diff --git a/docs/source/api/modelscope.pipelines.rst b/docs/source/api/modelscope.pipelines.rst
index 167b5cd3..e5758046 100644
--- a/docs/source/api/modelscope.pipelines.rst
+++ b/docs/source/api/modelscope.pipelines.rst
@@ -1,21 +1,6 @@
 modelscope.pipelines package
 ============================
 
-.. automodule:: modelscope.pipelines
-   :members:
-   :undoc-members:
-   :show-inheritance:
-
-Subpackages
------------
-
-.. toctree::
-   :maxdepth: 4
-
-   modelscope.pipelines.audio
-   modelscope.pipelines.cv
-   modelscope.pipelines.multi_modal
-   modelscope.pipelines.nlp
 
 Submodules
 ----------
@@ -25,8 +10,6 @@ modelscope.pipelines.base module
 
 .. automodule:: modelscope.pipelines.base
    :members:
-   :undoc-members:
-   :show-inheritance:
 
 modelscope.pipelines.builder module
 -----------------------------------
diff --git a/docs/source/change_log.md b/docs/source/change_log.md
index 047df483..1081c148 100644
--- a/docs/source/change_log.md
+++ b/docs/source/change_log.md
@@ -1,3 +1,60 @@
+## v 0.2.2 (05/07/2022)
+Second internal release.
+
+### Highlights
+
+### Algorithms
+#### CV
+* add cv-person-image-cartoon pipeline
+* add action recognition pipeline
+* add ocr detection pipeline
+* add animal recognition model
+* add cmdssl video embedding extraction pipeline
+
+#### NLP
+* add speech AEC pipeline
+* add palm2.0
+* add space model
+* add MPLUG model
+* add dialog_intent, dialog_modeling, dialog state tracking pipleline
+* add maskedlm model and fill_mask pipeline
+* add nli pipeline
+* add sentence similarity pipeline
+* add sentiment_classification pipeline
+* add text generation pipeline
+* add translation pipeline
+* add chinese word segmentation pipeline
+* add zero-shot classification
+
+#### Audio
+* add tts pipeline
+* add kws kwsbp pipline
+* add linear aec pipeline
+* add ans pipeline
+
+#### Multi-Modal
+* add image captioning pipeline
+* add multi-modal feature extraction pipeline
+* add text to image synthesis pipeline
+* add VQA pipeline
+
+### Framework
+* add msdataset interface
+* add hub interface and cache support
+* support multiple models in single pipeline
+* add default model configuration for each pipeline
+* remove task field image and video, using cv instead
+* dockerfile support
+* multi-level tests support
+* sphinx-docs use book theme
+* formalize the output of pipeline and make pipeline reusable
+* pipeline refactor and standardize module_name
+* self-host repo support
+
+### Bug Fix
+* support kwargs in pipeline
+* fix errors in task name definition
+
 ## v 0.1.0 (20/05/2022)
 
 First internal release for pipeline inference
diff --git a/docs/source/faq.md b/docs/source/faq.md
index 6ed3b305..2f6a43dc 100644
--- a/docs/source/faq.md
+++ b/docs/source/faq.md
@@ -41,3 +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]
+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
+```
diff --git a/docs/source/index.rst b/docs/source/index.rst
index e93c7aed..11b6ee49 100644
--- a/docs/source/index.rst
+++ b/docs/source/index.rst
@@ -29,15 +29,15 @@ ModelScope doc
 
    change_log.md
 
-.. toctree::
-   :maxdepth: 10
-   :caption: API Doc
-
-   api/modelscope.preprocessors
-   api/modelscope.models
-   api/modelscope.pipelines
-   api/modelscope.fileio
-   api/modelscope.utils
+.. .. toctree::
+..    :maxdepth: 10
+..    :caption: API Doc
+
+..    api/modelscope.preprocessors
+..    api/modelscope.models
+..    api/modelscope.pipelines
+..    api/modelscope.fileio
+..    api/modelscope.utils
 
 
 Indices and tables
diff --git a/docs/source/quick_start.md b/docs/source/quick_start.md
index 0d1f47ac..5306cc97 100644
--- a/docs/source/quick_start.md
+++ b/docs/source/quick_start.md
@@ -1,6 +1,8 @@
 # 快速开始
-ModelScope Library目前支持tensorflow，pytorch深度学习框架进行模型训练、推理， 在Python 3.7+, Pytorch 1.8+, Tensorflow1.15+，Tensorflow 2.6上测试可运行。
-注： 当前（630）版本仅支持python3.7 以及linux环境，其他环境(mac,windows等)支持预计730完成。
+ModelScope Library目前支持tensorflow，pytorch深度学习框架进行模型训练、推理， 在Python 3.7+, Pytorch 1.8+, Tensorflow1.13-1.15，Tensorflow 2.x上测试可运行。
+
+注： 当前（630）版本 `语音相关`的功能仅支持 python3.7,tensorflow1.13-1.15的`linux`环境使用。  其他功能可以在windows、mac上安装使用。
+
 ## python环境配置
 首先，参考[文档](https://docs.anaconda.com/anaconda/install/) 安装配置Anaconda环境
 
@@ -25,7 +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/0.2/repo.html
+pip install "model_scope[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
 ```
 ### 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
@@ -38,9 +45,16 @@ cd modelscope
 ```
 安装依赖并设置PYTHONPATH
 ```shell
-pip install -r requirements.txt
+pip install -e ".[all]" -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html
 export PYTHONPATH=`pwd`
 ```
+注： 6.30版本需要把cv、nlp、multi-modal领域依赖都装上，7.30号各个领域依赖会作为选装，用户需要使用哪个领域安装对应领域依赖即可。
+
+如需使用语音功能，请执行如下命令安装语音功能所需依赖
+```shell
+pip install -e ".[audio]" -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo
+```
+
 ### 安装验证
 安装成功后，可以执行如下命令进行验证安装是否正确
 ```shell
diff --git a/modelscope/models/__init__.py b/modelscope/models/__init__.py
index d11151b5..2e12d6ad 100644
--- a/modelscope/models/__init__.py
+++ b/modelscope/models/__init__.py
@@ -1,29 +1,34 @@
 # Copyright (c) Alibaba, Inc. and its affiliates.
+from modelscope.utils.error import (AUDIO_IMPORT_ERROR,
+                                    TENSORFLOW_IMPORT_WARNING)
 from .base import Model
 from .builder import MODELS, build_model
 
 try:
     from .audio.tts.am import SambertNetHifi16k
     from .audio.tts.vocoder import Hifigan16k
+    from .audio.kws import GenericKeyWordSpotting
+    from .audio.ans.frcrn import FRCRNModel
+except ModuleNotFoundError as e:
+    print(AUDIO_IMPORT_ERROR.format(e))
 
+try:
+    from .nlp.csanmt_for_translation import CsanmtForTranslation
 except ModuleNotFoundError as e:
     if str(e) == "No module named 'tensorflow'":
-        pass
+        print(TENSORFLOW_IMPORT_WARNING.format('CsanmtForTranslation'))
     else:
         raise ModuleNotFoundError(e)
 
 try:
-    from .audio.kws import GenericKeyWordSpotting
     from .multi_modal import OfaForImageCaptioning
     from .nlp import (BertForMaskedLM, BertForSequenceClassification,
-                      CsanmtForTranslation, SbertForNLI,
-                      SbertForSentenceSimilarity,
+                      SbertForNLI, SbertForSentenceSimilarity,
                       SbertForSentimentClassification,
                       SbertForTokenClassification,
                       SbertForZeroShotClassification, SpaceForDialogIntent,
                       SpaceForDialogModeling, SpaceForDialogStateTracking,
                       StructBertForMaskedLM, VecoForMaskedLM)
-    from .audio.ans.frcrn import FRCRNModel
 except ModuleNotFoundError as e:
     if str(e) == "No module named 'pytorch'":
         pass
diff --git a/modelscope/models/nlp/__init__.py b/modelscope/models/nlp/__init__.py
index fb1ff063..fb97fbb5 100644
--- a/modelscope/models/nlp/__init__.py
+++ b/modelscope/models/nlp/__init__.py
@@ -1,5 +1,6 @@
+# Copyright (c) Alibaba, Inc. and its affiliates.
+from modelscope.utils.error import TENSORFLOW_IMPORT_WARNING
 from .bert_for_sequence_classification import *  # noqa F403
-from .csanmt_for_translation import *  # noqa F403
 from .masked_language_model import *  # noqa F403
 from .palm_for_text_generation import *  # noqa F403
 from .sbert_for_nli import *  # noqa F403
@@ -10,3 +11,11 @@ from .sbert_for_zero_shot_classification import *  # noqa F403
 from .space.dialog_intent_prediction_model import *  # noqa F403
 from .space.dialog_modeling_model import *  # noqa F403
 from .space.dialog_state_tracking_model import *  # noqa F403
+
+try:
+    from .csanmt_for_translation import CsanmtForTranslation
+except ModuleNotFoundError as e:
+    if str(e) == "No module named 'tensorflow'":
+        print(TENSORFLOW_IMPORT_WARNING.format('CsanmtForTranslation'))
+    else:
+        raise ModuleNotFoundError(e)
diff --git a/modelscope/pipelines/__init__.py b/modelscope/pipelines/__init__.py
index 74f5507f..5452df28 100644
--- a/modelscope/pipelines/__init__.py
+++ b/modelscope/pipelines/__init__.py
@@ -1,7 +1,12 @@
-from .audio import LinearAECPipeline
-from .audio.ans_pipeline import ANSPipeline
+from modelscope.utils.error import AUDIO_IMPORT_ERROR
 from .base import Pipeline
 from .builder import pipeline
 from .cv import *  # noqa F403
 from .multi_modal import *  # noqa F403
 from .nlp import *  # noqa F403
+
+try:
+    from .audio import LinearAECPipeline
+    from .audio.ans_pipeline import ANSPipeline
+except ModuleNotFoundError as e:
+    print(AUDIO_IMPORT_ERROR.format(e))
diff --git a/modelscope/pipelines/audio/__init__.py b/modelscope/pipelines/audio/__init__.py
index c4dc0100..80c03c23 100644
--- a/modelscope/pipelines/audio/__init__.py
+++ b/modelscope/pipelines/audio/__init__.py
@@ -1,3 +1,7 @@
+# Copyright (c) Alibaba, Inc. and its affiliates.
+
+from modelscope.utils.error import TENSORFLOW_IMPORT_ERROR
+
 try:
     from .kws_kwsbp_pipeline import *  # noqa F403
     from .linear_aec_pipeline import LinearAECPipeline
@@ -11,6 +15,6 @@ try:
     from .text_to_speech_pipeline import *  # noqa F403
 except ModuleNotFoundError as e:
     if str(e) == "No module named 'tensorflow'":
-        pass
+        print(TENSORFLOW_IMPORT_ERROR.format('tts'))
     else:
         raise ModuleNotFoundError(e)
diff --git a/modelscope/pipelines/cv/__init__.py b/modelscope/pipelines/cv/__init__.py
index ce769c44..18dd1e3a 100644
--- a/modelscope/pipelines/cv/__init__.py
+++ b/modelscope/pipelines/cv/__init__.py
@@ -1,3 +1,7 @@
+# Copyright (c) Alibaba, Inc. and its affiliates.
+
+from modelscope.utils.error import TENSORFLOW_IMPORT_ERROR
+
 try:
     from .action_recognition_pipeline import ActionRecognitionPipeline
     from .animal_recog_pipeline import AnimalRecogPipeline
@@ -14,6 +18,8 @@ try:
     from .ocr_detection_pipeline import OCRDetectionPipeline
 except ModuleNotFoundError as e:
     if str(e) == "No module named 'tensorflow'":
-        pass
+        print(
+            TENSORFLOW_IMPORT_ERROR.format(
+                'image-cartoon image-matting ocr-detection'))
     else:
         raise ModuleNotFoundError(e)
diff --git a/modelscope/pipelines/nlp/__init__.py b/modelscope/pipelines/nlp/__init__.py
index c0671de7..c109d49d 100644
--- a/modelscope/pipelines/nlp/__init__.py
+++ b/modelscope/pipelines/nlp/__init__.py
@@ -1,3 +1,15 @@
+# Copyright (c) Alibaba, Inc. and its affiliates.
+
+from modelscope.utils.error import TENSORFLOW_IMPORT_WARNING
+
+try:
+    from .translation_pipeline import *  # noqa F403
+except ModuleNotFoundError as e:
+    if str(e) == "No module named 'tensorflow'":
+        print(TENSORFLOW_IMPORT_WARNING.format('translation'))
+    else:
+        raise ModuleNotFoundError(e)
+
 try:
     from .dialog_intent_prediction_pipeline import *  # noqa F403
     from .dialog_modeling_pipeline import *  # noqa F403
@@ -8,7 +20,6 @@ try:
     from .sentiment_classification_pipeline import *  # noqa F403
     from .sequence_classification_pipeline import *  # noqa F403
     from .text_generation_pipeline import *  # noqa F403
-    from .translation_pipeline import *  # noqa F403
     from .word_segmentation_pipeline import *  # noqa F403
     from .zero_shot_classification_pipeline import *  # noqa F403
 except ModuleNotFoundError as e:
diff --git a/modelscope/preprocessors/__init__.py b/modelscope/preprocessors/__init__.py
index 962e9f6e..29839c2b 100644
--- a/modelscope/preprocessors/__init__.py
+++ b/modelscope/preprocessors/__init__.py
@@ -1,5 +1,6 @@
 # Copyright (c) Alibaba, Inc. and its affiliates.
 
+from modelscope.utils.error import AUDIO_IMPORT_ERROR, TENSORFLOW_IMPORT_ERROR
 from .base import Preprocessor
 from .builder import PREPROCESSORS, build_preprocessor
 from .common import Compose
@@ -9,6 +10,10 @@ from .text_to_speech import *  # noqa F403
 
 try:
     from .audio import LinearAECAndFbank
+except ModuleNotFoundError as e:
+    print(AUDIO_IMPORT_ERROR.format(e))
+
+try:
     from .multi_modal import *  # noqa F403
     from .nlp import *  # noqa F403
     from .space.dialog_intent_prediction_preprocessor import *  # noqa F403
@@ -16,6 +21,6 @@ try:
     from .space.dialog_state_tracking_preprocessor import *  # noqa F403
 except ModuleNotFoundError as e:
     if str(e) == "No module named 'tensorflow'":
-        pass
+        print(TENSORFLOW_IMPORT_ERROR.format('tts'))
     else:
         raise ModuleNotFoundError(e)
diff --git a/modelscope/utils/error.py b/modelscope/utils/error.py
new file mode 100644
index 00000000..ba200379
--- /dev/null
+++ b/modelscope/utils/error.py
@@ -0,0 +1,77 @@
+# Copyright (c) Alibaba, Inc. and its affiliates.
+
+# docstyle-ignore
+AUDIO_IMPORT_ERROR = """
+Audio model import failed: {0}, if you want to use audio releated function, please execute
+`pip install modelscope[audio] -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html`
+"""
+
+# docstyle-ignore
+PROTOBUF_IMPORT_ERROR = """
+{0} requires the protobuf library but it was not found in your environment. Checkout the instructions on the
+installation page of its repo: https://github.com/protocolbuffers/protobuf/tree/master/python#installation and
+follow the ones that match your environment.
+"""
+
+# docstyle-ignore
+SENTENCEPIECE_IMPORT_ERROR = """
+{0} requires the SentencePiece library but it was not found in your environment. Checkout the instructions on the
+installation page of its repo: https://github.com/google/sentencepiece#installation and follow the ones
+that match your environment.
+"""
+
+# docstyle-ignore
+SKLEARN_IMPORT_ERROR = """
+{0} requires the scikit-learn library but it was not found in your environment. You can install it with:
+```
+pip install -U scikit-learn
+```
+In a notebook or a colab, you can install it by executing a cell with
+```
+!pip install -U scikit-learn
+```
+"""
+
+# docstyle-ignore
+TENSORFLOW_IMPORT_ERROR = """
+{0} requires the TensorFlow library but it was not found in your environment. Checkout the instructions on the
+installation page: https://www.tensorflow.org/install and follow the ones that match your environment.
+"""
+
+# docstyle-ignore
+TENSORFLOW_IMPORT_WARNING = """
+{0} requires the TensorFlow library but it was not found in your environment.
+If you don't want to use them, please ignore this message
+If you want to use them, please Checkout the instructions on the
+installation page: https://www.tensorflow.org/install and follow the ones that match your environment.
+"""
+
+# docstyle-ignore
+TIMM_IMPORT_ERROR = """
+{0} requires the timm library but it was not found in your environment. You can install it with pip:
+`pip install timm`
+"""
+
+# docstyle-ignore
+TOKENIZERS_IMPORT_ERROR = """
+{0} requires the 🤗 Tokenizers library but it was not found in your environment. You can install it with:
+```
+pip install tokenizers
+```
+In a notebook or a colab, you can install it by executing a cell with
+```
+!pip install tokenizers
+```
+"""
+
+# docstyle-ignore
+PYTORCH_IMPORT_ERROR = """
+{0} requires the PyTorch library but it was not found in your environment. Checkout the instructions on the
+installation page: https://pytorch.org/get-started/locally/ and follow the ones that match your environment.
+"""
+
+# docstyle-ignore
+SCIPY_IMPORT_ERROR = """
+{0} requires the scipy library but it was not found in your environment. You can install it with pip:
+`pip install scipy`
+"""
diff --git a/modelscope/utils/import_utils.py b/modelscope/utils/import_utils.py
index e4192082..43c6869a 100644
--- a/modelscope/utils/import_utils.py
+++ b/modelscope/utils/import_utils.py
@@ -18,6 +18,12 @@ import json
 from packaging import version
 
 from modelscope.utils.constant import Fields
+from modelscope.utils.error import (PROTOBUF_IMPORT_ERROR,
+                                    PYTORCH_IMPORT_ERROR, SCIPY_IMPORT_ERROR,
+                                    SENTENCEPIECE_IMPORT_ERROR,
+                                    SKLEARN_IMPORT_ERROR,
+                                    TENSORFLOW_IMPORT_ERROR, TIMM_IMPORT_ERROR,
+                                    TOKENIZERS_IMPORT_ERROR)
 from modelscope.utils.logger import get_logger
 
 if sys.version_info < (3, 8):
@@ -111,19 +117,27 @@ ENV_VARS_TRUE_VALUES = {'1', 'ON', 'YES', 'TRUE'}
 ENV_VARS_TRUE_AND_AUTO_VALUES = ENV_VARS_TRUE_VALUES.union({'AUTO'})
 USE_TF = os.environ.get('USE_TF', 'AUTO').upper()
 USE_TORCH = os.environ.get('USE_TORCH', 'AUTO').upper()
+
 _torch_version = 'N/A'
 if USE_TORCH in ENV_VARS_TRUE_AND_AUTO_VALUES and USE_TF not in ENV_VARS_TRUE_VALUES:
     _torch_available = importlib.util.find_spec('torch') is not None
     if _torch_available:
         try:
             _torch_version = importlib_metadata.version('torch')
-            logger.info(f'PyTorch version {_torch_version} available.')
+            logger.info(f'PyTorch version {_torch_version} Found.')
         except importlib_metadata.PackageNotFoundError:
             _torch_available = False
 else:
     logger.info('Disabling PyTorch because USE_TF is set')
     _torch_available = False
 
+_timm_available = importlib.util.find_spec('timm') is not None
+try:
+    _timm_version = importlib_metadata.version('timm')
+    logger.debug(f'Successfully imported timm version {_timm_version}')
+except importlib_metadata.PackageNotFoundError:
+    _timm_available = False
+
 _tf_version = 'N/A'
 if USE_TF in ENV_VARS_TRUE_AND_AUTO_VALUES and USE_TORCH not in ENV_VARS_TRUE_VALUES:
     _tf_available = importlib.util.find_spec('tensorflow') is not None
@@ -153,18 +167,11 @@ if USE_TF in ENV_VARS_TRUE_AND_AUTO_VALUES and USE_TORCH not in ENV_VARS_TRUE_VA
         if version.parse(_tf_version) < version.parse('2'):
             pass
         else:
-            logger.info(f'TensorFlow version {_tf_version} available.')
+            logger.info(f'TensorFlow version {_tf_version} Found.')
 else:
     logger.info('Disabling Tensorflow because USE_TORCH is set')
     _tf_available = False
 
-_timm_available = importlib.util.find_spec('timm') is not None
-try:
-    _timm_version = importlib_metadata.version('timm')
-    logger.debug(f'Successfully imported timm version {_timm_version}')
-except importlib_metadata.PackageNotFoundError:
-    _timm_available = False
-
 
 def is_scipy_available():
     return importlib.util.find_spec('scipy') is not None
@@ -211,68 +218,6 @@ def is_tf_available():
     return _tf_available
 
 
-# docstyle-ignore
-PROTOBUF_IMPORT_ERROR = """
-{0} requires the protobuf library but it was not found in your environment. Checkout the instructions on the
-installation page of its repo: https://github.com/protocolbuffers/protobuf/tree/master/python#installation and
-follow the ones that match your environment.
-"""
-
-# docstyle-ignore
-SENTENCEPIECE_IMPORT_ERROR = """
-{0} requires the SentencePiece library but it was not found in your environment. Checkout the instructions on the
-installation page of its repo: https://github.com/google/sentencepiece#installation and follow the ones
-that match your environment.
-"""
-
-# docstyle-ignore
-SKLEARN_IMPORT_ERROR = """
-{0} requires the scikit-learn library but it was not found in your environment. You can install it with:
-```
-pip install -U scikit-learn
-```
-In a notebook or a colab, you can install it by executing a cell with
-```
-!pip install -U scikit-learn
-```
-"""
-
-# docstyle-ignore
-TENSORFLOW_IMPORT_ERROR = """
-{0} requires the TensorFlow library but it was not found in your environment. Checkout the instructions on the
-installation page: https://www.tensorflow.org/install and follow the ones that match your environment.
-"""
-
-# docstyle-ignore
-TIMM_IMPORT_ERROR = """
-{0} requires the timm library but it was not found in your environment. You can install it with pip:
-`pip install timm`
-"""
-
-# docstyle-ignore
-TOKENIZERS_IMPORT_ERROR = """
-{0} requires the 🤗 Tokenizers library but it was not found in your environment. You can install it with:
-```
-pip install tokenizers
-```
-In a notebook or a colab, you can install it by executing a cell with
-```
-!pip install tokenizers
-```
-"""
-
-# docstyle-ignore
-PYTORCH_IMPORT_ERROR = """
-{0} requires the PyTorch library but it was not found in your environment. Checkout the instructions on the
-installation page: https://pytorch.org/get-started/locally/ and follow the ones that match your environment.
-"""
-
-# docstyle-ignore
-SCIPY_IMPORT_ERROR = """
-{0} requires the scipy library but it was not found in your environment. You can install it with pip:
-`pip install scipy`
-"""
-
 REQUIREMENTS_MAAPING = OrderedDict([
     ('protobuf', (is_protobuf_available, PROTOBUF_IMPORT_ERROR)),
     ('sentencepiece', (is_sentencepiece_available,
diff --git a/modelscope/version.py b/modelscope/version.py
index fc79d63d..020ed73d 100644
--- a/modelscope/version.py
+++ b/modelscope/version.py
@@ -1 +1 @@
-__version__ = '0.2.1'
+__version__ = '0.2.2'
diff --git a/requirements/audio.txt b/requirements/audio.txt
index 4c009d27..1e1e577b 100644
--- a/requirements/audio.txt
+++ b/requirements/audio.txt
@@ -6,7 +6,7 @@ librosa
 lxml
 matplotlib
 nara_wpe
-numpy
+numpy<=1.18
 protobuf>3,<=3.20
 ptflops
 pytorch_wavelets==1.3.0
diff --git a/requirements/multi-modal.txt b/requirements/multi-modal.txt
index b96bdd01..9fced4c2 100644
--- a/requirements/multi-modal.txt
+++ b/requirements/multi-modal.txt
@@ -5,3 +5,4 @@ pycocoevalcap>=1.2
 pycocotools>=2.0.4
 rouge_score
 timm
+torchvision
diff --git a/setup.py b/setup.py
index 3b40ac8b..2edfd685 100644
--- a/setup.py
+++ b/setup.py
@@ -176,6 +176,10 @@ 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])