diff --git a/deepspeech/__init__.py b/deepspeech/__init__.py
index a1617c78a54d6b1d7083f095aedb015e9d7434a6..8be0e93f4961781ba4d9a8bf195f98b5e097de86 100644
--- a/deepspeech/__init__.py
+++ b/deepspeech/__init__.py
@@ -11,7 +11,6 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-import logging
 from typing import Union
 from typing import List
 from typing import Tuple
@@ -23,30 +22,11 @@ from paddle.nn import functional as F
 #TODO(Hui Zhang): remove  fluid import
 from paddle.fluid import core
 
-root = logging.getLogger()
-root.setLevel(logging.DEBUG)
-
-format = '[%(levelname)s %(asctime)s %(filename)s:%(lineno)d] %(message)s'
-formatter = logging.Formatter(fmt=format, datefmt='%Y/%m/%d %H:%M:%S')
-
-stream_handler = logging.StreamHandler()
-stream_handler.setLevel(logging.INFO)
-stream_handler.setFormatter(formatter)
-root.addHandler(stream_handler)
-save_path = 'global.log'
-file_handler = logging.FileHandler(save_path)
-file_handler.setLevel(logging.DEBUG)
-file_handler.setFormatter(formatter)
-root.addHandler(file_handler)
-
-# stop propagate for propagating may print
-# log multiple times
-root.propagate = False
-
-logger = logging.getLogger(__name__)
+from deepspeech.utils.log import Log
+logger = Log(__name__).getlog()
 
 ########### hcak logging #############
-logger.warn = logging.warning
+logger.warn = logger.warning
 
 ########### hcak paddle #############
 paddle.bool = 'bool'
diff --git a/deepspeech/exps/deepspeech2/model.py b/deepspeech/exps/deepspeech2/model.py
index fcc303a3af3079846da1df98a72bdec3cf7dc12f..ffe7d717ceb7529abf386bc5ef1e7e7ae8f92663 100644
--- a/deepspeech/exps/deepspeech2/model.py
+++ b/deepspeech/exps/deepspeech2/model.py
@@ -14,7 +14,7 @@
 """Contains DeepSpeech2 model."""
 
 import time
-import logging
+from deepspeech.utils.log import Log
 import numpy as np
 from collections import defaultdict
 from pathlib import Path
@@ -38,7 +38,7 @@ from deepspeech.io.dataset import ManifestDataset
 from deepspeech.models.deepspeech2 import DeepSpeech2Model
 from deepspeech.models.deepspeech2 import DeepSpeech2InferModel
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 
 class DeepSpeech2Trainer(Trainer):
diff --git a/deepspeech/exps/u2/model.py b/deepspeech/exps/u2/model.py
index d02818c202ae0a9c4c21a443a8d0c291ef1b7b8d..646b672592e39f07e482d1edcdb682aacdaeea0b 100644
--- a/deepspeech/exps/u2/model.py
+++ b/deepspeech/exps/u2/model.py
@@ -15,7 +15,7 @@
 
 import sys
 import time
-import logging
+from deepspeech.utils.log import Log
 import numpy as np
 from collections import defaultdict
 from pathlib import Path
@@ -41,7 +41,7 @@ from deepspeech.io.dataset import ManifestDataset
 
 from deepspeech.models.u2 import U2Model
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 
 class U2Trainer(Trainer):
diff --git a/deepspeech/frontend/augmentor/spec_augment.py b/deepspeech/frontend/augmentor/spec_augment.py
index 2b648daba99b217fc86357792f18a6f518216cf7..59240be5a1f5504c479ce5ebefe3a22e49963cff 100644
--- a/deepspeech/frontend/augmentor/spec_augment.py
+++ b/deepspeech/frontend/augmentor/spec_augment.py
@@ -12,12 +12,12 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 """Contains the volume perturb augmentation model."""
-import logging
+from deepspeech.utils.log import Log
 import numpy as np
 
 from deepspeech.frontend.augmentor.base import AugmentorBase
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 
 class SpecAugmentor(AugmentorBase):
diff --git a/deepspeech/frontend/utility.py b/deepspeech/frontend/utility.py
index 8c5581eb09716ce9a749a15b06f988dea611f287..657a71a9f5f569e87129c72f64f7496a27b65f28 100644
--- a/deepspeech/frontend/utility.py
+++ b/deepspeech/frontend/utility.py
@@ -16,9 +16,9 @@ import numpy as np
 import math
 import json
 import codecs
-import logging
+from deepspeech.utils.log import Log
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = [
     "load_cmvn", "read_manifest", "rms_to_db", "rms_to_dbfs", "max_dbfs",
diff --git a/deepspeech/io/collator.py b/deepspeech/io/collator.py
index 4edb4ca99eb4c581c9c03e062d2df15e340fcd66..310333d04ed7d315956e5dd251b28146e33b61ee 100644
--- a/deepspeech/io/collator.py
+++ b/deepspeech/io/collator.py
@@ -12,16 +12,16 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import logging
+from deepspeech.utils.log import Log
 import numpy as np
 
 from deepspeech.io.utility import pad_sequence
 from deepspeech.frontend.utility import IGNORE_ID
 
-logger = logging.getLogger(__name__)
-
 __all__ = ["SpeechCollator"]
 
+logger = Log(__name__).getlog()
+
 
 class SpeechCollator():
     def __init__(self, keep_transcription_text=True):
diff --git a/deepspeech/io/dataset.py b/deepspeech/io/dataset.py
index a660ae49411f3aa5f5653a09029b2f353d6ea443..a33724947d105696005f7a39c6ecaa59a5dededc 100644
--- a/deepspeech/io/dataset.py
+++ b/deepspeech/io/dataset.py
@@ -16,7 +16,7 @@ import time
 import io
 import random
 import tarfile
-import logging
+from deepspeech.utils.log import Log
 from typing import Optional
 from yacs.config import CfgNode
 from collections import namedtuple
@@ -29,12 +29,12 @@ from deepspeech.frontend.featurizer.speech_featurizer import SpeechFeaturizer
 from deepspeech.frontend.speech import SpeechSegment
 from deepspeech.frontend.normalizer import FeatureNormalizer
 
-logger = logging.getLogger(__name__)
-
 __all__ = [
     "ManifestDataset",
 ]
 
+logger = Log(__name__).getlog()
+
 
 class ManifestDataset(Dataset):
     @classmethod
diff --git a/deepspeech/io/sampler.py b/deepspeech/io/sampler.py
index 7acb88cf24c79ce741f5b507137cec1224e00d7a..910169af67e4e86fc3b99e2d4145d927721d5a2e 100644
--- a/deepspeech/io/sampler.py
+++ b/deepspeech/io/sampler.py
@@ -13,20 +13,20 @@
 # limitations under the License.
 
 import math
-import logging
+from deepspeech.utils.log import Log
 import numpy as np
 
 from paddle.io import BatchSampler
 from paddle.io import DistributedBatchSampler
 from paddle import distributed as dist
 
-logger = logging.getLogger(__name__)
-
 __all__ = [
     "SortagradDistributedBatchSampler",
     "SortagradBatchSampler",
 ]
 
+logger = Log(__name__).getlog()
+
 
 def _batch_shuffle(indices, batch_size, epoch, clipped=False):
     """Put similarly-sized instances into minibatches for better efficiency
diff --git a/deepspeech/io/utility.py b/deepspeech/io/utility.py
index 4c2ff76bd353b6e539d0b9f75180b8f3b870987e..de0e5234907105e21c8f518dd7256c91b2c80528 100644
--- a/deepspeech/io/utility.py
+++ b/deepspeech/io/utility.py
@@ -12,14 +12,14 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import logging
+from deepspeech.utils.log import Log
 import numpy as np
 from typing import List
 
-logger = logging.getLogger(__name__)
-
 __all__ = ["pad_sequence"]
 
+logger = Log(__name__).getlog()
+
 
 def pad_sequence(sequences: List[np.ndarray],
                  batch_first: bool=True,
diff --git a/deepspeech/models/deepspeech2.py b/deepspeech/models/deepspeech2.py
index 884fa4b1f1428a9285eb70336020e5337f5f1000..a3ad9fed44f475c1bf7220c719df5f97ad3fda53 100644
--- a/deepspeech/models/deepspeech2.py
+++ b/deepspeech/models/deepspeech2.py
@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 """Deepspeech2 ASR Model"""
-import logging
+from deepspeech.utils.log import Log
 from typing import Optional
 from yacs.config import CfgNode
 
@@ -26,7 +26,7 @@ from deepspeech.modules.ctc import CTCDecoder
 from deepspeech.utils import checkpoint
 from deepspeech.utils import layer_tools
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = ['DeepSpeech2Model']
 
diff --git a/deepspeech/models/u2.py b/deepspeech/models/u2.py
index af7a11e1c8789f23253a3ec8b3a3c9dd1a193310..04a57c1ee0d677b3e263bf3dc2c44a002951b2ae 100644
--- a/deepspeech/models/u2.py
+++ b/deepspeech/models/u2.py
@@ -19,7 +19,7 @@ Unified Streaming and Non-streaming Two-pass End-to-end Model for Speech Recogni
 import time
 import sys
 from collections import defaultdict
-import logging
+from deepspeech.utils.log import Log
 from yacs.config import CfgNode
 from typing import List, Optional, Tuple, Dict
 
@@ -50,7 +50,7 @@ from deepspeech.utils.tensor_utils import th_accuracy
 from deepspeech.utils.tensor_utils import pad_sequence
 from deepspeech.utils.ctc_utils import remove_duplicates_and_blank
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = ["U2Model", "U2InferModel"]
 
diff --git a/deepspeech/modules/activation.py b/deepspeech/modules/activation.py
index 6fd2fa51c3bceead9c6c72929649715f7c6ca880..eb0297dd382e6ecb3872b8cf60f314452fe181bf 100644
--- a/deepspeech/modules/activation.py
+++ b/deepspeech/modules/activation.py
@@ -12,13 +12,13 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import logging
+from deepspeech.utils.log import Log
 from collections import OrderedDict
 
 import paddle
 from paddle import nn
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = ["get_activation", "brelu", "LinearGLUBlock", "ConvGLUBlock"]
 
diff --git a/deepspeech/modules/attention.py b/deepspeech/modules/attention.py
index f2eb77e4709fc1cc9f7243c82fa50c887f8ce0ec..19c3381777a4b1fb3b7d7111ffa6e560bdb919f0 100644
--- a/deepspeech/modules/attention.py
+++ b/deepspeech/modules/attention.py
@@ -13,14 +13,14 @@
 # limitations under the License.
 """Multi-Head Attention layer definition."""
 import math
-import logging
+from deepspeech.utils.log import Log
 from typing import Optional, Tuple
 
 import paddle
 from paddle import nn
 from paddle.nn import initializer as I
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = ["MultiHeadedAttention", "RelPositionMultiHeadedAttention"]
 
diff --git a/deepspeech/modules/cmvn.py b/deepspeech/modules/cmvn.py
index 562efd4170ab7942237b9f1935326d41f8ff0d62..0d65242bcc2231441d55018e21cc09d85f4ea464 100644
--- a/deepspeech/modules/cmvn.py
+++ b/deepspeech/modules/cmvn.py
@@ -12,12 +12,12 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import logging
+from deepspeech.utils.log import Log
 
 import paddle
 from paddle import nn
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = ['GlobalCMVN']
 
diff --git a/deepspeech/modules/conformer_convolution.py b/deepspeech/modules/conformer_convolution.py
index 1a9c59f42c5bb90de5409f793562c3736c6a8081..2e6d40bb1f159d568f3d8e32de63c0ad28e70e8c 100644
--- a/deepspeech/modules/conformer_convolution.py
+++ b/deepspeech/modules/conformer_convolution.py
@@ -15,12 +15,12 @@
 from typing import Optional, Tuple
 from typeguard import check_argument_types
 
-import logging
+from deepspeech.utils.log import Log
 
 import paddle
 from paddle import nn
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = ['ConvolutionModule']
 
diff --git a/deepspeech/modules/conv.py b/deepspeech/modules/conv.py
index 4c3374286ac8b61a45d340e28aeb29810b03cd10..c86b36f472e5f0cb35faa82f8c727e80e499c544 100644
--- a/deepspeech/modules/conv.py
+++ b/deepspeech/modules/conv.py
@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import logging
+from deepspeech.utils.log import Log
 
 from paddle import nn
 from paddle.nn import functional as F
@@ -20,7 +20,7 @@ from paddle.nn import functional as F
 from deepspeech.modules.mask import sequence_mask
 from deepspeech.modules.activation import brelu
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = ['ConvStack', "conv_output_size"]
 
diff --git a/deepspeech/modules/ctc.py b/deepspeech/modules/ctc.py
index 1cd7a3c850a12989ac70acade0d4c91c69ebbb67..fc0fc8f3839ed9d953354156198653a58ab51602 100644
--- a/deepspeech/modules/ctc.py
+++ b/deepspeech/modules/ctc.py
@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import logging
+from deepspeech.utils.log import Log
 from typeguard import check_argument_types
 
 import paddle
@@ -26,7 +26,7 @@ from deepspeech.decoders.swig_wrapper import Scorer
 from deepspeech.decoders.swig_wrapper import ctc_greedy_decoder
 from deepspeech.decoders.swig_wrapper import ctc_beam_search_decoder_batch
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = ['CTCDecoder']
 
diff --git a/deepspeech/modules/decoder.py b/deepspeech/modules/decoder.py
index 11f18f7bfcee16023189cdbeaecc9bea2fe51f00..6d6e0944aa299a341a3c777dcdca4354488eeb21 100644
--- a/deepspeech/modules/decoder.py
+++ b/deepspeech/modules/decoder.py
@@ -14,7 +14,7 @@
 """Decoder definition."""
 from typing import Tuple, List, Optional
 from typeguard import check_argument_types
-import logging
+from deepspeech.utils.log import Log
 
 import paddle
 from paddle import nn
@@ -26,7 +26,7 @@ from deepspeech.modules.positionwise_feed_forward import PositionwiseFeedForward
 from deepspeech.modules.mask import subsequent_mask
 from deepspeech.modules.mask import make_non_pad_mask
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = ["TransformerDecoder"]
 
diff --git a/deepspeech/modules/decoder_layer.py b/deepspeech/modules/decoder_layer.py
index a781ea5d5b4f0ddad2b16dd48d21c5ea6a596cdd..84490290e96c9a446dbdf0e70fa81723aa18864d 100644
--- a/deepspeech/modules/decoder_layer.py
+++ b/deepspeech/modules/decoder_layer.py
@@ -13,12 +13,12 @@
 # limitations under the License.
 """Decoder self-attention layer definition."""
 from typing import Optional, Tuple
-import logging
+from deepspeech.utils.log import Log
 
 import paddle
 from paddle import nn
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = ["DecoderLayer"]
 
diff --git a/deepspeech/modules/embedding.py b/deepspeech/modules/embedding.py
index df8e346a7d45cec75fdab9b940ca59eb5f5a8f4c..8d7a842e10e9aa80de6ec883606713d93ea2c598 100644
--- a/deepspeech/modules/embedding.py
+++ b/deepspeech/modules/embedding.py
@@ -14,13 +14,13 @@
 """Positonal Encoding Module."""
 
 import math
-import logging
+from deepspeech.utils.log import Log
 from typing import Tuple
 
 import paddle
 from paddle import nn
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = ["PositionalEncoding", "RelPositionalEncoding"]
 
diff --git a/deepspeech/modules/encoder.py b/deepspeech/modules/encoder.py
index e2ea57d3917e0a1ccd7d4ae17a9c4e92185c62f5..70b01eb4aaadec7b83bb629626766744119b35b8 100644
--- a/deepspeech/modules/encoder.py
+++ b/deepspeech/modules/encoder.py
@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 """Encoder definition."""
-import logging
+from deepspeech.utils.log import Log
 from typing import Tuple, List, Optional
 from typeguard import check_argument_types
 
@@ -35,7 +35,7 @@ from deepspeech.modules.mask import make_non_pad_mask
 from deepspeech.modules.mask import add_optional_chunk_mask
 from deepspeech.modules.activation import get_activation
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = ["BaseEncoder", 'TransformerEncoder', "ConformerEncoder"]
 
diff --git a/deepspeech/modules/encoder_layer.py b/deepspeech/modules/encoder_layer.py
index 6ae69852e7ef6ae539f865eddcd597f6aefea39a..5dec456526d2f96869252061db74b8c129b3dd47 100644
--- a/deepspeech/modules/encoder_layer.py
+++ b/deepspeech/modules/encoder_layer.py
@@ -13,12 +13,12 @@
 # limitations under the License.
 """Encoder self-attention layer definition."""
 from typing import Optional, Tuple
-import logging
+from deepspeech.utils.log import Log
 
 import paddle
 from paddle import nn
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = ["TransformerEncoderLayer", "ConformerEncoderLayer"]
 
diff --git a/deepspeech/modules/loss.py b/deepspeech/modules/loss.py
index 95ca644ad5f4ae4ced28b7b4cc499a250fcf63bc..d78f689d02d26c0e1a7167f9f63613db3fb4f948 100644
--- a/deepspeech/modules/loss.py
+++ b/deepspeech/modules/loss.py
@@ -12,13 +12,13 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import logging
+from deepspeech.utils.log import Log
 
 import paddle
 from paddle import nn
 from paddle.nn import functional as F
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = ['CTCLoss', "LabelSmoothingLoss"]
 
diff --git a/deepspeech/modules/mask.py b/deepspeech/modules/mask.py
index beb67db7db9c88d3ad2a61da4542006d289fcf7b..2877b063ea74697f5865eb0477b114759a0c297d 100644
--- a/deepspeech/modules/mask.py
+++ b/deepspeech/modules/mask.py
@@ -12,11 +12,11 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import logging
+from deepspeech.utils.log import Log
 
 import paddle
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = [
     'sequence_mask', "make_pad_mask", "make_non_pad_mask", "subsequent_mask",
diff --git a/deepspeech/modules/positionwise_feed_forward.py b/deepspeech/modules/positionwise_feed_forward.py
index 4c4c3a66a74e1780996449393900e90b60078ef2..3242dd6f1b9af5fd41856636d1bbc98eaa418abe 100644
--- a/deepspeech/modules/positionwise_feed_forward.py
+++ b/deepspeech/modules/positionwise_feed_forward.py
@@ -12,12 +12,12 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 """Positionwise feed forward layer definition."""
-import logging
+from deepspeech.utils.log import Log
 
 import paddle
 from paddle import nn
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = ["PositionwiseFeedForward"]
 
diff --git a/deepspeech/modules/rnn.py b/deepspeech/modules/rnn.py
index 22dd13a44bdeacca6b6930fbd47058c5c68d535b..6d26fb4f886275feb42ade580b434c260e488a1b 100644
--- a/deepspeech/modules/rnn.py
+++ b/deepspeech/modules/rnn.py
@@ -13,7 +13,7 @@
 # limitations under the License.
 
 import math
-import logging
+from deepspeech.utils.log import Log
 
 import paddle
 from paddle import nn
@@ -23,7 +23,7 @@ from paddle.nn import initializer as I
 from deepspeech.modules.mask import sequence_mask
 from deepspeech.modules.activation import brelu
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = ['RNNStack']
 
diff --git a/deepspeech/modules/subsampling.py b/deepspeech/modules/subsampling.py
index b337089620285bcd354fda1df8ed73ed5cb558dc..8ffb20cfbedfc67047ab755c65966a382e899aef 100644
--- a/deepspeech/modules/subsampling.py
+++ b/deepspeech/modules/subsampling.py
@@ -14,14 +14,14 @@
 """Subsampling layer definition."""
 
 from typing import Tuple
-import logging
+from deepspeech.utils.log import Log
 
 import paddle
 from paddle import nn
 
 from deepspeech.modules.embedding import PositionalEncoding
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = [
     "LinearNoSubsampling", "Conv2dSubsampling4", "Conv2dSubsampling6",
diff --git a/deepspeech/training/gradclip.py b/deepspeech/training/gradclip.py
index b52bb981df8e48562714ad3a3f9f4e54ca051d1a..8f62383d4121c6f2ab8f9aaad21e1d5b252615ca 100644
--- a/deepspeech/training/gradclip.py
+++ b/deepspeech/training/gradclip.py
@@ -12,17 +12,17 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import logging
+from deepspeech.utils.log import Log
 
 import paddle
 from paddle.fluid.dygraph import base as imperative_base
 from paddle.fluid import layers
 from paddle.fluid import core
 
-logger = logging.getLogger(__name__)
-
 __all__ = ["ClipGradByGlobalNormWithLog"]
 
+logger = Log(__name__).getlog()
+
 
 class ClipGradByGlobalNormWithLog(paddle.nn.ClipGradByGlobalNorm):
     def __init__(self, clip_norm):
diff --git a/deepspeech/training/scheduler.py b/deepspeech/training/scheduler.py
index 5144ca5aef469c9df19f9dd7e8b7653be927a3e7..b2400a73a04a812579f42a38e14cbb20cb384fdf 100644
--- a/deepspeech/training/scheduler.py
+++ b/deepspeech/training/scheduler.py
@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import logging
+from deepspeech.utils.log import Log
 from typing import Union
 from typeguard import check_argument_types
 
@@ -20,7 +20,7 @@ from paddle.optimizer.lr import LRScheduler
 
 __all__ = ["WarmupLR"]
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 
 class WarmupLR(LRScheduler):
diff --git a/deepspeech/training/trainer.py b/deepspeech/training/trainer.py
index 7a9748cf6bfb9fa9258f50805f36f78a812f34d1..d172d02dda2932e486a962211646946e7df537b0 100644
--- a/deepspeech/training/trainer.py
+++ b/deepspeech/training/trainer.py
@@ -13,7 +13,7 @@
 # limitations under the License.
 
 import time
-import logging
+from deepspeech.utils.log import Log
 from pathlib import Path
 
 import paddle
@@ -25,7 +25,7 @@ from deepspeech.utils import mp_tools
 
 __all__ = ["Trainer"]
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 
 class Trainer():
diff --git a/deepspeech/utils/checkpoint.py b/deepspeech/utils/checkpoint.py
index c674b11179905c3ff5244e2d343a0c7d5293d4bc..1174d8559fb7da04581a7e4b84d0ea9acfbf6f1d 100644
--- a/deepspeech/utils/checkpoint.py
+++ b/deepspeech/utils/checkpoint.py
@@ -13,7 +13,7 @@
 # limitations under the License.
 
 import os
-import logging
+from deepspeech.utils.log import Log
 import re
 import json
 from typing import Union
@@ -24,7 +24,7 @@ from paddle.optimizer import Optimizer
 
 from deepspeech.utils import mp_tools
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = ["load_parameters", "save_parameters"]
 
diff --git a/deepspeech/utils/ctc_utils.py b/deepspeech/utils/ctc_utils.py
index 9517b8d647c69a8f730a8c80bde266e75ff1c4f7..fe646450259034ef8ea2703f5bfec24f5d88f65f 100644
--- a/deepspeech/utils/ctc_utils.py
+++ b/deepspeech/utils/ctc_utils.py
@@ -12,13 +12,13 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import logging
+from deepspeech.utils.log import Log
 import numpy as np
 from typing import List
 
 import paddle
 
-logger = logging.getLogger(__name__)
+logger = Log(__name__).getlog()
 
 __all__ = ["forced_align", "remove_duplicates_and_blank", "insert_blank"]
 
diff --git a/deepspeech/utils/log.py b/deepspeech/utils/log.py
new file mode 100644
index 0000000000000000000000000000000000000000..b5168342dd32f52cb2ea9f82b7169a9b34d98a7a
--- /dev/null
+++ b/deepspeech/utils/log.py
@@ -0,0 +1,52 @@
+# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import logging
+import os
+
+
+class Log():
+    def __init__(self, logger=None, log_cate='global'):
+        self.logger = logging.getLogger(logger)
+        self.logger.setLevel(logging.DEBUG)
+
+        file_dir = os.getcwd() + '/log'
+        if not os.path.exists(file_dir):
+            os.mkdir(file_dir)
+        self.log_path = file_dir
+        self.log_name = self.log_path + "/" + log_cate + '.log'
+
+        fh = logging.FileHandler(self.log_name)
+        fh.setLevel(logging.DEBUG)
+
+        ch = logging.StreamHandler()
+        ch.setLevel(logging.INFO)
+
+        format = '[%(levelname)s %(asctime)s %(filename)s:%(lineno)d] %(message)s'
+        formatter = logging.Formatter(fmt=format, datefmt='%Y/%m/%d %H:%M:%S')
+        fh.setFormatter(formatter)
+        ch.setFormatter(formatter)
+
+        self.logger.addHandler(fh)
+        self.logger.addHandler(ch)
+
+        # stop propagate for propagating may print
+        # log multiple times
+        # self.logger.propagate = False
+
+        fh.close()
+        ch.close()
+
+    def getlog(self):
+        return self.logger
diff --git a/deepspeech/utils/tensor_utils.py b/deepspeech/utils/tensor_utils.py
index 66d40e673f641e70f5e7eb184a4aaef0d0771ece..0a78125f83d4dbfe688a860e85b9551ab123c427 100644
--- a/deepspeech/utils/tensor_utils.py
+++ b/deepspeech/utils/tensor_utils.py
@@ -12,15 +12,15 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 """Unility functions for Transformer."""
-import logging
+from deepspeech.utils.log import Log
 from typing import Tuple, List
 
 import paddle
 
-logger = logging.getLogger(__name__)
-
 __all__ = ["pad_sequence", "add_sos_eos", "th_accuracy"]
 
+logger = Log(__name__).getlog()
+
 
 def pad_sequence(sequences: List[paddle.Tensor],
                  batch_first: bool=False,