Update app.py

app.py

@@ -31,7 +31,7 @@ phoneme_vocab = config['vocab']
 # Download the model and voice files from Hugging Face Hub
 # ------------------------------------------------------------------------------
 model_repo = "onnx-community/Kokoro-82M-v1.0-ONNX"
-model_name = "onnx/model_quantized.onnx"
+model_name = "onnx/model_q8f16.onnx"
 voice_file_pattern = "*.bin"
 local_dir = "."
 snapshot_download(
@@ -241,14 +241,9 @@ def tts_streaming(text: str, voice: str = "af_heart", speed: float = 1.0, format
             # audio_tensor = torch.from_numpy(audio_int16.astype(np.float32) / 32767)
 
             # Yield the encoded audio chunk.
-            if format.lower() == "wav":
-                yield audio_int16
-            elif format.lower() == "opus":
-                yield audio_tensor_to_opus_bytes(audio_tensor, sample_rate=sample_rate)
-            else:
-                raise HTTPException(status_code=400, detail=f"Unsupported audio format: {format}")
+            yield audio_int16.tobytes()
 
-    media_type = "audio/wav" if format.lower() == "wav" else "audio/opus"
+    media_type = "audio/wav"
     return StreamingResponse(
         audio_generator(),
         media_type=media_type,