From 48f8302ebfbc72b4f71a2cb5bf2ece57bbe2e972 Mon Sep 17 00:00:00 2001
From: Naiyuan Qing <145280634+NevilleQingNY@users.noreply.github.com>
Date: Mon, 9 Feb 2026 08:37:57 +0800
Subject: [PATCH] feat(telegram): handle voice and audio messages with
 <media:audio> placeholder

Forward voice messages and audio files to the agent as <media:audio>
placeholder text. In groups, only process voice/audio that replies to
the bot. Includes caption text if present.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 src/channels/plugins/telegram.ts | 45 ++++++++++++++++++++++++++++++++
 1 file changed, 45 insertions(+)

diff --git a/src/channels/plugins/telegram.ts b/src/channels/plugins/telegram.ts
index e1922b57..3174c5a8 100644
--- a/src/channels/plugins/telegram.ts
+++ b/src/channels/plugins/telegram.ts
@@ -120,6 +120,51 @@ export const telegramChannel: ChannelPlugin = {
         });
       });
 
+      // Handle voice messages — forward as <media:audio> placeholder
+      bot.on("message:voice", (ctx) => {
+        const msg = ctx.message;
+        const isGroup = msg.chat.type === "group" || msg.chat.type === "supergroup";
+
+        // In groups, only respond if replied to the bot
+        if (isGroup) {
+          const isReplyToBot = msg.reply_to_message?.from?.is_bot === true;
+          if (!isReplyToBot) return;
+        }
+
+        console.log(`[Telegram] Received voice message: chatId=${msg.chat.id} from=${msg.from?.id} duration=${msg.voice.duration}s`);
+
+        const caption = msg.caption ? ` ${msg.caption}` : "";
+        onMessage({
+          messageId: String(msg.message_id),
+          conversationId: String(msg.chat.id),
+          senderId: String(msg.from?.id ?? "unknown"),
+          text: `<media:audio>${caption}`,
+          chatType: isGroup ? "group" : "direct",
+        });
+      });
+
+      // Handle audio file messages (music files, etc.)
+      bot.on("message:audio", (ctx) => {
+        const msg = ctx.message;
+        const isGroup = msg.chat.type === "group" || msg.chat.type === "supergroup";
+
+        if (isGroup) {
+          const isReplyToBot = msg.reply_to_message?.from?.is_bot === true;
+          if (!isReplyToBot) return;
+        }
+
+        console.log(`[Telegram] Received audio file: chatId=${msg.chat.id} from=${msg.from?.id} title=${msg.audio.title ?? "unknown"}`);
+
+        const caption = msg.caption ? ` ${msg.caption}` : "";
+        onMessage({
+          messageId: String(msg.message_id),
+          conversationId: String(msg.chat.id),
+          senderId: String(msg.from?.id ?? "unknown"),
+          text: `<media:audio>${caption}`,
+          chatType: isGroup ? "group" : "direct",
+        });
+      });
+
       // Graceful shutdown on abort
       signal.addEventListener("abort", () => {
         console.log("[Telegram] Bot stopped");