diff --git a/docs/requirements.txt b/docs/requirements.txt
index a5409a5448231dbb6793a6feacc2b6cd5ee15809..08a049c1be0089cb236cfd1439985ae2665b44fd 100644
--- a/docs/requirements.txt
+++ b/docs/requirements.txt
@@ -22,6 +22,7 @@ onnxruntime
 pandas
 paddlenlp
 paddlespeech_feat
+Pillow>=9.0.0
 praatio==5.0.0
 pypinyin
 pypinyin-dict
diff --git a/paddlespeech/s2t/frontend/augmentor/spec_augment.py b/paddlespeech/s2t/frontend/augmentor/spec_augment.py
index 296129a942b2860646785b2ef7aa0eb11ab2aa6b..380712851e9b0d5fcb031366da91b7233e1c9ec5 100644
--- a/paddlespeech/s2t/frontend/augmentor/spec_augment.py
+++ b/paddlespeech/s2t/frontend/augmentor/spec_augment.py
@@ -16,7 +16,6 @@ import random
 
 import numpy as np
 from PIL import Image
-from PIL.Image import Resampling
 
 from paddlespeech.s2t.frontend.augmentor.base import AugmentorBase
 from paddlespeech.s2t.utils.log import Log
@@ -164,9 +163,9 @@ class SpecAugmentor(AugmentorBase):
                                       window) + 1  # 1 ... t - 1
 
             left = Image.fromarray(x[:center]).resize((x.shape[1], warped),
-                                                      Resampling.BICUBIC)
+                                                      Image.BICUBIC)
             right = Image.fromarray(x[center:]).resize((x.shape[1], t - warped),
-                                                       Resampling.BICUBIC)
+                                                       Image.BICUBIC)
             if self.inplace:
                 x[:warped] = left
                 x[warped:] = right
diff --git a/paddlespeech/s2t/transform/spec_augment.py b/paddlespeech/s2t/transform/spec_augment.py
index fe24a41618704b458afd4decc9ad38bdc4d73d64..7b3485b1ce497fe519274f52b4ce30f35d91e7ee 100644
--- a/paddlespeech/s2t/transform/spec_augment.py
+++ b/paddlespeech/s2t/transform/spec_augment.py
@@ -14,10 +14,8 @@
 # Modified from espnet(https://github.com/espnet/espnet)
 """Spec Augment module for preprocessing i.e., data augmentation"""
 import random
-
 import numpy
 from PIL import Image
-from PIL.Image import Resampling
 
 from paddlespeech.s2t.transform.functional import FuncTrans
 
@@ -47,9 +45,9 @@ def time_warp(x, max_time_warp=80, inplace=False, mode="PIL"):
                                   window) + 1  # 1 ... t - 1
 
         left = Image.fromarray(x[:center]).resize((x.shape[1], warped),
-                                                  Resampling.BICUBIC)
+                                                  Image.BICUBIC)
         right = Image.fromarray(x[center:]).resize((x.shape[1], t - warped),
-                                                   Resampling.BICUBIC)
+                                                   Image.BICUBIC)
         if inplace:
             x[:warped] = left
             x[warped:] = right
diff --git a/setup.py b/setup.py
index 716c03bc5fd3e0cccd3a034d988ae8e127f8b735..d00b469bef732151958268e042629981fe49dd6e 100644
--- a/setup.py
+++ b/setup.py
@@ -48,6 +48,7 @@ base = [
     "pandas",
     "paddlenlp",
     "paddlespeech_feat",
+    "Pillow>=9.0.0"
     "praatio==5.0.0",
     "pypinyin",
     "pypinyin-dict",
@@ -87,7 +88,6 @@ requirements = {
         "gpustat",
         "paddlespeech_ctcdecoders",
         "phkit",
-        "Pillow",
         "pybind11",
         "pypi-kenlm",
         "snakeviz",