File size: 11,681 Bytes
2b7fd6e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 |
# -*- coding: utf-8 -*-
import logging
import tempfile
from inference.infer_tool import Svc
from typing import *
import api.base
import os
import io
import wave
import numpy as np
from service.tool import audio_normalize, read_wav_file_to_numpy_array
from utils import get_hparams_from_file
logger = logging.getLogger(__name__)
_svc: Optional[Svc] = None
_model_paths: Optional[List] = None
def init():
global _svc, _model_paths
_svc = Svc()
_model_paths = []
# get the path to the parent directory
parent_dir = os.path.abspath(os.path.join(os.getcwd(), os.curdir))
# construct the path to the "checkpoints" directory
checkpoints_dir = os.path.join(parent_dir, "checkpoints")
logger.debug(f"CkPoints Dir: {checkpoints_dir}")
for root, dirs, files in os.walk(checkpoints_dir):
for dir in dirs:
_model_paths.append(dir)
# noinspection PyAbstractClass
class ModelListHandler(api.base.ApiHandler):
async def get(self):
self.write({
"code": 200,
"msg": "ok",
"data": _model_paths
})
# noinspection PyAbstractClass
class SwitchHandler(api.base.ApiHandler):
async def post(self):
model_name = self.get_argument("model", "") # model name
mode = self.get_argument("mode", "single") # running mode: single or batch
device = self.get_argument("device", "cuda") # "cpu" or "cuda"
if model_name == "":
self.set_status(400)
self.write({
"code": 400,
"msg": "未选择模型!",
"data": None
})
return
if mode not in ("single", "batch"):
self.set_status(400)
self.write({
"code": 400,
"msg": "运行模式选择错误!",
"data": None
})
return
if device not in ("cpu", "cuda"):
self.set_status(400)
self.write({
"code": 400,
"msg": "设备选择错误!",
"data": None
})
return
logger.debug(f"modelname: {model_name}\n"
f"mode: {mode}\n"
f"device: {device}\n")
try:
_svc.set_device(device=device)
logger.debug(f"Device set.")
_svc.load_checkpoint(path=model_name)
logger.debug(f"Model set.")
except Exception as e:
logger.exception(e)
self.set_status(500)
self.write({
"code": 500,
"msg": "system_error",
"data": None
})
return
self.write({
"code": 200,
"msg": "ok",
"data": {
"mode": mode
}
})
# noinspection PyAbstractClass
class SingleInferenceHandler(api.base.ApiHandler):
async def post(self):
try:
from scipy.io import wavfile
dsid = self.get_argument("dsid", "")
tran = self.get_argument("tran", "0")
th = self.get_argument("th", "-40.0")
ns = self.get_argument("ns", "0.4")
audiofile_dict = self.request.files.get("srcaudio", [])
if not audiofile_dict:
self.set_status(400)
self.write({
"code": 400,
"msg": "未上传文件!",
"data": None
})
return
if dsid == "":
self.set_status(400)
self.write({
"code": 400,
"msg": "未选择模型!",
"data": None
})
return
audiofile = audiofile_dict[0]
audio_filename = audiofile['filename']
audio_filebody = audiofile['body']
audio_fileext = os.path.splitext(audio_filename)[-1].lower()
with tempfile.NamedTemporaryFile(suffix=audio_fileext, delete=False) as temp_file:
temp_file.write(audio_filebody)
temp_file.close()
converted_file = await audio_normalize(full_filename=audio_filename, file_data=audio_filebody)
# if audio_fileext != ".wav":
# logger.debug(f"file format is {audio_fileext}, not wav\n"
# f"converting to standard wav data...")
# converted_file = await audio_normalize(full_filename=audio_filename, file_data=audio_filebody)
# logger.debug(f"wav conversion completed.")
# else:
# converted_file = temp_file.name
sampling_rate, audio_array = read_wav_file_to_numpy_array(converted_file)
os.remove(converted_file)
scraudio = (sampling_rate, audio_array)
logger.debug(f"read file {audio_filename}\n"
f"sampling rate: {sampling_rate}")
tran = float(tran)
th = float(th)
ns = float(ns)
hparams = get_hparams_from_file(f"checkpoints/{dsid}/config.json")
spk = hparams.spk
real_dsid = ""
for k, v in spk.items():
if v == 0:
real_dsid = k
logger.debug(f"read dsid is: {real_dsid}")
output_audio_sr, output_audio_array = _svc.inference(srcaudio=scraudio,
chara=real_dsid,
tran=tran,
slice_db=th,
ns=ns)
logger.debug(f"svc for {audio_filename} succeed. \n"
f"audio data type: {type(output_audio_array)}\n"
f"audio data sr: {output_audio_sr}")
logger.debug(f"start output data.")
# Convert the NumPy array to WAV format
with io.BytesIO() as wav_file:
wavfile.write(wav_file, sampling_rate, output_audio_array)
wav_data = wav_file.getvalue()
# set the response headers and body
self.set_header('Content-Type', 'audio/wav')
self.set_header('Content-Disposition', f'attachment; filename="svc_output.wav"')
self.write(wav_data)
await self.flush()
logger.debug(f"response completed.")
except Exception as e:
logger.exception(e)
self.set_status(500)
self.write({
"code": 500,
"msg": "system_error",
"data": None
})
return
# noinspection PyAbstractClass
class BatchInferenceHandler(api.base.ApiHandler):
async def post(self):
try:
from zipfile import ZipFile
from scipy.io import wavfile
import uuid
dsid = self.get_argument("dsid", "")
tran = self.get_argument("tran", "0")
th = self.get_argument("th", "-40.0")
ns = self.get_argument("ns", "0.4")
audiofile_dict = self.request.files.get("srcaudio", [])
logger.debug(len(self.request.files))
if not audiofile_dict:
self.set_status(400)
self.write({
"code": 400,
"msg": "未上传文件!",
"data": None
})
return
if dsid == "":
self.set_status(400)
self.write({
"code": 400,
"msg": "未选择模型!",
"data": None
})
return
temp_dir_name = "temp"
# get the path to the parent directory
parent_dir = os.path.abspath(os.path.join(os.getcwd(), os.curdir))
# construct the path to the "temp" directory
temp_dir = os.path.join(parent_dir, temp_dir_name)
logger.debug(f"TempDir: {temp_dir}")
if not os.path.exists(temp_dir):
os.mkdir(temp_dir)
tmp_workdir_name = f"{temp_dir}/batch_{uuid.uuid4()}"
if not os.path.exists(tmp_workdir_name):
os.mkdir(tmp_workdir_name)
output_files = []
tran = float(tran)
th = float(th)
ns = float(ns)
hparams = get_hparams_from_file(f"checkpoints/{dsid}/config.json")
spk = hparams.spk
real_dsid = ""
for k, v in spk.items():
if v == 0:
real_dsid = k
logger.debug(f"read dsid is: {real_dsid}")
for idx, file in enumerate(audiofile_dict):
audio_filename = file["filename"]
audio_filebody = file["body"]
filename = os.path.basename(audio_filename)
audio_fileext = os.path.splitext(audio_filename)[-1].lower()
with tempfile.NamedTemporaryFile(suffix=audio_fileext, delete=False) as temp_file:
temp_file.write(audio_filebody)
temp_file.close()
converted_file = await audio_normalize(full_filename=audio_filename, file_data=audio_filebody)
# if audio_fileext != ".wav":
# logger.debug(f"file format is {audio_fileext}, not wav\n"
# f"converting to standard wav data...")
# converted_file = await audio_normalize(full_filename=audio_filename, file_data=audio_filebody)
# logger.debug(f"wav conversion completed.")
# else:
# converted_file = temp_file.name
sampling_rate, audio_array = read_wav_file_to_numpy_array(converted_file)
os.remove(converted_file)
scraudio = (sampling_rate, audio_array)
print(f"{idx}, {len(audio_filebody)}, {filename}")
output_sampling_rate, output_audio = _svc.inference(scraudio, chara=real_dsid, tran=tran,
slice_db=th, ns=ns)
new_filepath = f"{tmp_workdir_name}/{filename}"
wavfile.write(filename=new_filepath, rate=output_sampling_rate, data=output_audio)
output_files.append(new_filepath)
zipfilename = f"{tmp_workdir_name}/output.zip"
with ZipFile(zipfilename, "w") as zip_obj:
for idx, filepath in enumerate(output_files):
zip_obj.write(filepath, os.path.basename(filepath))
# todo: remove data
logger.debug(f"start output data.")
# set response header and body
self.set_header("Content-Type", "application/zip")
self.set_header("Content-Disposition", "attachment; filename=output.zip")
with open(zipfilename, "rb") as file:
self.write(file.read())
await self.flush()
logger.debug(f"response completed.")
except Exception as e:
logger.exception(e)
self.set_status(500)
self.write({
"code": 500,
"msg": "system_error",
"data": None
})
return
if __name__ == "__main__":
init()
print(_model_paths)
|