diff --git a/onnxruntime/python/tools/transformers/models/whisper/convert_to_onnx.py b/onnxruntime/python/tools/transformers/models/whisper/convert_to_onnx.py
index 1b634717ba..eef35db20b 100644
--- a/onnxruntime/python/tools/transformers/models/whisper/convert_to_onnx.py
+++ b/onnxruntime/python/tools/transformers/models/whisper/convert_to_onnx.py
@@ -397,7 +397,7 @@ def main(argv=None):
         ort_session = create_onnxruntime_session(
             args.beam_model_output_dir,
             use_gpu=args.use_gpu,
-            provider=["CUDAExecutionProvider", "CPUExecutionProvider"] if args.use_gpu else ["CPUExecutionProvider"],
+            provider=args.provider,
         )
         device = torch.device("cuda:0" if args.use_gpu else "cpu")