Merge branch 'develop' of https://github.com/PaddlePaddle/DeepSpeech into unit_test

demos/speech_server/conf/application.yaml

@@ -9,9 +9,17 @@ port: 8090
 ##################################################################
 #                     CONFIG FILE                                #
 ##################################################################
-# add engine type (Options: asr, tts) and config file here.
+# The engine_type of speech task needs to keep the same type as the config file of speech task.
+# E.g: The engine_type of asr is 'python', the engine_backend of asr is 'XX/asr.yaml'
+# E.g: The engine_type of asr is 'inference', the engine_backend of asr is 'XX/asr_pd.yaml'
+#
+# add engine type (Options: python, inference) 
+engine_type:
+    asr: 'inference'
+    tts: 'inference'
 
+# add engine backend type (Options: asr, tts) and config file here.
+# Adding a speech task to engine_backend means starting the service.
 engine_backend:
-    asr: 'conf/asr/asr.yaml'
-    tts: 'conf/tts/tts.yaml'
-
+    asr: 'conf/asr/asr_pd.yaml'
+    tts: 'conf/tts/tts_pd.yaml'

demos/speech_server/conf/asr/asr.yaml

@@ -1,7 +1,8 @@
 model: 'conformer_wenetspeech'
 lang: 'zh'
 sample_rate: 16000
-cfg_path: 
-ckpt_path: 
+cfg_path: # [optional]
+ckpt_path: # [optional]
 decode_method: 'attention_rescoring'
-force_yes: False
+force_yes: True
+device: 'cpu'  # set 'gpu:id' or 'cpu'

demos/speech_server/conf/asr/asr_pd.yaml

@@ -0,0 +1,25 @@
+# This is the parameter configuration file for ASR server.
+# These are the static models that support paddle inference.
+
+##################################################################
+#                  ACOUSTIC MODEL SETTING                        #
+# am choices=['deepspeech2offline_aishell'] TODO
+##################################################################
+model_type: 'deepspeech2offline_aishell'
+am_model: # the pdmodel file of am static model [optional]
+am_params:  # the pdiparams file of am static model [optional]
+lang: 'zh'
+sample_rate: 16000
+cfg_path: 
+decode_method: 
+force_yes: True
+
+am_predictor_conf:
+  device: 'cpu'  # set 'gpu:id' or 'cpu'
+  enable_mkldnn: True
+  switch_ir_optim: True
+
+
+##################################################################
+#                            OTHERS                              #
+##################################################################

demos/speech_server/conf/tts/tts.yaml

@@ -29,4 +29,4 @@ voc_stat:
 #                            OTHERS                              #
 ##################################################################
 lang: 'zh'
-device: 'gpu:2'
+device: 'cpu'  # set 'gpu:id' or 'cpu'

demos/speech_server/conf/tts/tts_pd.yaml

@@ -6,8 +6,8 @@
 # am choices=['speedyspeech_csmsc', 'fastspeech2_csmsc']
 ##################################################################
 am: 'fastspeech2_csmsc'   
-am_model: # the pdmodel file of am static model
-am_params: # the pdiparams file of am static model
+am_model: # the pdmodel file of your am static model (XX.pdmodel)
+am_params: # the pdiparams file of your am static model (XX.pdipparams)
 am_sample_rate: 24000
 phones_dict: 
 tones_dict: 
@@ -15,9 +15,9 @@ speaker_dict:
 spk_id: 0
 
 am_predictor_conf:
-  use_gpu: True
-  enable_mkldnn: True
-  switch_ir_optim: True
+  device: 'cpu'  # set 'gpu:id' or 'cpu'
+  enable_mkldnn: False
+  switch_ir_optim: False
 
 
 ##################################################################
@@ -25,17 +25,16 @@ am_predictor_conf:
 # voc choices=['pwgan_csmsc', 'mb_melgan_csmsc','hifigan_csmsc']
 ##################################################################
 voc: 'pwgan_csmsc'
-voc_model: # the pdmodel file of vocoder static model
-voc_params: # the pdiparams file of vocoder static model 
+voc_model: # the pdmodel file of your vocoder static model (XX.pdmodel)
+voc_params: # the pdiparams file of your vocoder static model (XX.pdipparams)
 voc_sample_rate: 24000
 
 voc_predictor_conf:
-  use_gpu: True
-  enable_mkldnn: True  
-  switch_ir_optim: True  
+  device: 'cpu'  # set 'gpu:id' or 'cpu'
+  enable_mkldnn: False  
+  switch_ir_optim: False  
 
 ##################################################################
 #                            OTHERS                              #
 ##################################################################
 lang: 'zh'
-device: paddle.get_device()

paddlespeech/server/bin/main.py

@@ -12,8 +12,8 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 import argparse
+
 import uvicorn
-import yaml
 from fastapi import FastAPI
 
 from paddlespeech.server.engine.engine_pool import init_engine_pool

paddlespeech/server/bin/paddlespeech_client.py

@@ -48,8 +48,9 @@ class TTSClientExecutor(BaseExecutor):
         self.parser.add_argument(
             '--input',
             type=str,
-            default="你好，欢迎使用语音合成服务",
-            help='A sentence to be synthesized.')
+            default=None,
+            help='Text to be synthesized.',
+            required=True)
         self.parser.add_argument(
             '--spk_id', type=int, default=0, help='Speaker id')
         self.parser.add_argument(
@@ -123,7 +124,7 @@ class TTSClientExecutor(BaseExecutor):
             logger.info("RTF: %f " % (time_consume / duration))
 
             return True
-        except:
+        except BaseException:
             logger.error("Failed to synthesized audio.")
             return False
 
@@ -163,7 +164,7 @@ class TTSClientExecutor(BaseExecutor):
             print("Audio duration: %f s." % (duration))
             print("Response time: %f s." % (time_consume))
             print("RTF: %f " % (time_consume / duration))
-        except:
+        except BaseException:
             print("Failed to synthesized audio.")
 
 
@@ -181,8 +182,9 @@ class ASRClientExecutor(BaseExecutor):
         self.parser.add_argument(
             '--input',
             type=str,
-            default="./paddlespeech/server/tests/16_audio.wav",
-            help='Audio file to be recognized')
+            default=None,
+            help='Audio file to be recognized',
+            required=True)
         self.parser.add_argument(
             '--sample_rate', type=int, default=16000, help='audio sample rate')
         self.parser.add_argument(
@@ -209,7 +211,7 @@ class ASRClientExecutor(BaseExecutor):
             logger.info(r.json())
             logger.info("time cost %f s." % (time_end - time_start))
             return True
-        except:
+        except BaseException:
             logger.error("Failed to speech recognition.")
             return False
 
@@ -240,5 +242,5 @@ class ASRClientExecutor(BaseExecutor):
             time_end = time.time()
             print(r.json())
             print("time cost %f s." % (time_end - time_start))
-        except:
-            print("Failed to speech recognition.")
+        except BaseException:
+            print("Failed to speech recognition.")

paddlespeech/server/bin/paddlespeech_server.py

@@ -20,7 +20,7 @@ from fastapi import FastAPI
 from ..executor import BaseExecutor
 from ..util import cli_server_register
 from ..util import stats_wrapper
-from paddlespeech.server.engine.engine_factory import EngineFactory
+from paddlespeech.server.engine.engine_pool import init_engine_pool
 from paddlespeech.server.restful.api import setup_router
 from paddlespeech.server.utils.config import get_config
 
@@ -41,7 +41,8 @@ class ServerExecutor(BaseExecutor):
             "--config_file",
             action="store",
             help="yaml file of the app",
-            default="./conf/application.yaml")
+            default=None,
+            required=True)
 
         self.parser.add_argument(
             "--log_file",
@@ -51,8 +52,10 @@ class ServerExecutor(BaseExecutor):
 
     def init(self, config) -> bool:
         """system initialization
+
         Args:
             config (CfgNode): config object
+
         Returns:
             bool: 
         """
@@ -61,13 +64,8 @@ class ServerExecutor(BaseExecutor):
         api_router = setup_router(api_list)
         app.include_router(api_router)
 
-        # init engine
-        engine_pool = []
-        for engine in config.engine_backend:
-            engine_pool.append(EngineFactory.get_engine(engine_name=engine))
-            if not engine_pool[-1].init(
-                    config_file=config.engine_backend[engine]):
-                return False
+        if not init_engine_pool(config):
+            return False
 
         return True
 

paddlespeech/server/conf/application.yaml

@@ -9,12 +9,17 @@ port: 8090
 ##################################################################
 #                     CONFIG FILE                                #
 ##################################################################
+# The engine_type of speech task needs to keep the same type as the config file of speech task.
+# E.g: The engine_type of asr is 'python', the engine_backend of asr is 'XX/asr.yaml'
+# E.g: The engine_type of asr is 'inference', the engine_backend of asr is 'XX/asr_pd.yaml'
+#
 # add engine type (Options: python, inference) 
 engine_type:
-    asr: 'inference'
-    # tts: 'inference'
+    asr: 'python'
+    tts: 'python'
 
 # add engine backend type (Options: asr, tts) and config file here.
+# Adding a speech task to engine_backend means starting the service.
 engine_backend:
-    asr: 'conf/asr/asr_pd.yaml'
-    #tts: 'conf/tts/tts_pd.yaml'
+    asr: 'conf/asr/asr.yaml'
+    tts: 'conf/tts/tts.yaml'

paddlespeech/server/conf/asr/asr.yaml

@@ -5,3 +5,4 @@ cfg_path: # [optional]
 ckpt_path: # [optional]
 decode_method: 'attention_rescoring'
 force_yes: True
+device: 'cpu'  # set 'gpu:id' or 'cpu'

paddlespeech/server/conf/asr/asr_pd.yaml

@@ -15,7 +15,7 @@ decode_method:
 force_yes: True
 
 am_predictor_conf:
-  use_gpu: True
+  device: 'cpu'  # set 'gpu:id' or 'cpu'
   enable_mkldnn: True
   switch_ir_optim: True
 

paddlespeech/server/conf/tts/tts.yaml

@@ -29,4 +29,4 @@ voc_stat:
 #                            OTHERS                              #
 ##################################################################
 lang: 'zh'
-device: paddle.get_device()
+device: 'cpu'  # set 'gpu:id' or 'cpu'

paddlespeech/server/conf/tts/tts_pd.yaml

@@ -6,18 +6,18 @@
 # am choices=['speedyspeech_csmsc', 'fastspeech2_csmsc']
 ##################################################################
 am: 'fastspeech2_csmsc'   
-am_model: # the pdmodel file of am static model
-am_params: # the pdiparams file of am static model
-am_sample_rate: 24000
+am_model: # the pdmodel file of your am static model (XX.pdmodel)
+am_params: # the pdiparams file of your am static model (XX.pdipparams)
+am_sample_rate: 24000    # must match the model
 phones_dict: 
 tones_dict: 
 speaker_dict: 
 spk_id: 0
 
 am_predictor_conf:
-  use_gpu: True
-  enable_mkldnn: True
-  switch_ir_optim: True
+  device: 'cpu'  # set 'gpu:id' or 'cpu'
+  enable_mkldnn: False
+  switch_ir_optim: False
 
 
 ##################################################################
@@ -25,17 +25,16 @@ am_predictor_conf:
 # voc choices=['pwgan_csmsc', 'mb_melgan_csmsc','hifigan_csmsc']
 ##################################################################
 voc: 'pwgan_csmsc'
-voc_model: # the pdmodel file of vocoder static model
-voc_params: # the pdiparams file of vocoder static model 
-voc_sample_rate: 24000
+voc_model: # the pdmodel file of your vocoder static model (XX.pdmodel)
+voc_params: # the pdiparams file of your vocoder static model (XX.pdipparams)
+voc_sample_rate: 24000    #must match the model
 
 voc_predictor_conf:
-  use_gpu: True
-  enable_mkldnn: True  
-  switch_ir_optim: True  
+  device: 'cpu'   # set 'gpu:id' or 'cpu'
+  enable_mkldnn: False
+  switch_ir_optim: False  
 
 ##################################################################
 #                            OTHERS                              #
 ##################################################################
 lang: 'zh'
-device: paddle.get_device()

paddlespeech/server/engine/asr/paddleinference/asr_engine.py

@@ -13,31 +13,24 @@
 # limitations under the License.
 import io
 import os
-from typing import List
 from typing import Optional
-from typing import Union
 
-import librosa
 import paddle
-import soundfile
 from yacs.config import CfgNode
 
-from paddlespeech.cli.utils import MODEL_HOME
-from paddlespeech.s2t.modules.ctc import CTCDecoder
 from paddlespeech.cli.asr.infer import ASRExecutor
 from paddlespeech.cli.log import logger
+from paddlespeech.cli.utils import MODEL_HOME
 from paddlespeech.s2t.frontend.featurizer.text_featurizer import TextFeaturizer
-from paddlespeech.s2t.transform.transformation import Transformation
-from paddlespeech.s2t.utils.dynamic_import import dynamic_import
+from paddlespeech.s2t.modules.ctc import CTCDecoder
 from paddlespeech.s2t.utils.utility import UpdateConfig
+from paddlespeech.server.engine.base_engine import BaseEngine
 from paddlespeech.server.utils.config import get_config
 from paddlespeech.server.utils.paddle_predictor import init_predictor
 from paddlespeech.server.utils.paddle_predictor import run_model
-from paddlespeech.server.engine.base_engine import BaseEngine
 
 __all__ = ['ASREngine']
 
-
 pretrained_models = {
     "deepspeech2offline_aishell-zh-16k": {
         'url':
@@ -143,7 +136,6 @@ class ASRServerExecutor(ASRExecutor):
             batch_average=True,  # sum / batch_size
             grad_norm_type=self.config.get('ctc_grad_norm_type', None))
 
-
     @paddle.no_grad()
     def infer(self, model_type: str):
         """
@@ -161,9 +153,8 @@ class ASRServerExecutor(ASRExecutor):
                 cfg.beam_size, cfg.cutoff_prob, cfg.cutoff_top_n,
                 cfg.num_proc_bsearch)
 
-            output_data = run_model(
-                                self.am_predictor,
-                                [audio.numpy(), audio_len.numpy()])
+            output_data = run_model(self.am_predictor,
+                                    [audio.numpy(), audio_len.numpy()])
 
             probs = output_data[0]
             eouts_len = output_data[1]
@@ -208,14 +199,14 @@ class ASREngine(BaseEngine):
 
         paddle.set_device(paddle.get_device())
         self.executor._init_from_path(
-                    model_type=self.config.model_type,
-                    am_model=self.config.am_model,
-                    am_params=self.config.am_params,
-                    lang=self.config.lang,
-                    sample_rate=self.config.sample_rate,
-                    cfg_path=self.config.cfg_path,
-                    decode_method=self.config.decode_method,
-                    am_predictor_conf=self.config.am_predictor_conf)
+            model_type=self.config.model_type,
+            am_model=self.config.am_model,
+            am_params=self.config.am_params,
+            lang=self.config.lang,
+            sample_rate=self.config.sample_rate,
+            cfg_path=self.config.cfg_path,
+            decode_method=self.config.decode_method,
+            am_predictor_conf=self.config.am_predictor_conf)
 
         logger.info("Initialize ASR server engine successfully.")
         return True
@@ -230,7 +221,8 @@ class ASREngine(BaseEngine):
                 io.BytesIO(audio_data), self.config.sample_rate,
                 self.config.force_yes):
             logger.info("start running asr engine")
-            self.executor.preprocess(self.config.model_type, io.BytesIO(audio_data))
+            self.executor.preprocess(self.config.model_type,
+                                     io.BytesIO(audio_data))
             self.executor.infer(self.config.model_type)
             self.output = self.executor.postprocess()  # Retrieve result of asr.
             logger.info("end inferring asr engine")

paddlespeech/server/engine/asr/python/asr_engine.py

@@ -12,21 +12,11 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 import io
-import os
-from typing import List
-from typing import Optional
-from typing import Union
 
-import librosa
 import paddle
-import soundfile
 
 from paddlespeech.cli.asr.infer import ASRExecutor
 from paddlespeech.cli.log import logger
-from paddlespeech.s2t.frontend.featurizer.text_featurizer import TextFeaturizer
-from paddlespeech.s2t.transform.transformation import Transformation
-from paddlespeech.s2t.utils.dynamic_import import dynamic_import
-from paddlespeech.s2t.utils.utility import UpdateConfig
 from paddlespeech.server.engine.base_engine import BaseEngine
 from paddlespeech.server.utils.config import get_config
 
@@ -63,7 +53,10 @@ class ASREngine(BaseEngine):
         self.executor = ASRServerExecutor()
 
         self.config = get_config(config_file)
-        paddle.set_device(paddle.get_device())
+        if self.config.device is None:
+            paddle.set_device(paddle.get_device())
+        else:
+            paddle.set_device(self.config.device)
         self.executor._init_from_path(
             self.config.model, self.config.lang, self.config.sample_rate,
             self.config.cfg_path, self.config.decode_method,

paddlespeech/server/engine/base_engine.py

@@ -12,8 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 import os
-from typing import Any
-from typing import List
 from typing import Union
 
 from pattern_singleton import Singleton

paddlespeech/server/engine/engine_factory.py

@@ -13,7 +13,6 @@
 # limitations under the License.
 from typing import Text
 
-
 __all__ = ['EngineFactory']
 
 

paddlespeech/server/engine/engine_pool.py

@@ -29,8 +29,10 @@ def init_engine_pool(config) -> bool:
     """
     global ENGINE_POOL
     for engine in config.engine_backend:
-        ENGINE_POOL[engine] = EngineFactory.get_engine(engine_name=engine, engine_type=config.engine_type[engine])
-        if not ENGINE_POOL[engine].init(config_file=config.engine_backend[engine]):
+        ENGINE_POOL[engine] = EngineFactory.get_engine(
+            engine_name=engine, engine_type=config.engine_type[engine])
+        if not ENGINE_POOL[engine].init(
+                config_file=config.engine_backend[engine]):
             return False
 
     return True

paddlespeech/server/engine/tts/paddleinference/tts_engine.py

@@ -344,7 +344,6 @@ class TTSEngine(BaseEngine):
 
         try:
             self.config = get_config(config_file)
-
             self.executor._init_from_path(
                 am=self.config.am,
                 am_model=self.config.am_model,
@@ -361,8 +360,8 @@ class TTSEngine(BaseEngine):
                 am_predictor_conf=self.config.am_predictor_conf,
                 voc_predictor_conf=self.config.voc_predictor_conf, )
 
-        except:
-            logger.info("Initialize TTS server engine Failed.")
+        except BaseException:
+            logger.error("Initialize TTS server engine Failed.")
             return False
 
         logger.info("Initialize TTS server engine successfully.")
@@ -406,11 +405,13 @@ class TTSEngine(BaseEngine):
         # transform speed
         try:  # windows not support soxbindings
             wav_speed = change_speed(wav_vol, speed, target_fs)
-        except:
+        except ServerBaseException:
             raise ServerBaseException(
                 ErrorCode.SERVER_INTERNAL_ERR,
                 "Transform speed failed. Can not install soxbindings on your system. \
                  You need to set speed value 1.0.")
+        except BaseException:
+            logger.error("Transform speed failed.")
 
         # wav to base64
         buf = io.BytesIO()
@@ -463,9 +464,11 @@ class TTSEngine(BaseEngine):
         try:
             self.executor.infer(
                 text=sentence, lang=lang, am=self.config.am, spk_id=spk_id)
-        except:
+        except ServerBaseException:
             raise ServerBaseException(ErrorCode.SERVER_INTERNAL_ERR,
                                       "tts infer failed.")
+        except BaseException:
+            logger.error("tts infer failed.")
 
         try:
             target_sample_rate, wav_base64 = self.postprocess(
@@ -475,8 +478,10 @@ class TTSEngine(BaseEngine):
                 volume=volume,
                 speed=speed,
                 audio_path=save_path)
-        except:
+        except ServerBaseException:
             raise ServerBaseException(ErrorCode.SERVER_INTERNAL_ERR,
                                       "tts postprocess failed.")
+        except BaseException:
+            logger.error("tts postprocess failed.")
 
         return lang, target_sample_rate, wav_base64

paddlespeech/server/engine/tts/python/tts_engine.py

@@ -54,7 +54,10 @@ class TTSEngine(BaseEngine):
 
         try:
             self.config = get_config(config_file)
-            paddle.set_device(self.config.device)
+            if self.config.device is None:
+                paddle.set_device(paddle.get_device())
+            else:
+                paddle.set_device(self.config.device)
 
             self.executor._init_from_path(
                 am=self.config.am,
@@ -69,8 +72,8 @@ class TTSEngine(BaseEngine):
                 voc_ckpt=self.config.voc_ckpt,
                 voc_stat=self.config.voc_stat,
                 lang=self.config.lang)
-        except:
-            logger.info("Initialize TTS server engine Failed.")
+        except BaseException:
+            logger.error("Initialize TTS server engine Failed.")
             return False
 
         logger.info("Initialize TTS server engine successfully.")
@@ -114,10 +117,13 @@ class TTSEngine(BaseEngine):
         # transform speed
         try:  # windows not support soxbindings
             wav_speed = change_speed(wav_vol, speed, target_fs)
-        except:
+        except ServerBaseException:
             raise ServerBaseException(
                 ErrorCode.SERVER_INTERNAL_ERR,
-                "Can not install soxbindings on your system.")
+                "Transform speed failed. Can not install soxbindings on your system. \
+                 You need to set speed value 1.0.")
+        except BaseException:
+            logger.error("Transform speed failed.")
 
         # wav to base64
         buf = io.BytesIO()
@@ -170,9 +176,11 @@ class TTSEngine(BaseEngine):
         try:
             self.executor.infer(
                 text=sentence, lang=lang, am=self.config.am, spk_id=spk_id)
-        except:
+        except ServerBaseException:
             raise ServerBaseException(ErrorCode.SERVER_INTERNAL_ERR,
                                       "tts infer failed.")
+        except BaseException:
+            logger.error("tts infer failed.")
 
         try:
             target_sample_rate, wav_base64 = self.postprocess(
@@ -182,8 +190,10 @@ class TTSEngine(BaseEngine):
                 volume=volume,
                 speed=speed,
                 audio_path=save_path)
-        except:
+        except ServerBaseException:
             raise ServerBaseException(ErrorCode.SERVER_INTERNAL_ERR,
                                       "tts postprocess failed.")
+        except BaseException:
+            logger.error("tts postprocess failed.")
 
         return lang, target_sample_rate, wav_base64

paddlespeech/server/restful/asr_api.py

@@ -14,6 +14,7 @@
 import base64
 import traceback
 from typing import Union
+
 from fastapi import APIRouter
 
 from paddlespeech.server.engine.engine_pool import get_engine_pool
@@ -83,7 +84,7 @@ def asr(request_body: ASRRequest):
 
     except ServerBaseException as e:
         response = failed_response(e.error_code, e.msg)
-    except:
+    except BaseException:
         response = failed_response(ErrorCode.SERVER_UNKOWN_ERR)
         traceback.print_exc()
 

paddlespeech/server/restful/request.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.
-from typing import List
 from typing import Optional
 
 from pydantic import BaseModel

paddlespeech/server/restful/response.py

@@ -11,9 +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.
-from typing import List
-from typing import Optional
-
 from pydantic import BaseModel
 
 __all__ = ['ASRResponse', 'TTSResponse']

paddlespeech/server/restful/tts_api.py

@@ -16,7 +16,7 @@ from typing import Union
 
 from fastapi import APIRouter
 
-from paddlespeech.server.engine.tts.paddleinference.tts_engine import TTSEngine
+from paddlespeech.server.engine.engine_pool import get_engine_pool
 from paddlespeech.server.restful.request import TTSRequest
 from paddlespeech.server.restful.response import ErrorResponse
 from paddlespeech.server.restful.response import TTSResponse
@@ -60,28 +60,41 @@ def tts(request_body: TTSRequest):
     Returns:
         json: [description]
     """
-    # json to dict 
-    item_dict = request_body.dict()
-    sentence = item_dict['text']
-    spk_id = item_dict['spk_id']
-    speed = item_dict['speed']
-    volume = item_dict['volume']
-    sample_rate = item_dict['sample_rate']
-    save_path = item_dict['save_path']
+    # get params
+    text = request_body.text
+    spk_id = request_body.spk_id
+    speed = request_body.speed
+    volume = request_body.volume
+    sample_rate = request_body.sample_rate
+    save_path = request_body.save_path
 
     # Check parameters
-    if speed <=0 or speed > 3 or volume <=0 or volume > 3 or \
-        sample_rate not in [0, 16000, 8000] or \
-        (save_path is not None and not save_path.endswith("pcm") and not save_path.endswith("wav")):
-        return failed_response(ErrorCode.SERVER_PARAM_ERR)
-
-    # single
-    tts_engine = TTSEngine()
+    if speed <= 0 or speed > 3:
+        return failed_response(
+            ErrorCode.SERVER_PARAM_ERR,
+            "invalid speed value, the value should be between 0 and 3.")
+    if volume <= 0 or volume > 3:
+        return failed_response(
+            ErrorCode.SERVER_PARAM_ERR,
+            "invalid volume value, the value should be between 0 and 3.")
+    if sample_rate not in [0, 16000, 8000]:
+        return failed_response(
+            ErrorCode.SERVER_PARAM_ERR,
+            "invalid sample_rate value, the choice of value is 0, 8000, 16000.")
+    if save_path is not None and not save_path.endswith(
+            "pcm") and not save_path.endswith("wav"):
+        return failed_response(
+            ErrorCode.SERVER_PARAM_ERR,
+            "invalid save_path, saved audio formats support pcm and wav")
 
     # run
     try:
+        # get single engine from engine pool
+        engine_pool = get_engine_pool()
+        tts_engine = engine_pool['tts']
+
         lang, target_sample_rate, wav_base64 = tts_engine.run(
-            sentence, spk_id, speed, volume, sample_rate, save_path)
+            text, spk_id, speed, volume, sample_rate, save_path)
 
         response = {
             "success": True,
@@ -101,7 +114,7 @@ def tts(request_body: TTSRequest):
         }
     except ServerBaseException as e:
         response = failed_response(e.error_code, e.msg)
-    except:
+    except BaseException:
         response = failed_response(ErrorCode.SERVER_UNKOWN_ERR)
         traceback.print_exc()
 

paddlespeech/server/tests/asr/http_client.py

@@ -10,11 +10,11 @@
 # 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 
-import requests
+import base64
 import json
 import time
-import base64
-import io
+
+import requests
 
 
 def readwav2base64(wav_file):
@@ -34,23 +34,23 @@ def main():
     url = "http://127.0.0.1:8090/paddlespeech/asr"
 
     # start Timestamp
-    time_start=time.time()
+    time_start = time.time()
 
     test_audio_dir = "./16_audio.wav"
     audio = readwav2base64(test_audio_dir)
 
     data = {
-            "audio": audio,
-            "audio_format": "wav",
-            "sample_rate": 16000,
-            "lang": "zh_cn",
-            }
+        "audio": audio,
+        "audio_format": "wav",
+        "sample_rate": 16000,
+        "lang": "zh_cn",
+    }
 
     r = requests.post(url=url, data=json.dumps(data))
 
     # ending Timestamp
-    time_end=time.time()
-    print('time cost',time_end - time_start, 's')
+    time_end = time.time()
+    print('time cost', time_end - time_start, 's')
 
     print(r.json())
 

paddlespeech/server/tests/tts/test_client.py

@@ -25,6 +25,7 @@ import soundfile
 
 from paddlespeech.server.utils.audio_process import wav2pcm
 
+
 # Request and response
 def tts_client(args):
     """ Request and response
@@ -99,5 +100,5 @@ if __name__ == "__main__":
         print("Inference time: %f" % (time_consume))
         print("The duration of synthesized audio: %f" % (duration))
         print("The RTF is: %f" % (rtf))
-    except:
+    except BaseException:
         print("Failed to synthesized audio.")

paddlespeech/server/util.py

@@ -219,7 +219,7 @@ class ConfigCache:
             try:
                 cfg = yaml.load(file, Loader=yaml.FullLoader)
                 self._data.update(cfg)
-            except:
+            except BaseException:
                 self.flush()
 
     @property

paddlespeech/server/utils/paddle_predictor.py

@@ -41,8 +41,9 @@ def init_predictor(model_dir: Optional[os.PathLike]=None,
         config = Config(model_file, params_file)
 
     config.enable_memory_optim()
-    if predictor_conf["use_gpu"]:
-        config.enable_use_gpu(1000, 0)
+    if "gpu" in predictor_conf["device"]:
+        gpu_id = predictor_conf["device"].split(":")[-1]
+        config.enable_use_gpu(1000, int(gpu_id))
     if predictor_conf["enable_mkldnn"]:
         config.enable_mkldnn()
     if predictor_conf["switch_ir_optim"]:

requirements.txt

@@ -1,48 +0,0 @@
-ConfigArgParse
-coverage
-editdistance
-g2p_en
-g2pM
-gpustat
-h5py
-inflect
-jieba
-jsonlines
-kaldiio
-librosa
-loguru
-matplotlib
-nara_wpe
-nltk
-paddleaudio
-paddlenlp
-paddlespeech_ctcdecoders
-paddlespeech_feat
-pandas
-phkit
-Pillow
-praatio==5.0.0
-pre-commit
-pybind11
-pypi-kenlm
-pypinyin
-python-dateutil
-pyworld
-resampy==0.2.2
-sacrebleu
-scipy
-sentencepiece~=0.1.96
-snakeviz
-soundfile~=0.10
-sox
-soxbindings
-textgrid
-timer
-tqdm
-typeguard
-unidecode
-visualdl
-webrtcvad
-yacs~=0.1.8
-yq
-zhon

setup.py

@@ -27,46 +27,52 @@ from setuptools.command.install import install
 
 HERE = Path(os.path.abspath(os.path.dirname(__file__)))
 
-VERSION = '0.1.1'
+VERSION = '0.1.2'
+
+base = [
+    "editdistance",
+    "g2p_en",
+    "g2pM",
+    "h5py",
+    "inflect",
+    "jieba",
+    "jsonlines",
+    "kaldiio",
+    "librosa==0.8.1",
+    "loguru",
+    "matplotlib",
+    "nara_wpe",
+    "pandas",
+    "paddleaudio",
+    "paddlenlp",
+    "paddlespeech_feat",
+    "praatio==5.0.0",
+    "pypinyin",
+    "python-dateutil",
+    "pyworld",
+    "resampy==0.2.2",
+    "sacrebleu",
+    "scipy",
+    "sentencepiece~=0.1.96",
+    "soundfile~=0.10",
+    "textgrid",
+    "timer",
+    "tqdm",
+    "typeguard",
+    "visualdl",
+    "webrtcvad",
+    "yacs~=0.1.8",
+]
+
+server = [
+    "fastapi",
+    "uvicorn",
+    "pattern_singleton",
+]
 
 requirements = {
-    "install": [
-        "editdistance",
-        "g2p_en",
-        "g2pM",
-        "h5py",
-        "inflect",
-        "jieba",
-        "jsonlines",
-        "kaldiio",
-        "librosa",
-        "loguru",
-        "matplotlib",
-        "nara_wpe",
-        "pandas",
-        "paddleaudio",
-        "paddlenlp",
-        "paddlespeech_feat",
-        "praatio==5.0.0",
-        "pypinyin",
-        "python-dateutil",
-        "pyworld",
-        "resampy==0.2.2",
-        "sacrebleu",
-        "scipy",
-        "sentencepiece~=0.1.96",
-        "soundfile~=0.10",
-        "textgrid",
-        "timer",
-        "tqdm",
-        "typeguard",
-        "visualdl",
-        "webrtcvad",
-        "yacs~=0.1.8",
-        # fastapi server
-        "fastapi",
-        "uvicorn",
-    ],
+    "install":
+    base + server,
     "develop": [
         "ConfigArgParse",
         "coverage",