diff --git a/@applications/api/src/modules/chat/chat.module.ts b/@applications/api/src/modules/chat/chat.module.ts
index b1aab74..42c6224 100644
--- a/@applications/api/src/modules/chat/chat.module.ts
+++ b/@applications/api/src/modules/chat/chat.module.ts
@@ -4,9 +4,10 @@ import { ChatController } from './chat.controller';
 import { AiCoreClient } from '../../clients/ai-core.client';
 import { ModelBossClient } from '../../clients/model-boss.client';
 import { SessionModule } from '../session/session.module';
+import { VoiceSharedModule } from '../voice/voice-shared.module';
 
 @Module({
-  imports: [SessionModule],
+  imports: [SessionModule, VoiceSharedModule],
   controllers: [ChatController],
   providers: [ChatService, AiCoreClient, ModelBossClient],
   exports: [AiCoreClient, ModelBossClient],
diff --git a/@applications/api/src/modules/chat/chat.service.ts b/@applications/api/src/modules/chat/chat.service.ts
index bcfc630..b9c8ca3 100644
--- a/@applications/api/src/modules/chat/chat.service.ts
+++ b/@applications/api/src/modules/chat/chat.service.ts
@@ -1,9 +1,12 @@
 import { Injectable, Logger } from '@nestjs/common';
 import { ConfigService } from '@nestjs/config';
+import { randomUUID } from 'node:crypto';
 import type { Response } from 'express';
-import { AiCoreClient } from '../../clients/ai-core.client';
-import { ModelBossClient, ChatMessage } from '../../clients/model-boss.client';
+import { AiCoreClient, type ProcessSegment } from '../../clients/ai-core.client';
+import { ModelBossClient } from '../../clients/model-boss.client';
 import { SessionService } from '../session/session.service';
+import { VoiceServerRef } from '../voice/voice-server.ref';
+import { VoiceSessionStore } from '../voice/voice-session.store';
 
 /**
  * POST /chat SSE pipeline.
@@ -15,16 +18,21 @@ import { SessionService } from '../session/session.service';
  *   4. POST @model-boss /v1/chat/completions (SSE) → token stream
  *   5. Open Socket.IO @ai /process → send init + tokens + done → receive segments
  *   6. SSE each segment to browser
- *   7. Persist user + assistant messages to DB
+ *   7. For each segment: POST @model-boss /api/v1/tts/synthesize → emit binary audio via VoiceGateway
+ *   8. Persist user + assistant messages to DB
  */
 @Injectable()
 export class ChatService {
   private readonly logger = new Logger(ChatService.name);
+  private _seq = 0;
+
   constructor(
     private readonly config: ConfigService,
     private readonly sessionService: SessionService,
     private readonly aiCore: AiCoreClient,
     private readonly modelBoss: ModelBossClient,
+    private readonly voiceServerRef: VoiceServerRef,
+    private readonly voiceSessionStore: VoiceSessionStore,
   ) {}
 
   async streamChat(sessionId: string, userMessage: string, res: Response): Promise<void> {
@@ -35,13 +43,13 @@ export class ChatService {
 
     // 2. Build message history
     const history = await this.sessionService.getHistory(sessionId);
-    const messages: ChatMessage[] = [
-      { role: 'system', content: `/no_think\n${compose.system_prompt}` },
+    const messages = [
+      { role: 'system' as const, content: `/no_think\n${compose.system_prompt}` },
       ...history.map((m) => ({
         role: m.role as 'user' | 'assistant',
         content: m.content,
       })),
-      { role: 'user', content: userMessage },
+      { role: 'user' as const, content: userMessage },
     ];
 
     // 3. Persist user message before streaming
@@ -63,20 +71,29 @@ export class ChatService {
 
     const collectedSegments: Array<{ text: string; emotion: string }> = [];
 
+    // TTS chain: ensures segments are spoken in order (each waits for the previous to finish playing)
+    let ttsChain: Promise<void> = Promise.resolve();
+
     processSocket.onSegment((segment) => {
       collectedSegments.push({ text: segment.text, emotion: segment.emotion });
 
-      const eventData = JSON.stringify({
-        type: 'segment',
-        session_id: sessionId,
-        part_index: segment.partIndex,
-        text: segment.text,
-        emotion: segment.emotion,
-        tts_params: segment.ttsParams,
-        final: false,
-      });
+      res.write(
+        `data: ${JSON.stringify({
+          type: 'segment',
+          session_id: sessionId,
+          part_index: segment.partIndex,
+          text: segment.text,
+          emotion: segment.emotion,
+          tts_params: segment.ttsParams,
+          final: false,
+        })}\n\n`,
+      );
 
-      res.write(`data: ${eventData}\n\n`);
+      ttsChain = ttsChain.then(() =>
+        this.speakSegment(sessionId, segment, compose.tts.emotion).catch((err: Error) => {
+          this.logger.warn(`TTS dispatch failed [${sessionId}] part ${segment.partIndex}: ${err.message}`);
+        }),
+      );
     });
 
     try {
@@ -125,4 +142,70 @@ export class ChatService {
       }
     }
   }
+
+  /**
+   * Synthesise one segment via @model-boss TTS and push the audio to the browser
+   * over the active VoiceGateway Socket.IO connection for this session.
+   *
+   * Resolves only after the audio duration has elapsed so the TTS chain
+   * plays segments sequentially rather than overlapping.
+   *
+   * Silently no-ops if no voice WebSocket is open for this session.
+   */
+  private async speakSegment(
+    sessionId: string,
+    segment: ProcessSegment,
+    emotionConfig: Record<string, unknown>,
+  ): Promise<void> {
+    const voiceSession = this.voiceSessionStore.get(sessionId);
+    if (!voiceSession) return;
+
+    const { server } = this.voiceServerRef;
+    if (!server) return;
+
+    const browserSocket = server.sockets.sockets.get(voiceSession.browserSocketId);
+    if (!browserSocket) return;
+
+    const { pcm, durationMs } = await this.modelBoss.synthesizeTts({
+      text: segment.text,
+      exaggeration: segment.ttsParams.exaggeration,
+      cfgWeight: segment.ttsParams.cfgWeight,
+    });
+
+    const utteranceId = randomUUID().replace(/-/g, '').slice(0, 16);
+    const frame = buildDownstreamFrame(this._seq++, utteranceId, pcm);
+
+    browserSocket.emit('event', {
+      type: 'tts.start',
+      session_id: sessionId,
+      utterance_id: utteranceId,
+      part_index: segment.partIndex,
+      text: segment.text,
+      emotion: segment.emotion,
+    });
+
+    browserSocket.emit('binary', frame);
+
+    // Hold the chain until the audio has had time to play before starting next segment
+    await new Promise<void>((resolve) => setTimeout(resolve, durationMs));
+
+    browserSocket.emit('event', {
+      type: 'tts.end',
+      session_id: sessionId,
+      utterance_id: utteranceId,
+      part_index: segment.partIndex,
+    });
+  }
+}
+
+/**
+ * Build a downstream binary frame for the browser VoiceClient.
+ * Wire format: [0x01][seq: 4B big-endian][utterance_id: 16B ASCII][pcm: Int16 bytes]
+ */
+function buildDownstreamFrame(seq: number, utteranceId: string, pcm: Int16Array): Buffer {
+  const header = Buffer.alloc(21); // 1 type + 4 seq + 16 utteranceId
+  header.writeUInt8(0x01, 0);
+  header.writeUInt32BE(seq, 1);
+  Buffer.from(utteranceId.padEnd(16, '\0').slice(0, 16), 'ascii').copy(header, 5);
+  return Buffer.concat([header, Buffer.from(pcm.buffer, pcm.byteOffset, pcm.byteLength)]);
 }