Simplify Windows TTS playback path

src/audio/local-voice-session.ts

@@ -511,6 +511,11 @@ export class LocalVoiceSession {
   }
 
   private async playToWindowsDefaultSink(playback: PreparedSpeechAudio, signal: AbortSignal): Promise<void> {
+    if (playback.sourceFilePath) {
+      await this.playWindowsWaveFile(playback.sourceFilePath, signal);
+      return;
+    }
+
     const chunks: Buffer[] = [];
 
     await new Promise<void>((resolve, reject) => {
@@ -584,6 +589,47 @@ export class LocalVoiceSession {
     }
   }
 
+  private async playWindowsWaveFile(filePath: string, signal: AbortSignal): Promise<void> {
+    const psScript = [
+      "Add-Type -AssemblyName System;",
+      `$player = New-Object System.Media.SoundPlayer('${filePath.replace(/'/g, "''")}');`,
+      "$player.PlaySync();",
+    ].join(" ");
+
+    const player = spawn("powershell", ["-NoProfile", "-Command", psScript], {
+      stdio: ["ignore", "ignore", "pipe"],
+    });
+    this.currentPlayer = player;
+
+    player.stderr.on("data", (chunk: Buffer) => {
+      const text = chunk.toString().trim();
+      if (text.length > 0) {
+        this.options.logger.debug("[powershell-player]", text);
+      }
+    });
+
+    signal.addEventListener(
+      "abort",
+      () => {
+        if (!player.killed) {
+          player.kill("SIGKILL");
+        }
+      },
+      { once: true },
+    );
+
+    const [code, playSignal] = (await once(player, "exit")) as [number | null, NodeJS.Signals | null];
+    this.currentPlayer = null;
+
+    if (signal.aborted) {
+      return;
+    }
+
+    if (code !== 0) {
+      throw new Error(`powershell playback exited with code=${code ?? "null"} signal=${playSignal ?? "null"}`);
+    }
+  }
+
   private getFfmpegPath(): string {
     return requireFfmpegPath();
   }

src/services/tts.ts

@@ -2,6 +2,7 @@ import type { Readable } from "node:stream";
 
 export interface PreparedSpeechAudio {
   stream: Readable;
+  sourceFilePath?: string;
   dispose: () => void;
 }
 

src/services/windows-system-tts.ts

@@ -4,8 +4,6 @@ import { unlink } from "node:fs/promises";
 import os from "node:os";
 import path from "node:path";
 
-import prism from "prism-media";
-
 import { resolveFfmpegPath } from "../audio/ffmpeg-path.js";
 import type { PreparedSpeechAudio, TtsService } from "./tts.js";
 
@@ -90,45 +88,19 @@ export class WindowsSystemTtsService implements TtsService {
       throw error;
     });
 
-    const input = createReadStream(tempPath);
-    const ffmpeg = new prism.FFmpeg({
-      args: [
-        "-analyzeduration",
-        "0",
-        "-loglevel",
-        "0",
-        "-i",
-        tempPath,
-        "-f",
-        "s16le",
-        "-ar",
-        "48000",
-        "-ac",
-        "2",
-        "pipe:1",
-      ],
-    });
-
-    signal?.addEventListener(
-      "abort",
-      () => {
-        input.destroy();
-        ffmpeg.destroy();
-        void unlink(tempPath).catch(() => null);
-      },
-      { once: true },
-    );
-
     return {
-      stream: ffmpeg,
+      stream: createReadStream(tempPath),
+      sourceFilePath: tempPath,
       dispose: () => {
-        input.destroy();
+        this.cleanupTempWave(tempPath);
-        ffmpeg.destroy();
-        void unlink(tempPath).catch(() => null);
       },
     };
   }
 
+  private cleanupTempWave(filePath: string): void {
+    void unlink(filePath).catch(() => null);
+  }
+
   async destroy(): Promise<void> {
     return;
   }