added engine type and asr inference , test=doc

paddlespeech/__init__.py

@@ -11,3 +11,14 @@
 # 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.
+
+
+
+
+
+
+
+
+
+
+

paddlespeech/s2t/io/utility.py

@@ -12,6 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 from typing import List
+from io import BytesIO
 
 import numpy as np
 
@@ -88,6 +89,10 @@ def pad_sequence(sequences: List[np.ndarray],
 
 
 def feat_type(filepath):
+ # deal with Byteio type for paddlespeech server
+ if isinstance(filepath, BytesIO):
+ return 'sound'
+
  suffix = filepath.split(":")[0].split('.')[-1].lower()
  if suffix == 'ark':
  return 'mat'

paddlespeech/server/__init__.py

@@ -11,3 +11,14 @@
 # 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 _locale
+
+from .base_commands import ClientBaseCommand
+from .base_commands import ClientHelpCommand
+from .base_commands import ServerBaseCommand
+from .base_commands import ServerHelpCommand
+from .bin.paddlespeech_client import ASRClientExecutor
+from .bin.paddlespeech_client import TTSClientExecutor
+from .bin.paddlespeech_server import ServerExecutor
+
+_locale._getdefaultlocale = (lambda *args: ['en_US', 'utf8'])

paddlespeech/server/base_commands.py

@@ -0,0 +1,82 @@
+# 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:https://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.
+from typing import List
+
+from .entry import client_commands
+from .entry import server_commands
+from .util import cli_client_register
+from .util import cli_server_register
+from .util import get_client_command
+from .util import get_server_command
+
+__all__ = [
+ 'ServerBaseCommand',
+ 'ServerHelpCommand',
+ 'ClientBaseCommand',
+ 'ClientHelpCommand',
+]
+
+
+@cli_server_register(name='paddlespeech_server')
+class ServerBaseCommand:
+ def execute(self, argv: List[str]) -> bool:
+ help = get_server_command('paddlespeech_server.help')
+ return help().execute(argv)
+
+
+@cli_server_register(
+ name='paddlespeech_server.help', description='Show help for commands.')
+class ServerHelpCommand:
+ def execute(self, argv: List[str]) -> bool:
+ msg = 'Usage:\n'
+ msg += ' paddlespeech_server <command> <options>\n\n'
+ msg += 'Commands:\n'
+ for command, detail in server_commands['paddlespeech_server'].items():
+ if command.startswith('_'):
+ continue
+
+ if '_description' not in detail:
+ continue
+ msg += ' {:<15} {}\n'.format(command,
+ detail['_description'])
+
+ print(msg)
+ return True
+
+
+@cli_client_register(name='paddlespeech_client')
+class ClientBaseCommand:
+ def execute(self, argv: List[str]) -> bool:
+ help = get_client_command('paddlespeech_client.help')
+ return help().execute(argv)
+
+
+@cli_client_register(
+ name='paddlespeech_client.help', description='Show help for commands.')
+class ClientHelpCommand:
+ def execute(self, argv: List[str]) -> bool:
+ msg = 'Usage:\n'
+ msg += ' paddlespeech_client <command> <options>\n\n'
+ msg += 'Commands:\n'
+ for command, detail in client_commands['paddlespeech_client'].items():
+ if command.startswith('_'):
+ continue
+
+ if '_description' not in detail:
+ continue
+ msg += ' {:<15} {}\n'.format(command,
+ detail['_description'])
+
+ print(msg)
+ return True

paddlespeech/server/bin/__init__.py

@@ -0,0 +1,16 @@
+# 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:https://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.
+from .paddlespeech_client import ASRClientExecutor
+from .paddlespeech_client import TTSClientExecutor
+from .paddlespeech_server import ServerExecutor

paddlespeech/server/bin/main.py

@@ -16,10 +16,9 @@
 import yaml
 from fastapi import FastAPI
 
-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
-from paddlespeech.server.utils.log import logger
 
 app = FastAPI(
  title="PaddleSpeech Serving API", description="Api", version="0.0.1")
@@ -39,12 +38,8 @@ def init(config):
  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/bin/paddlespeech_client.py

@@ -0,0 +1,156 @@
+# 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:https://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 argparse
+import base64
+import io
+import json
+import os
+import random
+import time
+from typing import List
+
+import numpy as np
+import requests
+import soundfile
+
+from ..util import cli_client_register
+from paddlespeech.server.utils.audio_process import wav2pcm
+from paddlespeech.server.utils.util import wav2base64
+
+__all__ = ['TTSClientExecutor', 'ASRClientExecutor']
+
+
+@cli_client_register(
+ name='paddlespeech_client.tts', description='visit tts service')
+class TTSClientExecutor():
+ def __init__(self):
+ super().__init__()
+ self.parser = argparse.ArgumentParser()
+ self.parser.add_argument(
+ '--server_ip', type=str, default='127.0.0.1', help='server ip')
+ self.parser.add_argument(
+ '--port', type=int, default=8090, help='server port')
+ self.parser.add_argument(
+ '--text',
+ type=str,
+ default="你好，欢迎使用语音合成服务",
+ help='A sentence to be synthesized')
+ self.parser.add_argument(
+ '--spk_id', type=int, default=0, help='Speaker id')
+ self.parser.add_argument(
+ '--speed', type=float, default=1.0, help='Audio speed')
+ self.parser.add_argument(
+ '--volume', type=float, default=1.0, help='Audio volume')
+ self.parser.add_argument(
+ '--sample_rate',
+ type=int,
+ default=0,
+ help='Sampling rate, the default is the same as the model')
+ self.parser.add_argument(
+ '--output',
+ type=str,
+ default="./out.wav",
+ help='Synthesized audio file')
+
+ # Request and response
+ def tts_client(self, args):
+ """ Request and response
+ Args:
+ text: A sentence to be synthesized
+ outfile: Synthetic audio file
+ """
+ url = 'http:https://' + args.server_ip + ":" + str(
+ args.port) + '/paddlespeech/tts'
+ request = {
+ "text": args.text,
+ "spk_id": args.spk_id,
+ "speed": args.speed,
+ "volume": args.volume,
+ "sample_rate": args.sample_rate,
+ "save_path": args.output
+ }
+
+ response = requests.post(url, json.dumps(request))
+ response_dict = response.json()
+ print(response_dict["message"])
+ wav_base64 = response_dict["result"]["audio"]
+
+ audio_data_byte = base64.b64decode(wav_base64)
+ # from byte
+ samples, sample_rate = soundfile.read(
+ io.BytesIO(audio_data_byte), dtype='float32')
+
+ # transform audio
+ outfile = args.output
+ if outfile.endswith(".wav"):
+ soundfile.write(outfile, samples, sample_rate)
+ elif outfile.endswith(".pcm"):
+ temp_wav = str(random.getrandbits(128)) + ".wav"
+ soundfile.write(temp_wav, samples, sample_rate)
+ wav2pcm(temp_wav, outfile, data_type=np.int16)
+ os.system("rm %s" % (temp_wav))
+ else:
+ print("The format for saving audio only supports wav or pcm")
+
+ return len(samples), sample_rate
+
+ def execute(self, argv: List[str]) -> bool:
+ args = self.parser.parse_args(argv)
+ st = time.time()
+ try:
+ samples_length, sample_rate = self.tts_client(args)
+ time_consume = time.time() - st
+ print("Save synthesized audio successfully on %s." % (args.output))
+ print("Inference time: %f s." % (time_consume))
+ except:
+ print("Failed to synthesized audio.")
+
+
+@cli_client_register(
+ name='paddlespeech_client.asr', description='visit asr service')
+class ASRClientExecutor():
+ def __init__(self):
+ super().__init__()
+ self.parser = argparse.ArgumentParser()
+ self.parser.add_argument(
+ '--server_ip', type=str, default='127.0.0.1', help='server ip')
+ self.parser.add_argument(
+ '--port', type=int, default=8090, help='server port')
+ self.parser.add_argument(
+ '--audio_file',
+ type=str,
+ default="./paddlespeech/server/tests/16_audio.wav",
+ help='Audio file to be recognized')
+ self.parser.add_argument(
+ '--sample_rate', type=int, default=16000, help='audio sample rate')
+
+ def execute(self, argv: List[str]) -> bool:
+ args = self.parser.parse_args(argv)
+ url = 'http:https://' + args.server_ip + ":" + str(
+ args.port) + '/paddlespeech/asr'
+ audio = wav2base64(args.audio_file)
+ data = {
+ "audio": audio,
+ "audio_format": "wav",
+ "sample_rate": args.sample_rate,
+ "lang": "zh_cn",
+ }
+ time_start = time.time()
+ try:
+ r = requests.post(url=url, data=json.dumps(data))
+ # ending Timestamp
+ time_end = time.time()
+ print('time cost', time_end - time_start, 's')
+ except:
+ print("Failed to speech recognition.")