diff --git a/audio/setup.py b/audio/setup.py
index 0fe6e59954741fb32eb8ecae3f7f38a51ca74ede..f7d4594469c89355597a4ba0630b39c3d5c085f7 100644
--- a/audio/setup.py
+++ b/audio/setup.py
@@ -38,8 +38,10 @@ VERSION = '1.2.0'
 COMMITID = 'none'
 
 base = [
+    # paddleaudio align with librosa==0.8.1, which need numpy==1.23.x
+    "librosa==0.8.1",
+    "numpy==1.23.5",
     "kaldiio",
-    "librosa>=0.10.0",
     "pathos",
     "pybind11",
     "parameterized",
diff --git a/setup.py b/setup.py
index 07b411bd048e965bf0107f53b749a2a66fa9c968..af7c4dc3dfc3f1dfb15ccf11c5d61080f196efdb 100644
--- a/setup.py
+++ b/setup.py
@@ -40,6 +40,8 @@ base = [
     "hyperpyyaml",
     "inflect",
     "jsonlines",
+    # paddleaudio align with librosa==0.8.1, which need numpy==1.23.x
+    "numpy==1.23.5",
     "librosa==0.8.1",
     "scipy>=1.4.0",
     "loguru",
@@ -260,6 +262,7 @@ setup_info = dict(
     long_description=read("README.md"),
     long_description_content_type="text/markdown",
     keywords=[
+        "SSL"
         "speech",
         "asr",
         "tts",
@@ -268,12 +271,19 @@ setup_info = dict(
         "text frontend",
         "MFA",
         "paddlepaddle",
+        "paddleaudio",
+        "streaming asr",
+        "streaming tts",
         "beam search",
         "ctcdecoder",
         "deepspeech2",
+        "wav2vec2",
+        "hubert",
+        "wavlm",
         "transformer",
         "conformer",
         "fastspeech2",
+        "hifigan",
         "gan vocoders",
     ],
     python_requires='>=3.7',