy or offline use, but slower and
+              uses more system resources than cloud.
+            </span>
           </div>
         </label>
 
@@ -110,21 +133,13 @@
             bind:group={variant}
           />
           <div class="variant-info">
-            <span class="variant-name">GPU Accelerated (CUDA)</span>
+            <span class="variant-name">Local - GPU (NVIDIA CUDA)</span>
-            <span class="variant-desc">Faster transcription with NVIDIA GPU (~2 GB download)</span>
+            <span class="variant-desc">~2 GB download</span>
-          </div>
+            <span class="variant-detail">
-        </label>
+              Runs Whisper AI models locally using your NVIDIA GPU for fast
-
+              transcription. No internet needed after download. Requires an
-        <label class="variant-option" class:selected={variant === "cloud"}>
+              NVIDIA GPU with CUDA support.
-          <input
+            </span>
-            type="radio"
-            name="variant"
-            value="cloud"
-            bind:group={variant}
-          />
-          <div class="variant-info">
-            <span class="variant-name">Cloud Only (Deepgram)</span>
-            <span class="variant-desc">Lightweight, requires Deepgram API key (~50 MB download)</span>
           </div>
         </label>
       </div>
@@ -273,6 +288,30 @@
     color: #888;
   }
 
+  .variant-detail {
+    font-size: 11px;
+    color: #666;
+    line-height: 1.4;
+    margin-top: 2px;
+  }
+
+  .variant-tag {
+    display: inline-block;
+    font-size: 10px;
+    font-weight: 600;
+    text-transform: uppercase;
+    letter-spacing: 0.5px;
+    padding: 2px 6px;
+    border-radius: 3px;
+    margin-top: 4px;
+    width: fit-content;
+  }
+
+  .variant-tag.recommended {
+    background: rgba(76, 175, 80, 0.15);
+    color: #4CAF50;
+  }
+
   .download-btn {
     display: block;
     width: 100%;