diff --git a/xtts/bridge.js b/xtts/bridge.js
index 5b9fc11..0a6da4e 100644
--- a/xtts/bridge.js
+++ b/xtts/bridge.js
@@ -195,32 +195,27 @@ async function _runTTSRequest(payload) {
 }
 
 /**
- * Ruft /tts_to_audio/ auf und streamt das Response-Body chunkweise an
- * den Callback. Kein echtes Server-Streaming (XTTS rendert komplett
- * bevor es antwortet), aber stabil und mit der Queue + grosszuegigem
- * AudioTrack-Buffer klingt's akzeptabel.
+ * Ruft /tts_stream (GET) auf — echter Streaming-Endpoint bei daswer123.
+ * Samples fliessen waehrend XTTS rendert (chunked transfer).
+ * Time-to-first-audio ~300-500ms statt 2-4s beim batch-Endpoint.
  *
- * /tts_stream ist elegant, funktioniert aber nicht in allen Versionen
- * von daswer123/xtts-api-server.
+ * Parameter werden als Query-String uebergeben (GET-API).
  */
 function streamXTTSAsPCM(text, language, speakerWav, onPcmChunk) {
   return new Promise((resolve, reject) => {
-    const body = JSON.stringify({
+    const qs = new URLSearchParams({
       text,
-      language,
-      speaker_wav: speakerWav || "",
+      language: language || "de",
+      speaker_wav: speakerWav ? speakerWav : "",
+      stream_chunk_size: "40",
     });
 
-    const url = new URL(`${XTTS_API_URL}/tts_to_audio/`);
+    const url = new URL(`${XTTS_API_URL}/tts_stream?${qs.toString()}`);
     const options = {
       hostname: url.hostname,
       port: url.port,
-      path: url.pathname,
-      method: "POST",
-      headers: {
-        "Content-Type": "application/json",
-        "Content-Length": Buffer.byteLength(body),
-      },
+      path: `${url.pathname}?${url.searchParams.toString()}`,
+      method: "GET",
       timeout: 60000,
     };
 
@@ -282,7 +277,6 @@ function streamXTTSAsPCM(text, language, speakerWav, onPcmChunk) {
 
     req.on("error", reject);
     req.on("timeout", () => { req.destroy(); reject(new Error("XTTS API Timeout (60s)")); });
-    req.write(body);
     req.end();
   });
 }