From 229817fdc79942dd1b307ac63a873fc267f135be Mon Sep 17 00:00:00 2001
From: mariko-code-bot <mariko-code-bot@users.noreply.github.com>
Date: Wed, 11 Mar 2026 22:57:47 +0000
Subject: [PATCH] fix(tts): correct return mimeType and include ok in send
 details

---
 src/agents/tools/tts-tool.test.ts | 40 +++++++++++++++++++++++++++++++
 src/agents/tools/tts-tool.ts      |  9 +++++--
 2 files changed, 47 insertions(+), 2 deletions(-)

diff --git a/src/agents/tools/tts-tool.test.ts b/src/agents/tools/tts-tool.test.ts
index 07284bf8105..5b07da266ae 100644
--- a/src/agents/tools/tts-tool.test.ts
+++ b/src/agents/tools/tts-tool.test.ts
@@ -39,10 +39,50 @@ describe("createTtsTool", () => {
       ok: true,
       deliveryMode: "return",
       audioPath: "/tmp/openclaw/tts-test/voice.mp3",
+      mimeType: "audio/mpeg",
       sent: false,
     });
   });
 
+  it("uses audio/ogg mimeType for voice-compatible return output", async () => {
+    textToSpeechMock.mockResolvedValueOnce({
+      success: true,
+      audioPath: "/tmp/openclaw/tts-test/voice.opus",
+      provider: "openai",
+      voiceCompatible: true,
+    });
+    const tool = createTtsTool();
+
+    const result = await tool.execute("call-voice", {
+      text: "hello",
+      channel: "telegram",
+      deliveryMode: "return",
+    });
+
+    expect(result.details).toMatchObject({
+      ok: true,
+      deliveryMode: "return",
+      mimeType: "audio/ogg",
+    });
+  });
+
+  it("includes ok=true in send mode success details", async () => {
+    textToSpeechMock.mockResolvedValueOnce({
+      success: true,
+      audioPath: "/tmp/openclaw/tts-test/voice.mp3",
+      provider: "openai",
+      voiceCompatible: false,
+    });
+    const tool = createTtsTool();
+
+    const result = await tool.execute("call-send", {
+      text: "hello",
+      deliveryMode: "send",
+    });
+
+    expect(result.details).toMatchObject({ ok: true, deliveryMode: "send" });
+  });
+
   it("returns validation error for invalid deliveryMode", async () => {
     const tool = createTtsTool();
     const result = await tool.execute("call-2", {
diff --git a/src/agents/tools/tts-tool.ts b/src/agents/tools/tts-tool.ts
index d313ed7140a..2d54ab8499f 100644
--- a/src/agents/tools/tts-tool.ts
+++ b/src/agents/tools/tts-tool.ts
@@ -72,7 +72,7 @@ export function createTtsTool(opts?: {
               ok: true,
               deliveryMode: "return",
               audioPath: result.audioPath,
-              mimeType: "audio/mpeg",
+              mimeType: result.voiceCompatible ? "audio/ogg" : "audio/mpeg",
               sent: false,
               provider: result.provider,
             },
@@ -87,7 +87,12 @@ export function createTtsTool(opts?: {
         lines.push(`MEDIA:${result.audioPath}`);
         return {
           content: [{ type: "text", text: lines.join("\n") }],
-          details: { audioPath: result.audioPath, provider: result.provider, deliveryMode: "send" },
+          details: {
+            ok: true,
+            audioPath: result.audioPath,
+            provider: result.provider,
+            deliveryMode: "send",
+          },
         };
       }