update

nexa/gguf/nexa_inference_audio_lm.py

@@ -153,7 +153,6 @@ def run(self):
                 user_input = nexa_prompt("Enter text (leave empty if no prompt): ")
                 with suppress_stdout_stderr():
                     response = self.inference(audio_path, user_input)
-                response = response.decode("utf-8") if isinstance(response, bytes) else response
                 print(f"{response}")
 
         except KeyboardInterrupt:
@@ -201,7 +200,7 @@ def inference(self, audio_path: str, prompt: str = "") -> str:
             response = audio_lm_cpp.process_full(
                 self.context, ctypes.byref(self.ctx_params), is_qwen=self.is_qwen
             )
-            return response
+            return response.decode("utf-8") if isinstance(response, bytes) else response
         except Exception as e:
             raise RuntimeError(f"Error during inference: {str(e)}")
         finally:

nexa/gguf/streamlit/streamlit_audio_lm.py

@@ -46,7 +46,7 @@ def process_audio(nexa_model, audio_file, prompt=""):
     try:
         # Use the model's inference method directly
         response = nexa_model.inference(temp_audio_path, prompt)
-        return response.decode("utf-8") if isinstance(response, bytes) else response
+        return response
 
     except Exception as e:
         st.error(f"Error during audio processing: {e}")