Merge pull request #1945 from PaddlePaddle/asr_line

[server][asr] refactor asr streaming server and remove useless code

demos/audio_searching/src/operations/load.py

@@ -26,9 +26,8 @@ def get_audios(path):
     """
     supported_formats = [".wav", ".mp3", ".ogg", ".flac", ".m4a"]
     return [
-        item
-        for sublist in [[os.path.join(dir, file) for file in files]
-                        for dir, _, files in list(os.walk(path))]
+        item for sublist in [[os.path.join(dir, file) for file in files]
+                             for dir, _, files in list(os.walk(path))]
         for item in sublist if os.path.splitext(item)[1] in supported_formats
     ]
 

demos/streaming_asr_server/websocket_client.py

@@ -13,9 +13,7 @@
 # limitations under the License.
 #!/usr/bin/python
 # -*- coding: UTF-8 -*-
-
 # script for calc RTF: grep -rn RTF log.txt | awk '{print $NF}' | awk -F "=" '{sum += $NF} END {print "all time",sum, "audio num", NR,  "RTF", sum/NR}'
-
 import argparse
 import asyncio
 import codecs

docs/source/asr/PPASR_cn.md

@@ -92,5 +92,3 @@ server 的 demo： [streaming_asr_server](https://github.com/PaddlePaddle/Paddle
 ## 4. 快速开始
 
 关于如果使用 PP-ASR，可以看这里的 [install](https://github.com/PaddlePaddle/PaddleSpeech/blob/develop/docs/source/install_cn.md)，其中提供了 **简单**、**中等**、**困难** 三种安装方式。如果想体验 paddlespeech 的推理功能，可以用 **简单** 安装方式。
-
-

paddlespeech/cli/utils.py

@@ -24,11 +24,11 @@ from typing import Any
 from typing import Dict
 
 import paddle
+import paddleaudio
 import requests
 import yaml
 from paddle.framework import load
 
-import paddleaudio
 from . import download
 from .entry import commands
 try:

paddlespeech/s2t/io/sampler.py

@@ -51,7 +51,7 @@ def _batch_shuffle(indices, batch_size, epoch, clipped=False):
     """
     rng = np.random.RandomState(epoch)
     shift_len = rng.randint(0, batch_size - 1)
-    batch_indices = list(zip(*[iter(indices[shift_len:])] * batch_size))
+    batch_indices = list(zip(* [iter(indices[shift_len:])] * batch_size))
     rng.shuffle(batch_indices)
     batch_indices = [item for batch in batch_indices for item in batch]
     assert clipped is False

paddlespeech/server/engine/acs/__init__.py

+# Copyright (c) 2022 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.

paddlespeech/server/engine/acs/python/__init__.py

+# Copyright (c) 2022 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.

paddlespeech/server/restful/api.py

@@ -17,12 +17,12 @@ from typing import List
 from fastapi import APIRouter
 
 from paddlespeech.cli.log import logger
+from paddlespeech.server.restful.acs_api import router as acs_router
 from paddlespeech.server.restful.asr_api import router as asr_router
 from paddlespeech.server.restful.cls_api import router as cls_router
 from paddlespeech.server.restful.text_api import router as text_router
 from paddlespeech.server.restful.tts_api import router as tts_router
 from paddlespeech.server.restful.vector_api import router as vec_router
-from paddlespeech.server.restful.acs_api import router as acs_router
 _router = APIRouter()
 
 

paddlespeech/server/utils/audio_handler.py

@@ -248,7 +248,7 @@ class ASRHttpHandler:
         }
 
         res = requests.post(url=self.url, data=json.dumps(data))
-        
+
         return res.json()
 
 

paddlespeech/server/utils/buffer.py

@@ -12,6 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+
 class Frame(object):
     """Represents a "frame" of audio data."""
 
@@ -45,7 +46,7 @@ class ChunkBuffer(object):
         self.shift_ms = shift_ms
         self.sample_rate = sample_rate
         self.sample_width = sample_width  # int16 = 2; float32 = 4
-        
+
         self.window_sec = float((self.window_n - 1) * self.shift_ms +
                                 self.window_ms) / 1000.0
         self.shift_sec = float(self.shift_n * self.shift_ms / 1000.0)
@@ -77,8 +78,8 @@ class ChunkBuffer(object):
 
         offset = 0
         while offset + self.window_bytes <= len(audio):
-            yield Frame(audio[offset:offset + self.window_bytes], self.timestamp,
-                        self.window_sec)
+            yield Frame(audio[offset:offset + self.window_bytes],
+                        self.timestamp, self.window_sec)
             self.timestamp += self.shift_sec
             offset += self.shift_bytes
 

paddlespeech/t2s/exps/speedyspeech/synthesize_e2e.py

@@ -176,7 +176,10 @@ def main():
     parser.add_argument(
         "--ngpu", type=int, default=1, help="if ngpu == 0, use cpu or xpu.")
     parser.add_argument(
-        "--nxpu", type=int, default=0, help="if nxpu == 0 and ngpu == 0, use cpu.")
+        "--nxpu",
+        type=int,
+        default=0,
+        help="if nxpu == 0 and ngpu == 0, use cpu.")
 
     args, _ = parser.parse_known_args()
 

paddlespeech/t2s/exps/speedyspeech/train.py

@@ -188,7 +188,10 @@ def main():
     parser.add_argument("--dev-metadata", type=str, help="dev data.")
     parser.add_argument("--output-dir", type=str, help="output dir.")
     parser.add_argument(
-        "--nxpu", type=int, default=0, help="if nxpu == 0 and ngpu == 0, use cpu.")
+        "--nxpu",
+        type=int,
+        default=0,
+        help="if nxpu == 0 and ngpu == 0, use cpu.")
     parser.add_argument(
         "--ngpu", type=int, default=1, help="if ngpu == 0, use cpu or xpu")
 

paddlespeech/t2s/modules/transformer/repeat.py

@@ -36,4 +36,4 @@ def repeat(N, fn):
     Returns:
         MultiSequential: Repeated model instance.
     """
-    return MultiSequential(*[fn(n) for n in range(N)])
+    return MultiSequential(* [fn(n) for n in range(N)])

setup.py

@@ -98,7 +98,6 @@ requirements = {
 }
 
 
-
 def check_call(cmd: str, shell=False, executable=None):
     try:
         sp.check_call(
@@ -112,12 +111,13 @@ def check_call(cmd: str, shell=False, executable=None):
             file=sys.stderr)
         raise e
 
+
 def check_output(cmd: str, shell=False):
     try:
         out_bytes = sp.check_output(cmd.split())
     except sp.CalledProcessError as e:
-        out_bytes = e.output       # Output generated before error
-        code      = e.returncode   # Return code
+        out_bytes = e.output  # Output generated before error
+        code = e.returncode  # Return code
         print(
             f"{__file__}:{inspect.currentframe().f_lineno}: CMD: {cmd}, Error:",
             out_bytes,
@@ -146,6 +146,7 @@ def _remove(files: str):
     for f in files:
         f.unlink()
 
+
 ################################# Install ##################################
 
 
@@ -308,6 +309,5 @@ setup_info = dict(
         ]
     })
 
-
 with version_info():
     setup(**setup_info)

tests/unit/cli/aishell_test_prepare.py

@@ -20,7 +20,6 @@ of each audio file in the data set.
 """
 import argparse
 import codecs
-import json
 import os
 from pathlib import Path
 
@@ -89,7 +88,7 @@ def create_manifest(data_dir, manifest_path_prefix):
                 duration = float(len(audio_data) / samplerate)
                 text = transcript_dict[audio_id]
                 json_lines.append(audio_path)
-                reference_lines.append(str(total_num+1) + "\t" + text)
+                reference_lines.append(str(total_num + 1) + "\t" + text)
 
                 total_sec += duration
                 total_text += len(text)
@@ -106,6 +105,7 @@ def create_manifest(data_dir, manifest_path_prefix):
 
         manifest_dir = os.path.dirname(manifest_path_prefix)
 
+
 def prepare_dataset(url, md5sum, target_dir, manifest_path=None):
     """Download, unpack and create manifest file."""
     data_dir = os.path.join(target_dir, 'data_aishell')