diff --git a/server/src/app.ts b/server/src/app.ts
index e851b90f..389a656c 100644
--- a/server/src/app.ts
+++ b/server/src/app.ts
@@ -32,6 +32,7 @@ import { assistantMemoryRoutes } from "./routes/assistant-memory.js";
 import { assistantHandoffRoutes } from "./routes/assistant-handoff.js";
 import { chatFileRoutes } from "./routes/chat-files.js";
 import { nexusSettingsRoutes } from "./routes/nexus-settings.js";
+import { voiceRoutes } from "./routes/voice.js";
 import { pluginRoutes } from "./routes/plugins.js";
 import { pluginUiStaticRoutes } from "./routes/plugin-ui-static.js";
 import { applyUiBranding } from "./ui-branding.js";
@@ -174,6 +175,7 @@ export async function createApp(
   api.use(assistantHandoffRoutes(db));
   api.use(chatFileRoutes(db, opts.storageService));
   api.use(nexusSettingsRoutes());
+  api.use(voiceRoutes());
   const hostServicesDisposers = new Map<string, () => void>();
   const workerManager = createPluginWorkerManager();
   const pluginRegistry = pluginRegistryService(db);
diff --git a/server/src/routes/chat-files.ts b/server/src/routes/chat-files.ts
index 5c7f993f..25b0c3f5 100644
--- a/server/src/routes/chat-files.ts
+++ b/server/src/routes/chat-files.ts
@@ -294,96 +294,5 @@ export function chatFileRoutes(db: Db, storage: StorageService) {
     res.json(updated);
   });
 
-  // POST /transcribe — Transcribe audio via local Whisper
-  const audioUpload = multer({
-    storage: multer.memoryStorage(),
-    limits: { fileSize: MAX_ATTACHMENT_BYTES, files: 1 },
-  });
-
-  async function runAudioUpload(
-    upload: ReturnType<typeof multer>,
-    req: Request,
-    res: Response,
-  ) {
-    await new Promise<void>((resolve, reject) => {
-      upload.single("audio")(req, res, (err: unknown) => {
-        if (err) reject(err);
-        else resolve();
-      });
-    });
-  }
-
-  router.post("/transcribe", async (req, res) => {
-    assertBoard(req);
-
-    try {
-      await runAudioUpload(audioUpload, req, res);
-    } catch (err) {
-      if (err instanceof multer.MulterError) {
-        res.status(400).json({ error: err.message });
-        return;
-      }
-      throw err;
-    }
-
-    const file = (req as Request & { file?: { buffer: Buffer; mimetype: string } }).file;
-    if (!file) {
-      res.status(400).json({ error: "Missing audio field" });
-      return;
-    }
-
-    // Write to temp file
-    const { writeFile, unlink } = await import("node:fs/promises");
-    const { tmpdir } = await import("node:os");
-    const tmpPath = path.join(tmpdir(), `nexus-audio-${Date.now()}.webm`);
-
-    try {
-      await writeFile(tmpPath, file.buffer);
-
-      // Try whisper CLI (whisper.cpp or openai-whisper)
-      const { promisify } = await import("node:util");
-      const { execFile: execFileCb } = await import("node:child_process");
-      const execFileAsync = promisify(execFileCb);
-
-      try {
-        // Try whisper.cpp first (outputs transcription to stdout with --no-timestamps)
-        const { stdout } = await execFileAsync("whisper-cpp", [
-          "--model", "base.en",
-          "--file", tmpPath,
-          "--no-timestamps",
-          "--output-txt",
-        ], { timeout: 30000 });
-        res.json({ text: stdout.trim() });
-      } catch {
-        try {
-          // Fallback: openai-whisper Python CLI
-          const { stdout } = await execFileAsync("whisper", [
-            tmpPath,
-            "--model", "base.en",
-            "--output_format", "txt",
-            "--output_dir", tmpdir(),
-          ], { timeout: 60000 });
-          // whisper CLI writes to a .txt file alongside the input
-          const txtPath = tmpPath.replace(/\.webm$/, ".txt");
-          try {
-            const { readFile } = await import("node:fs/promises");
-            const text = await readFile(txtPath, "utf-8");
-            await unlink(txtPath).catch(() => {});
-            res.json({ text: text.trim() });
-          } catch {
-            // Parse stdout as fallback
-            res.json({ text: stdout.trim() });
-          }
-        } catch {
-          res.status(503).json({
-            error: "Whisper not available. Install whisper-cpp or openai-whisper for voice input.",
-          });
-        }
-      }
-    } finally {
-      await unlink(tmpPath).catch(() => {});
-    }
-  });
-
   return router;
 }
diff --git a/server/src/routes/chat.ts b/server/src/routes/chat.ts
index 73c7ca28..f8b0e85f 100644
--- a/server/src/routes/chat.ts
+++ b/server/src/routes/chat.ts
@@ -90,7 +90,9 @@ export function chatRoutes(db: Db): Router {
   // POST /api/conversations/:id/stream -- SSE streaming endpoint (CHAT-01, PERF-02)
   router.post("/conversations/:id/stream", async (req, res) => {
     assertBoard(req);
-    const { content, agentId } = req.body;
+    const { content, agentId, voiceMode } = req.body as {
+      content: string; agentId?: string; voiceMode?: "text" | "voice_input" | "full_voice";
+    };
     if (!content || typeof content !== "string") {
       res.status(400).json({ error: "content is required" });
       return;
@@ -139,6 +141,20 @@ export function chatRoutes(db: Db): Router {
     // Add the new user message
     messagesWithMemory.push({ role: "user", content });
 
+    // Inject dual-output formatting prompt when voice mode is full_voice (VPIPE-06)
+    if (voiceMode === "full_voice") {
+      messagesWithMemory.push({
+        role: "system",
+        content: [
+          "Format your response with EXACTLY these two labeled sections:",
+          "",
+          "SPOKEN: [Natural speech prose only. No markdown. No bullet points. No code blocks. Max 2-3 sentences for spoken delivery.]",
+          "",
+          "DETAILED: [Your full response with all detail, code blocks, and markdown formatting.]",
+        ].join("\n"),
+      });
+    }
+
     // Set SSE headers and flush BEFORE any generation (PERF-02)
     res.setHeader("Content-Type", "text/event-stream");
     res.setHeader("Cache-Control", "no-cache");
@@ -168,6 +184,9 @@ export function chatRoutes(db: Db): Router {
           role: "assistant",
           content: fullContent.trim(),
           agentId: agentId || undefined,
+          messageType: voiceMode === "full_voice" ? "voice_full"
+            : voiceMode === "voice_input" ? "voice_input"
+            : undefined,
         });
         res.write(`data: ${JSON.stringify({ type: "done", messageId: message.id, content: fullContent.trim() })}\n\n`);