feat: implement tier-a4 tts voice output replies

src/daemon/routing.ts

@@ -1,6 +1,7 @@
 import type { AudioTranscriptionConfig } from '../models/media.js';
 import type { Attachment } from '../channels/types.js';
 import { isSupportedAudio, transcribeAudio } from '../models/media.js';
+import { synthesizeSpeechAttachment } from '../models/tts.js';
 import { supportsAudioInput } from '../models/capabilities.js';
 import { AgentOrchestrator, type DelegationConfig } from '../backends/index.js';
 import { OutboundAttachmentCollector } from '../backends/native/attachments.js';
@@ -84,6 +85,17 @@ function parseResearchPrefix(text: string): string | undefined {
   }
   return undefined;
 }
+
+function isTtsEnabledForChannel(config: Config, channel: string): boolean {
+  if (!config.tts?.enabled) {
+    return false;
+  }
+  const enabledChannels = config.tts.enabled_channels ?? [];
+  if (enabledChannels.length === 0) {
+    return true;
+  }
+  return enabledChannels.includes(channel);
+}
 /**
  * Create the unified message handler for the channel registry.
  * Each channel+sender pair gets its own AgentOrchestrator backed by a persistent session.
@@ -116,6 +128,31 @@ export function createMessageRouter(deps: {
   const agents = new Map<string, { orchestrator: AgentOrchestrator; collector: OutboundAttachmentCollector }>();
   const talkModeUntil = new Map<string, number>();
 
+  async function maybeBuildTtsAttachment(responseText: string, channel: string) {
+    if (!isTtsEnabledForChannel(deps.config, channel)) {
+      return undefined;
+    }
+
+    const provider = deps.config.tts?.provider;
+    const endpoint = provider?.endpoint ?? (provider?.type === 'openai' ? 'https://api.openai.com/v1/audio/speech' : undefined);
+    if (!endpoint) {
+      return undefined;
+    }
+
+    try {
+      return await synthesizeSpeechAttachment(responseText, {
+        endpoint,
+        apiKey: provider?.api_key,
+        model: provider?.model,
+        voice: provider?.voice,
+        format: provider?.format,
+      });
+    } catch (error) {
+      console.warn(`TTS synthesis failed for channel ${channel}:`, error instanceof Error ? error.message : 'Unknown error');
+      return undefined;
+    }
+  }
+
   function getOrCreateAgent(channel: string, senderId: string, metadata?: Record<string, unknown>, agentOverride?: string): { orchestrator: AgentOrchestrator; collector: OutboundAttachmentCollector } {
     // Resolve agent config name via routing (sender → channel → default fallback)
     const agentConfigName = agentOverride ?? deps.agentRouter?.resolve(channel, senderId);
@@ -998,7 +1035,12 @@ export function createMessageRouter(deps: {
             history,
           });
           session.addMessage({ role: 'assistant', content: response });
-          await reply({ text: response, replyTo: msg.id });
+          const ttsAttachment = await maybeBuildTtsAttachment(response, msg.channel);
+          await reply({
+            text: response,
+            replyTo: msg.id,
+            attachments: ttsAttachment ? [ttsAttachment] : undefined,
+          });
           return;
         } catch (error) {
           const detail = error instanceof Error ? error.message : String(error);
@@ -1031,10 +1073,14 @@ export function createMessageRouter(deps: {
         response = await agent.process(messageText, attachments);
       }
       const outboundAttachments = collector.drain();
+      const ttsAttachment = await maybeBuildTtsAttachment(response, msg.channel);
+      const mergedAttachments = ttsAttachment
+        ? [...outboundAttachments, ttsAttachment]
+        : outboundAttachments;
       await reply({
         text: response,
         replyTo: msg.id,
-        attachments: outboundAttachments.length > 0 ? outboundAttachments : undefined,
+        attachments: mergedAttachments.length > 0 ? mergedAttachments : undefined,
       });
     } catch (error) {
       console.error(`Error processing message from ${msg.channel}:${msg.senderId}:`, error);