test

src/server/main.py

@@ -341,11 +341,15 @@ async def synthesize_speech(tts_manager: TTSManager, text: str, ref_audio_name:
     logger.info(f"Synthesizing speech for text: {text[:50]}... with ref_audio: {ref_audio_name}")
     sample_rate, audio_data = load_audio_from_url(ref_audio_url)
     
-    async with await asyncio.to_thread(tempfile.NamedTemporaryFile, suffix=".wav", delete=False) as temp_audio:
+    # Use synchronous context manager and run synthesis in a thread
+    with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as temp_audio:
         sf.write(temp_audio.name, audio_data, samplerate=sample_rate, format='WAV')
         temp_audio.flush()
         audio = await asyncio.to_thread(tts_manager.synthesize, text, temp_audio.name, ref_text)
 
+    # Clean up temporary file
+    os.unlink(temp_audio.name)
+
     if audio.dtype == np.int16:
         audio = audio.astype(np.float32) / 32768.0
     buffer = io.BytesIO()