diff --git a/ppgan/utils/audio.py b/ppgan/utils/audio.py
index 432305bdc..d62c65e95 100644
--- a/ppgan/utils/audio.py
+++ b/ppgan/utils/audio.py
@@ -130,8 +130,8 @@ def _linear_to_mel(spectogram):
 def _build_mel_basis():
     assert audio_config.fmax <= audio_config.sample_rate // 2
     librosa = try_import('librosa')
-    return librosa.filters.mel(audio_config.sample_rate,
-                               audio_config.n_fft,
+    return librosa.filters.mel(sr=audio_config.sample_rate,
+                               n_fft=audio_config.n_fft,
                                n_mels=audio_config.num_mels,
                                fmin=audio_config.fmin,
                                fmax=audio_config.fmax)
diff --git a/requirements.txt b/requirements.txt
index 4df4b5b2a..99b68ae0b 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -5,7 +5,7 @@ scipy>=1.1.0
 opencv-python<=4.6.0.66
 imageio==2.9.0
 imageio-ffmpeg
-librosa==0.8.1
+librosa
 numba
 easydict
 munch
diff --git a/setup.py b/setup.py
index 50bdb31eb..42a524a18 100644
--- a/setup.py
+++ b/setup.py
@@ -43,6 +43,7 @@ def readme():
     url='https://github.com/PaddlePaddle/PaddleGAN',
     download_url='https://github.com/PaddlePaddle/PaddleGAN.git',
     keywords=['gan paddlegan'],
+    python_requires='3.9.20'
     classifiers=[
         'Intended Audience :: Developers', 'Operating System :: OS Independent',
         'Natural Language :: Chinese (Simplified)',