fix(google): emit opus voice-note tts

extensions/xiaomi/speech-provider.test.ts

@@ -1,9 +1,9 @@
 import { afterEach, beforeEach, describe, expect, it, vi } from "vitest";
 
-const runFfmpegMock = vi.hoisted(() => vi.fn());
+const transcodeAudioBufferToOpusMock = vi.hoisted(() => vi.fn());
 
 vi.mock("openclaw/plugin-sdk/media-runtime", () => ({
-  runFfmpeg: runFfmpegMock,
+  transcodeAudioBufferToOpus: transcodeAudioBufferToOpusMock,
 }));
 
 import { buildXiaomiSpeechProvider } from "./speech-provider.js";
@@ -123,7 +123,7 @@ describe("buildXiaomiSpeechProvider", () => {
 
     beforeEach(() => {
       vi.stubGlobal("fetch", vi.fn());
-      runFfmpegMock.mockReset();
+      transcodeAudioBufferToOpusMock.mockReset();
     });
 
     afterEach(() => {
@@ -170,7 +170,7 @@ describe("buildXiaomiSpeechProvider", () => {
         { role: "assistant", content: "Hello from OpenClaw." },
       ]);
       expect(body.audio).toEqual({ format: "mp3", voice: "default_en" });
-      expect(runFfmpegMock).not.toHaveBeenCalled();
+      expect(transcodeAudioBufferToOpusMock).not.toHaveBeenCalled();
     });
 
     it("transcodes Xiaomi output to Opus for voice-note targets", async () => {
@@ -181,15 +181,7 @@ describe("buildXiaomiSpeechProvider", () => {
           headers: { "Content-Type": "application/json" },
         }),
       );
-      runFfmpegMock.mockImplementationOnce(async (args: string[]) => {
-        const outputPath = args.at(-1);
-        if (typeof outputPath !== "string") {
-          throw new Error("missing ffmpeg output path");
-        }
-        await import("node:fs/promises").then((fs) =>
-          fs.writeFile(outputPath, Buffer.from("fake-opus-audio")),
-        );
-      });
+      transcodeAudioBufferToOpusMock.mockResolvedValueOnce(Buffer.from("fake-opus-audio"));
 
       const result = await provider.synthesize({
         text: "Hello from OpenClaw.",
@@ -203,10 +195,12 @@ describe("buildXiaomiSpeechProvider", () => {
       expect(result.fileExtension).toBe(".opus");
       expect(result.voiceCompatible).toBe(true);
       expect(result.audioBuffer.toString()).toBe("fake-opus-audio");
-      expect(runFfmpegMock).toHaveBeenCalledWith(
-        expect.arrayContaining(["-c:a", "libopus", "-ar", "48000"]),
-        { timeoutMs: 30000 },
-      );
+      expect(transcodeAudioBufferToOpusMock).toHaveBeenCalledWith({
+        audioBuffer: Buffer.from("fake-mp3-audio"),
+        inputExtension: "mp3",
+        tempPrefix: "tts-xiaomi-",
+        timeoutMs: 30000,
+      });
     });
 
     it("throws when API key is missing", async () => {

extensions/xiaomi/speech-provider.ts

@@ -1,6 +1,4 @@
-import { mkdir, mkdtemp, readFile, rm, writeFile } from "node:fs/promises";
-import path from "node:path";
-import { runFfmpeg } from "openclaw/plugin-sdk/media-runtime";
+import { transcodeAudioBufferToOpus } from "openclaw/plugin-sdk/media-runtime";
 import { assertOkOrThrowProviderError } from "openclaw/plugin-sdk/provider-http";
 import { normalizeResolvedSecretInputString } from "openclaw/plugin-sdk/secret-input";
 import type {
@@ -14,7 +12,6 @@ import {
   fetchWithSsrFGuard,
   ssrfPolicyFromHttpBaseUrlAllowedHostname,
 } from "openclaw/plugin-sdk/ssrf-runtime";
-import { resolvePreferredOpenClawTmpDir } from "openclaw/plugin-sdk/temp-path";
 
 export const DEFAULT_XIAOMI_TTS_BASE_URL = "https://api.xiaomimimo.com/v1";
 export const DEFAULT_XIAOMI_TTS_MODEL = "mimo-v2.5-tts";
@@ -242,45 +239,6 @@ export async function xiaomiTTS(params: {
   }
 }
 
-async function transcodeAudioToOpus(params: {
-  audioBuffer: Buffer;
-  inputExtension: string;
-  timeoutMs: number | undefined;
-}) {
-  const tempRoot = resolvePreferredOpenClawTmpDir();
-  await mkdir(tempRoot, { recursive: true, mode: 0o700 });
-  const tempDir = await mkdtemp(path.join(tempRoot, "tts-xiaomi-"));
-  try {
-    const inputPath = path.join(tempDir, `input.${params.inputExtension}`);
-    const outputPath = path.join(tempDir, "voice.opus");
-    await writeFile(inputPath, params.audioBuffer, { mode: 0o600 });
-    await runFfmpeg(
-      [
-        "-hide_banner",
-        "-loglevel",
-        "error",
-        "-y",
-        "-i",
-        inputPath,
-        "-vn",
-        "-c:a",
-        "libopus",
-        "-b:a",
-        "64k",
-        "-ar",
-        "48000",
-        "-ac",
-        "1",
-        outputPath,
-      ],
-      { timeoutMs: params.timeoutMs },
-    );
-    return await readFile(outputPath);
-  } finally {
-    await rm(tempDir, { recursive: true, force: true });
-  }
-}
-
 export function buildXiaomiSpeechProvider(): SpeechProviderPlugin {
   return {
     id: "xiaomi",
@@ -313,9 +271,10 @@ export function buildXiaomiSpeechProvider(): SpeechProviderPlugin {
         timeoutMs: req.timeoutMs,
       });
       if (req.target === "voice-note") {
-        const opusBuffer = await transcodeAudioToOpus({
+        const opusBuffer = await transcodeAudioBufferToOpus({
           audioBuffer,
           inputExtension: outputFormat,
+          tempPrefix: "tts-xiaomi-",
           timeoutMs: req.timeoutMs,
         });
         return {