diff --git a/.dev_scripts/citest.sh b/.dev_scripts/citest.sh
index c437193c..ce267b1a 100644
--- a/.dev_scripts/citest.sh
+++ b/.dev_scripts/citest.sh
@@ -1,4 +1,9 @@
-pip install -r requirements.txt
+pip install -r requirements.txt -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html
+pip install -r requirements/audio.txt -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html
+pip install -r requirements/cv.txt -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html
+pip install -r requirements/multi-modal.txt -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html
+pip install -r requirements/nlp.txt -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html
+
 pip install -r requirements/tests.txt
 
 
diff --git a/docs/source/faq.md b/docs/source/faq.md
index 2f6a43dc..f4881c5e 100644
--- a/docs/source/faq.md
+++ b/docs/source/faq.md
@@ -41,8 +41,8 @@ reference: [https://huggingface.co/docs/tokenizers/installation#installation-fro
 ```shell
 pip install -f https://download.pytorch.org/whl/torch_stable.html -i https://pypi.tuna.tsinghua.edu.cn/simple -r requirements.txt
 ```
-### 4. zsh: no matches found: model_scope-0.2.2-py3-none-any.whl[all]
+### 4. zsh: no matches found: modelscope-0.2.2-py3-none-any.whl[all]
 mac终端的zsh 对于[]需要做转义，执行如下命令
 ```shell
-pip install model_scope\[all\] -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html
+pip install modelscope\[all\] -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html
 ```
diff --git a/docs/source/quick_start.md b/docs/source/quick_start.md
index 5306cc97..6a188594 100644
--- a/docs/source/quick_start.md
+++ b/docs/source/quick_start.md
@@ -27,12 +27,12 @@ pip install --upgrade tensorflow
 ### pip安装
 执行如下命令：
 ```shell
-pip install "model_scope[all]" -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html
+pip install "modelscope[all]" -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html
 ```
 
 如需体验`语音功能`，请`额外`执行如下命令：
 ```shell
-pip install "model_scope[audio]" -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html
+pip install "modelscope[audio]" -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html
 ```
 ### 使用源码安装
 适合本地开发调试使用，修改源码后可以直接执行
diff --git a/modelscope/pipelines/audio/linear_aec_pipeline.py b/modelscope/pipelines/audio/linear_aec_pipeline.py
index 5ceb499f..3539fe81 100644
--- a/modelscope/pipelines/audio/linear_aec_pipeline.py
+++ b/modelscope/pipelines/audio/linear_aec_pipeline.py
@@ -7,17 +7,24 @@ import scipy.io.wavfile as wav
 import torch
 import yaml
 
+from modelscope.fileio import File
 from modelscope.metainfo import Pipelines
 from modelscope.preprocessors.audio import LinearAECAndFbank
 from modelscope.utils.constant import ModelFile, Tasks
+from modelscope.utils.logger import get_logger
 from ..base import Pipeline
 from ..builder import PIPELINES
 from ..outputs import OutputKeys
 
+logger = get_logger()
+
 FEATURE_MVN = 'feature.DEY.mvn.txt'
 
 CONFIG_YAML = 'dey_mini.yaml'
 
+AEC_LIB_URL = 'https://modelscope.oss-cn-beijing.aliyuncs.com/dependencies/ics_MaaS_AEC_lib_libmitaec_pyio.so'
+AEC_LIB_FILE = 'libmitaec_pyio.so'
+
 
 def initialize_config(module_cfg):
     r"""According to config items, load specific module dynamically with params.
@@ -61,6 +68,13 @@ class LinearAECPipeline(Pipeline):
             model: model id on modelscope hub.
         """
         super().__init__(model=model)
+
+        # auto download so for linux inference before light-weight docker got ready
+        if not os.path.exists(AEC_LIB_FILE):
+            logger.info(f'downloading {AEC_LIB_URL} to {AEC_LIB_FILE}')
+            with open(AEC_LIB_FILE, 'wb') as ofile:
+                ofile.write(File.read(AEC_LIB_URL))
+
         self.use_cuda = torch.cuda.is_available()
         with open(
                 os.path.join(self.model, CONFIG_YAML), encoding='utf-8') as f:
diff --git a/requirements/multi-modal.txt b/requirements/multi-modal.txt
index 9fced4c2..5c149aa1 100644
--- a/requirements/multi-modal.txt
+++ b/requirements/multi-modal.txt
@@ -1,6 +1,6 @@
 fairseq==maas
 ftfy>=6.0.3
-ofa==0.0.2
+ofa==0.0.2-3.6
 pycocoevalcap>=1.2
 pycocotools>=2.0.4
 rouge_score
diff --git a/setup.py b/setup.py
index 2edfd685..97778b89 100644
--- a/setup.py
+++ b/setup.py
@@ -176,17 +176,17 @@ if __name__ == '__main__':
     for field in dir(Fields):
         if field.startswith('_'):
             continue
-        # skip audio requirements due to its hard dependency which
-        # result in mac/windows compatibility problems
-        if field == Fields.audio:
-            continue
         extra_requires[field], _ = parse_requirements(
             f'requirements/{field}.txt')
-        all_requires.append(extra_requires[field])
+
+        # skip audio requirements due to its hard dependency which
+        # result in mac/windows compatibility problems
+        if field != Fields.audio:
+            all_requires.append(extra_requires[field])
     extra_requires['all'] = all_requires
 
     setup(
-        name='model-scope',
+        name='modelscope',
         version=get_version(),
         description='',
         long_description=readme(),
diff --git a/tests/pipelines/test_speech_signal_process.py b/tests/pipelines/test_speech_signal_process.py
index 9e3a8059..9f00c4ef 100644
--- a/tests/pipelines/test_speech_signal_process.py
+++ b/tests/pipelines/test_speech_signal_process.py
@@ -36,8 +36,6 @@ class SpeechSignalProcessTest(unittest.TestCase):
 
     @unittest.skipUnless(test_level() >= 1, 'skip test in current test level')
     def test_aec(self):
-        # A temporary hack to provide c++ lib. Download it first.
-        download(AEC_LIB_URL, AEC_LIB_FILE)
         # Download audio files
         download(NEAREND_MIC_URL, NEAREND_MIC_FILE)
         download(FAREND_SPEECH_URL, FAREND_SPEECH_FILE)