diff --git a/convert_torch2paddle.py b/convert_torch2paddle.py
deleted file mode 100644
index b13a793a8..000000000
--- a/convert_torch2paddle.py
+++ /dev/null
@@ -1,18 +0,0 @@
-import paddle
-import torch
-
-torch_model_dict = torch.load('large-v3-turbo.pt')['model_state_dict']
-
-paddle_model_state_dict = {}
-for key, val in torch_model_dict.items():
-    if key.endswith(
-            'weight'
-    ) and val.ndim == 2 and key != "decoder.token_embedding.weight":
-        val = val.T
-    paddle_model_state_dict[key] = paddle.to_tensor(
-        val.cpu().numpy()).astype("float32")
-
-# add other params in case if need, such as:
-paddle_model_state_dict['dims'] = torch.load('large-v3-turbo.pt')['dims']
-
-paddle.save(paddle_model_state_dict, 'weights.params')
diff --git a/paddlespeech/cli/whisper/infer.py b/paddlespeech/cli/whisper/infer.py
index 7207d7467..822ae2549 100644
--- a/paddlespeech/cli/whisper/infer.py
+++ b/paddlespeech/cli/whisper/infer.py
@@ -75,7 +75,7 @@ class WhisperExecutor(BaseExecutor):
             '--size',
             type=str,
             default='large',
-            choices=['large', 'medium', 'base', 'small', 'tiny'],
+            choices=['large', 'medium', 'base', 'small', 'tiny', 'turbo'],
             help='Choose model size. now only support large, large:[whisper-large-16k]'
         )
         self.parser.add_argument(