update digithuman

auto_deploy_dight_human.sh

@@ -0,0 +1,29 @@
+#!/bin/bash
+# 定义logs文件夹的名称
+LOG_DIR="logs"
+
+cd /data/redserver/red-agent-service/scene-digit-human
+export HAIRUO_ENV=prod
+
+# 检查logs文件夹是否存在
+if [ ! -d "$LOG_DIR" ]; then
+    # 如果logs文件夹不存在，则创建它
+    echo "Creating logs directory..."
+    mkdir "$LOG_DIR"
+fi
+# 设置conda环境名
+conda_env_name="agent-common"
+echo "staring upload_api..."
+
+# 使用命令替换和if语句来检查conda环境中是否存在
+if conda env list | grep -q "$conda_env_name"; then
+    # 如果存在，则先杀掉进程
+    ps -ef | grep upload_api | grep -v grep | awk '{print $2}'| xargs kill -9 2>/dev/null || true
+    python /data/config-manager/generate_service_configs.py --service_config_info_path configs/config-vars.yml  --config_path configs/cfg.yml
+    conda run -n "$conda_env_name" gunicorn main:app -n digithuiman_api -c digithuman.conf.py --daemon
+    echo "Conda environment name is set to: $conda_env_name"
+    echo "Gunicorn started in the background in $conda_env_name environment."
+else
+    # 如果不存在，则打印找不到
+    echo "can not find : $conda_env_name"
+fi

common/__init__.py

@@ -0,0 +1,26 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+#
+# Copyright @2024 INSPUR Inc. (inspur.com)
+#
+# @author: J.G. Chen <chenjianguo@inspur.com>
+# @date: 2024/06/14
+#
+
+import os
+from enum import Enum
+
+__version__ = "v1.4.2"
+
+
+class HairuoEnv(str, Enum):
+    UNK = "unk"
+    TEST = "test"
+    DEV = "dev"
+    PROD = "prod"
+
+
+# 默认所有的机器需要设置 "HAIRUO_ENV" 环境变量, test/dev/prod
+HAIRUO_ENV = HairuoEnv(os.getenv("HAIRUO_ENV", "unk"))
+
+assert HAIRUO_ENV != HairuoEnv.UNK, "env var `HAIRUO_ENV` is required."

common/config.py

@@ -0,0 +1,218 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+#
+# Copyright @2024 INSPUR Inc. (inspur.com)
+#
+# @author: J.G. Chen <chenjianguo@inspur.com>
+# @date: 2024/02/17
+#
+"""
+对配置文件的一些操作
+
+Notes
+    - pyyaml:
+        * 对科学计数法的支持有特殊要求，对于 a[eE][+-]b：a 必须有小数点，指数必须含正负号
+        `pyyaml issues#173 <https://github.com/yaml/pyyaml/issues/173#issuecomment-507918276>`_;
+        * 或者使用 ruamel.yaml 代替；
+    - conf(hocon):
+        * 变量替换：不能添加引号，`{ b = "hello"\n a = ${b} world }`;
+        * 环境变量：'{ a = ${HOME} }';
+"""
+
+import json
+import os
+from pathlib import Path
+from platform import python_version
+from pprint import pformat
+from typing import Dict
+from typing import Union
+
+import yaml
+from omegaconf import DictConfig
+from omegaconf import ListConfig
+from omegaconf import OmegaConf
+from packaging import version
+from pyhocon import ConfigFactory
+from pyhocon import HOCONConverter
+
+from common import HAIRUO_ENV
+from common import HairuoEnv
+
+try:
+    import torch
+except ImportError:
+    torch = None
+
+try:
+    from common.utils.logger import init_logger
+
+    logger = init_logger(__name__).info
+except ImportError:
+    init_logger = None
+    logger = print
+
+
+class ConfigBase:
+    config: Union[DictConfig, ListConfig]
+
+    def __init__(self, config_file: str = ""):
+        """load config from `config_file`
+
+        Args:
+            config_file(str): config_file name, could be `.yaml`, `.conf`, `.json`, `.bin`
+        Notes:
+            config_file is relative to current work dir(cwd), pass to abs path
+        """
+        self._file_path = self._resolve(config_file)
+        self.config = self.load(self._file_path)
+
+    @classmethod
+    def _resolve(cls, config_file: str = ""):
+        cwd = Path.cwd()
+        valid_extension = (".yaml", ".yml", ".conf", ".json", ".bin")
+
+        # check config file in current work dir
+        if not config_file:
+            files = filter(lambda f: f.is_file(), cwd.iterdir())
+            logger(f"loading config file from dir: {cwd.resolve()}")
+            for filename in sorted(files, key=lambda x: os.path.getmtime(x), reverse=True):
+                if filename.suffix in valid_extension:
+                    config_file = Path(cwd, filename)
+                    break
+        else:
+            # parse config file specified.
+            config_file = Path(config_file)
+
+        # check config file
+        if not (config_file and config_file.is_file() and config_file.suffix in valid_extension):
+            raise FileNotFoundError(f"Config '{config_file}' not find!")
+
+        logger(f"loading config from {config_file.resolve()}")
+        return config_file
+
+    @classmethod
+    def show(cls, config: Union[Dict, DictConfig, ListConfig] = None):
+        """show config given or parsed self.config."""
+        kwargs = {} if version.parse(python_version()) < version.parse("3.8") else {"sort_dicts": False}
+        if not config:
+            config = cls.config
+
+        if isinstance(config, (DictConfig, ListConfig)):
+            config = OmegaConf.to_object(config)
+
+        logger(f"======= resolved config ======>\n{pformat(config, **kwargs).encode('utf-8')}")
+
+    @classmethod
+    def load(cls, path: Union[str, Path]) -> Union[DictConfig, ListConfig]:
+        """load config file."""
+        path = Path(path)
+        logger(f"loading config from: {path}")
+
+        if path.suffix in [".yaml", ".yml"]:
+            with path.open(encoding="utf-8") as f:
+                config = OmegaConf.create(yaml.load(f, Loader=yaml.FullLoader), flags={"allow_objects": True})
+        elif path.suffix == ".conf":
+            config = ConfigFactory.parse_file(path.as_posix())
+            config = OmegaConf.create(HOCONConverter.to_json(config))
+        elif path.suffix == ".json":
+            with path.open() as f:
+                config = OmegaConf.create(json.load(f))
+        elif path.suffix == ".bin":
+            if torch is not None:
+                config = OmegaConf.create(torch.load(path))
+            else:
+                raise RuntimeError("`torch` required to load .bin config file.")
+        else:
+            raise RuntimeError("unsupported file format to load.")
+
+        logger(f"parse configs for ENV: {HAIRUO_ENV}")
+        # keep configs for given env only
+        # TODO: update fields recursively
+        config.update(config.get(HAIRUO_ENV, dict()))
+        for env in HairuoEnv:
+            config.pop(env, default=None)
+
+        ConfigBase.show(config)
+        return config
+
+    @classmethod
+    def save_config(cls, config: Union[dict, DictConfig, ListConfig], path: Union[str, Path]):
+        """save `config` to given `path`"""
+        path = Path(path)
+
+        def convert_to_ct(_config):
+            # convert dict to ConfigTree recursively.
+            if isinstance(_config, (dict, DictConfig)):
+                tmp = {}
+                for k, v in _config.items():
+                    tmp[k] = convert_to_ct(v)
+                return ConfigFactory.from_dict(tmp)
+            elif isinstance(_config, ListConfig):
+                return [convert_to_ct(i) for i in _config]
+            else:
+                return _config
+
+        if path.suffix in [".yaml", ".yml"]:
+            OmegaConf.save(config, path)
+        elif path.suffix == ".conf":
+            with open(path, "w") as writer:
+                writer.write(HOCONConverter.to_hocon(convert_to_ct(config)))
+                writer.write("\n")
+        elif path.suffix == ".json":
+            with open(path, "w") as writer:
+                json.dump(config, writer, ensure_ascii=False, indent=2)
+        elif path.suffix == ".bin":
+            if torch is not None:
+                torch.save(config, path)
+            else:
+                raise RuntimeError("`torch` required to load .bin config file.")
+        else:
+            raise RuntimeError("unsupported file format.")
+        logger(f"saving config to: {path}")
+
+    def save(self, path: Union[str, Path] = "", key: str = ""):
+        """save resolved config obj to `path` or path provided by `key` in config file.
+
+        Args:
+            path: save
+            key: the `path` is resolved from `key` in resolved config.
+        """
+
+        if path:
+            pass
+        elif key and self.config.get(key, ""):
+            path = self.config.get(key)
+        else:
+            raise RuntimeError("target dir not found")
+
+        if Path(path).is_dir():
+            path = Path(path) / self._file_path.name
+        else:
+            path = Path(path)
+
+        self.save_config(self.config, path)
+
+
+if __name__ == "__main__":
+    import argparse
+
+    parser = argparse.ArgumentParser(
+        description="tool to convert config file format",
+        formatter_class=argparse.ArgumentDefaultsHelpFormatter,
+    )
+    args = parser.add_argument
+    args("path", metavar="config_file", help="path to config file.")
+    args("-f", "--format", type=str, default="conf", choices=["conf", "yaml", "json"], help="save conf in format.")
+    args("-o", "--output", type=str, help="output config to file.")
+    args("-s", "--show", action="store_true", help="just show the resolved config.")
+    params = parser.parse_args()
+
+    conf = ConfigBase.load(params.path)
+    output = params.output
+    if not output:
+        output = Path(params.path).with_suffix(f".{params.format.strip('.')}")
+
+    if params.show:
+        ConfigBase.show(conf)
+    else:
+        ConfigBase.save_config(conf, output)

common/security_check.py

@@ -0,0 +1,40 @@
+import requests
+
+
+def security_check(SECURITY_URL: str, auth_code: str, question: str, isRejection: bool, isRefusal: bool):
+    '''
+    SECURITY_URL: 拒识拒答url
+    auth_code: token
+    question: 被检测的字符串
+    isRejection: 是否进行拒识检测
+    isRefusal: 是否进行拒答检测
+    threshold: 知识库检索最低阈值
+    '''
+    headers = {
+        'content-type': "application/json",
+        'authorization': auth_code,
+    }
+    security_json = {
+        "query": question,
+        "isRejection": isRejection,
+        "isRefusal": isRefusal,
+    }
+    try:
+        security_res = requests.post(SECURITY_URL, json=security_json, headers=headers)
+        # {'code': 2, 'message': '拒识检测未通过!', 'result': False}
+        security_res_json = security_res.json()
+        if not security_res_json['result']:
+            return {
+                "result": False,
+                "msg": security_res_json["message"]
+            }
+        elif security_res_json['result']:
+            return {
+                "result": True,
+                "msg": security_res_json["message"]
+            }
+    except Exception as e:
+        return {
+            "result": False,
+            "msg": f"error: {e}"
+        }

common/use_model.py

@@ -0,0 +1,87 @@
+import sys
+
+sys.path.append("..")
+from utils.utils import get_logger
+import time
+import requests
+import json
+
+logger = get_logger("hairuo_general_vllm")
+
+
+def ask_question(query_url, messages, stream):
+    if stream:
+        return vllm_chat_stream(query_url, messages)
+    else:
+        return vllm_chat_non_flow(query_url, messages)
+
+
+def vllm_chat_non_flow(query_url, messages):
+    try:
+        T1 = time.time()
+        inter_q = [
+                      {
+                          "role": "system",
+                          "content": "You are a helpful assistant."
+                      }
+                  ] + messages
+        post_json = {
+            "model": "general_model",
+            "stream": False,
+            "stop": "<|im_end|>",
+            "messages": inter_q
+        }
+        headers = {
+            'Content-Type': 'application/json'
+        }
+        response = requests.post(query_url, headers=headers, json=post_json, timeout=60)
+        T2 = time.time()
+        logger.info('程序运行时间:%s毫秒' % ((T2 - T1) * 1000))
+        logger.info(f'Ask result:{response.json()}')
+        if response.status_code == 200:
+            data_ok = response.json()
+            first_choice = data_ok['choices'][0]
+            message_content = first_choice['message']['content']
+            return message_content
+        else:
+            return response.json()
+    except Exception as e:
+        logger.info(f"Ask occur an error: {e}")
+        return "Model running abnormally!"
+
+
+def vllm_chat_stream(query_url, messages):
+    inter_q = [
+                  {
+                      "role": "system",
+                      "content": "You are a helpful assistant."
+                  }
+              ] + messages
+    req_body = {
+        "model": 'general_model',
+        "stream": True,
+        "stop": "<|im_end|>",
+        "messages": inter_q
+    }
+
+    headers = {'Content-Type': 'application/json;charset=utf-8'}
+    response = requests.post(query_url, json=req_body, headers=headers, stream=True)
+    answer = ""
+    # answer_result = []
+    for chunk in response.iter_lines():
+        if chunk:
+            chunk = chunk.decode('utf-8')
+            chunk = chunk.replace("data: ", "")
+            try:
+                try:
+                    chunk = json.loads(chunk)
+                except:
+                    logger.info("-----最后一条------------------------------------")
+                if type(chunk) != str:
+                    if chunk['choices'][0]['delta'].get('content', '') != '':
+                        answer = answer + chunk['choices'][0]['delta']['content']
+                        if chunk['choices'][0]["finish_reason"] != "stop":
+                            yield answer
+            except Exception as e:
+                logger.info("流式输出error: %s", e)
+    return answer

configs/cfg.yml

@@ -0,0 +1,10 @@
+# 数字人
+dighthuman:
+     dev:
+         dh_webui: http://100.200.128.72:14041/agentstore/api/v1/multimodal_models/dh/dighthuman
+     test:
+         dh_webui: http://100.200.128.72:14041/agentstore/api/v1/multimodal_models/dh/dighthuman
+     prod:
+         dh_webui: http://100.200.128.72:14041/agentstore/api/v1/multimodal_models/dh/dighthuman
+     model_name: HaiRuo-AudioVisual-ST
+     static_token: 7c3eafb5-2d6e-100d-ab0f-7b2c1cdafb3c

configs/config-template.j2

@@ -0,0 +1,9 @@
+dighthuman:
+     dev:
+         dh_webui: {{ red_server.dighthuman.dev.dh_webui }}
+     test:
+         dh_webui: {{ red_server.dighthuman.test.dh_webui }}
+     prod:
+         dh_webui: {{ red_server.dighthuman.prod.dh_webui }}
+     model_name: {{ red_server.dighthuman.model_name }}
+     static_token: {{ global.authorization.static_token }}

configs/config-vars-dev.yml

@@ -0,0 +1,10 @@
+red_server:
+  dighthuman:
+      dev:
+          dh_webui: http://127.0.0.1:14040/agentstore/api/v1/multimodal_models/dh/dighthuman
+      test:
+          dh_webui: http://127.0.0.1:14040/agentstore/api/v1/multimodal_models/dh/dighthuman
+      prod:
+          dh_webui: http://127.0.0.1:14040/agentstore/api/v1/multimodal_models/dh/dighthuman
+      model_name: HaiRuo-AudioVisual-ST
+      static_token: 7c3eafb5-2d6e-100d-ab0f-7b2c1cdafb3c

configs/config-vars-mindie.yml

@@ -0,0 +1,10 @@
+red_server:
+  dighthuman:
+      dev:
+          dh_webui: http://127.0.0.1:14040/agentstore/api/v1/multimodal_models/dh/dighthuman
+      test:
+          dh_webui: http://127.0.0.1:14040/agentstore/api/v1/multimodal_models/dh/dighthuman
+      prod:
+          dh_webui: http://127.0.0.1:14040/agentstore/api/v1/multimodal_models/dh/dighthuman
+      model_name: HaiRuo-AudioVisual-ST
+      static_token: 7c3eafb5-2d6e-100d-ab0f-7b2c1cdafb3c

configs/config-vars-prod.yml

@@ -0,0 +1,10 @@
+red_server:
+  dighthuman:
+      dev:
+          dh_webui: http://127.0.0.1:14040/agentstore/api/v1/multimodal_models/dh/dighthuman
+      test:
+          dh_webui: http://127.0.0.1:14040/agentstore/api/v1/multimodal_models/dh/dighthuman
+      prod:
+          dh_webui: http://127.0.0.1:14040/agentstore/api/v1/multimodal_models/dh/dighthuman
+      model_name: HaiRuo-AudioVisual-ST
+      static_token: 7c3eafb5-2d6e-100d-ab0f-7b2c1cdafb3c

configs/config-vars-test.yml

@@ -0,0 +1,10 @@
+red_server:
+  dighthuman:
+      dev:
+          dh_webui: http://100.200.128.83:14040/agentstore/api/v1/multimodal_models/dh/dighthuman
+      test:
+          dh_webui: http://100.200.128.83:14040/agentstore/api/v1/multimodal_models/dh/dighthuman
+      prod:
+          dh_webui: http://100.200.128.83:14040/agentstore/api/v1/multimodal_models/dh/dighthuman
+      model_name: HaiRuo-AudioVisual-ST
+      static_token: 7c3eafb5-2d6e-100d-ab0f-7b2c1cdafb3c

digit_human.conf.py

@@ -0,0 +1,30 @@
+    # -*- coding: utf-8 -*-
+
+import os
+
+path_of_current_file = os.path.abspath(__file__)
+path_of_current_dir = os.path.split(path_of_current_file)[0]
+
+# worker_class为sync会报错
+# uvicorn.workers.UvicornWorker
+worker_class = 'uvicorn.workers.UvicornWorker'
+# workers = multiprocessing.cpu_count() * 2 + 1
+workers = 1     # 按需启动的进程数
+threads = 1     # 各进程包含的线程数
+
+chdir = path_of_current_dir
+
+worker_connections = 0
+timeout = 0
+max_requests = 0
+graceful_timeout = 0
+
+loglevel = 'info'
+access_log_format = '%(t)s %(p)s %(h)s "%(r)s" %(s)s %(L)s %(b)s %(f)s" "%(a)s"'
+reload = True
+debug = False
+bind = "%s:%s" % ("0.0.0.0", 14040)
+pidfile = '%s/digithuiman.pid' % (path_of_current_dir)
+errorlog = '%s/logs/digithuiman.log' % (path_of_current_dir)
+accesslog = '%s/logs/digithuiman_access.log' % (path_of_current_dir)
+proc_name = "digithuiman_api"

main.py

@@ -0,0 +1,94 @@
+# -*- encoding: utf-8 -*-
+'''
+@Email   :   liaoxiju@inspur.com
+'''
+import edge_tts
+import os
+import re
+import sys
+sys.path.append("../")
+sys.path.append("../..")
+import datetime
+from datetime import timedelta
+import logging
+import yaml
+import uvicorn
+from common import HAIRUO_ENV
+from common import HairuoEnv
+from fastapi import FastAPI, Request
+from agent_common_utils.logger import get_logger
+from agent_common_utils.function_monitor import log_function_call
+from model_utils import audio_driven_video
+import tempfile
+
+logger = get_logger("digithuman")
+
+dh_audio_driven_video = log_function_call(logger)(audio_driven_video)
+#dh_audio_driven_video = audio_driven_video
+
+app = FastAPI()
+
+def upload_to_s3(filepath):
+    os.system(f's3cmd put --expiry-days 10 {filepath} -r s3://ihp/tmp/liao/service/edge-tts/result/')
+
+def text_to_wav(text):
+    save_path = "result"
+    if not os.path.exists(save_path):
+        os.system(f"mkdir -p {save_path}")
+    if text is None or type(text) != str or len(text) == 0:
+        raise "input text is Error"
+    with tempfile.TemporaryDirectory(dir=save_path) as tmp_dir:
+        output_stream = edge_tts.Communicate(text.strip(), "zh-CN-XiaoyiNeural")
+        wav_save_path = f"{tmp_dir}/text2audio.wav"
+        output_stream.save_sync(wav_save_path)
+        #return save_path, tmp_dir
+        upload_to_s3(tmp_dir)
+    return f"https://ihp.oss.cn-north-4.inspurcloudoss.com/tmp/liao/service/edge-tts/{wav_save_path}"
+
+
+@app.post("/hairuo/digithuman")
+async def digithuman(request: Request, req: dict):
+    '''
+    对输入文本文本描述，生成wav音频，然后调用dh-webui生成视频
+    '''
+
+    logger.info("<digithuman> Verification authorization.")
+    body_data = await request.body()
+    body_data = body_data.decode("utf-8")
+
+    logger.info("<body> {}".format(body_data))
+    logger.info("<digithuman> input text.")
+    try:
+        text = req.get("text")
+
+        ##edge tts生成音频，上传oss，获取音频url
+        wav_url = text_to_wav(text)
+
+        code, video_url = dh_audio_driven_video(wav_url)
+        if code != 0:
+            logger.info("<digithuman> dh model error")
+            resp = {
+                     "code": "1",
+                     "message": "dh model error",
+                     "result": ''
+                    }
+            return resp
+        resp = {
+            "code": "0",
+            "message": "success",
+            "result": {"video_url":video_url}
+        }
+        return resp
+        
+    except:
+        logger.info("<digithuman> model error")
+        resp = {     
+                     "code": "1",
+                     "message": "dh model call error",
+                     "result": ""
+                    }
+        return resp
+
+
+if __name__ == "__main__":
+    uvicorn.run(app, host="0.0.0.0", port=14040)

model_utils.py

@@ -0,0 +1,40 @@
+# -*- encoding: utf-8 -*-
+'''
+@File    :   model_utils.py
+@Time    :   2024/07/24 10:28:28
+@Author  :   liangzz1991
+@Email   :   zhaoliang03@inspur.com
+'''
+import json
+import time
+import yaml
+import requests
+from common.config import ConfigBase
+from common import HAIRUO_ENV
+from common import HairuoEnv
+
+def get_cfg(config):
+    config = config.dighthuman
+    config.update(config.get(HAIRUO_ENV, dict()))
+    for env in HairuoEnv:
+        config.pop(env, default=None)
+    ConfigBase.show(config)
+    return config
+
+configs = ConfigBase.load('configs/cfg.yml')
+configs = get_cfg(configs)
+STATIC_TOKEN = configs['static_token']
+DH_URL = configs['dh_webui']
+model_name = configs['model_name']
+
+def audio_driven_video(wav_url):
+    data = {
+         "audio_url": wav_url
+         }
+    response = requests.post(DH_URL, data=json.dumps(data),timeout=60,headers={'Content-Type': 'application/json', 'Authorization': STATIC_TOKEN})
+    if response.status_code != 200:
+        return 1, ""
+    else:
+        video_url = response.json()['result']
+        return 0, video_url
+    

sadtalker-server/sadtalker.conf.py

@@ -0,0 +1,30 @@
+    # -*- coding: utf-8 -*-
+
+import os
+
+path_of_current_file = os.path.abspath(__file__)
+path_of_current_dir = os.path.split(path_of_current_file)[0]
+
+# worker_class为sync会报错
+# uvicorn.workers.UvicornWorker
+worker_class = 'uvicorn.workers.UvicornWorker'
+# workers = multiprocessing.cpu_count() * 2 + 1
+workers = 1     # 按需启动的进程数
+threads = 1     # 各进程包含的线程数
+
+chdir = path_of_current_dir
+
+worker_connections = 0
+timeout = 0
+max_requests = 0
+graceful_timeout = 0
+
+loglevel = 'info'
+access_log_format = '%(t)s %(p)s %(h)s "%(r)s" %(s)s %(L)s %(b)s %(f)s" "%(a)s"'
+reload = True
+debug = False
+bind = "%s:%s" % ("0.0.0.0", 14041)
+pidfile = '%s/sadtalker.pid' % (path_of_current_dir)
+errorlog = '%s/sadtalker.log' % (path_of_current_dir)
+accesslog = '%s/sadtalker_access.log' % (path_of_current_dir)
+proc_name = "sadtalker_api"

sadtalker-server/sadtalker_server.py

@@ -0,0 +1,107 @@
+# -*- encoding: utf-8 -*-
+'''
+@Email   :   liaoxiju@inspur.com
+'''
+import os
+import re
+import sys
+import uvicorn
+from fastapi import FastAPI, Request
+from modelscope.pipelines import pipeline
+import requests
+import tempfile
+import torch
+import time
+
+if torch.cuda.is_available():
+    torch.cuda.set_device(0)
+
+app = FastAPI()
+
+model_path = os.getenv('MODEL_PATH','./wwd123/sadtalker')
+
+inference = pipeline('talking-head', model=model_path, model_revision='v1.0.0', device='cuda', use_gpu=True, cache="./")
+
+def upload_to_s3(filepath):
+    timestap = int(time.time())
+    os.system(f's3cmd put --expiry-days 10 {filepath} -r s3://ihp/tmp/liao/service/sadtalker/results/{timestap}.mp4')
+    return f"results/{timestap}.mp4"
+
+def download_wav(wav_url):
+    try:
+        if not os.path.exists("results"):
+            os.system("mkdir -p results")
+        res = requests.get(wav_url)
+        content = res.content
+        tmp_dir = tempfile.TemporaryDirectory(dir="results").name
+        if not os.path.exists(tmp_dir):
+            os.system(f"mkdir -p {tmp_dir}")
+        wav_path = os.path.join(tmp_dir, "input_audio.wav")
+        fout = open(wav_path, "wb")
+        fout.write(content)
+        fout.close()
+        return wav_path, tmp_dir
+    except:
+        import traceback
+        print("download error", traceback.format_exc())
+        return None
+
+#@app.post("/hairuo/audiodrivenvido")
+@app.post("/agentstore/api/v1/multimodal_models/dh/dighthuman")
+async def digithuman(request: Request, req: dict):
+    '''
+    获取音频url-wav音频，然后使用sadtalker
+    '''
+    body_data = await request.body()
+    body_data = body_data.decode("utf-8")
+    try:   
+        timestamp = int(time.time())
+        save_path = f"results_{timestamp}"
+        if not os.path.exists(save_path):
+            os.system(f"mkdir -p {save_path}")
+
+        audio_url = req.get("audio_url")
+        wav_path, tmp_dir = download_wav(audio_url)
+        source_image = 'nv.jpg'
+
+        kwargs = {
+            'preprocess' : 'crop', # 'crop', 'resize', 'full'
+            'still_mode' : True,
+            'use_enhancer' : False,
+            'batch_size' : 1,
+            'size' : 256, # 256, 512
+            'pose_style' : 0,
+            'exp_scale' : 1,
+            'result_dir': save_path
+        }
+        
+        video_path = inference(source_image, driven_audio=wav_path, **kwargs)
+
+        video_name = upload_to_s3(video_path)
+
+        ##清除临时文件
+        os.system(f"rm -rf {tmp_dir}")
+        os.system(f"rm {video_path}")
+        os.system(f"rm -rf {save_path}")
+
+        video_url = os.path.join("https://ihp.oss.cn-north-4.inspurcloudoss.com/tmp/liao/service/sadtalker/", video_name)
+        resp = {
+            "code": "0",
+            "message": "success",
+            "result": video_url
+        }
+        return resp
+        
+    except:
+        import traceback
+        print(traceback.format_exc())
+        resp = {     
+                     "code": "1",
+                     "message": "audiodrivenvido model call error",
+                     "result": ""
+                    }
+        return resp
+
+
+if __name__ == "__main__":
+    uvicorn.run(app, host="0.0.0.0", port=14041)

sadtalker-server/start_sadtalker.sh

@@ -0,0 +1,25 @@
+#!/bin/bash
+
+sconfig_common
+
+# 检查命令是否存在
+if npu-smi >/dev/null 2>&1; then
+    echo "npu-smi命令存在, 使用ASCEND_RT_VISIBLE_DEVICES执行相关操作..."
+    # 在这里执行需要的操作
+    ASCEND_RT_VISIBLE_DEVICES=1 gunicorn --env MODEL_PATH="./wwd123/sadtalker" sadtalker_server:app -n sadtalker_api -c sadtalker.conf.py --daemon
+    exit 0
+else
+    echo "npu-smi命令不存在, 跳过该部分脚本。"
+fi
+
+# 检查命令是否存在
+if cnmon >/dev/null 2>&1; then
+    echo "cnmon命令存在, 使用MLU_VISIBLE_DEVICES执行相关操作..."
+    # 在这里执行需要的操作
+    MLU_VISIBLE_DEVICES=1 gunicorn --env MODEL_PATH="./wwd123/sadtalker" sadtalker:app -n sadtalker_api -c sadtalker.conf.py --daemon
+    exit 0
+else
+    echo "cnmon命令不存在, 跳过该部分脚本。"
+fi
+
+CUDA_VISIBLE_DEVICES=1 gunicorn --env MODEL_PATH="./wwd123/sadtalker" sadtalker_server:app -n sadtalker_api -c sadtalker.conf.py --daemon

sadtalker-server/stop_sadtalker.sh

@@ -0,0 +1 @@
+ps -ef | grep sadtalker_api | grep -v grep | awk '{print $2}'| xargs kill -9

start.sh

@@ -0,0 +1,8 @@
+SELF_DIR=$(cd $(dirname "$0"); pwd)
+PROJECT_ROOT=${SELF_DIR}/..
+export PYTHONPATH=$PROJECT_ROOT:$PYTHONPATH
+cd $PROJECT_ROOT/scene-digit-human
+# python main.py
+mkdir logs
+python /data/config-manager/generate_service_configs.py --service_config_info_path configs/config-vars.yml  --config_path configs/cfg.yml
+gunicorn main:app -n digithuman_api -c digit_human.conf.py --daemon

stop.sh

@@ -0,0 +1 @@
+ps -ef | grep digithuman_api | grep -v grep | awk '{print $2}'| xargs kill -9

test.py

@@ -0,0 +1,32 @@
+"""
+@Email: liaoxiju@inspur.com
+"""
+from modelscope.piplines import pipeline
+
+# Create a pipeline instance for talking head generation using the specified model and revision.
+inference = pipeline('talking-head', model='./wwd123/sadtalker', model_revision='v1.0.0')
+
+# Define the input source image and audio file paths.
+source_image = "liao.jpg"
+driven_audio = "xx_cn.wav"
+
+# Set the output directory where results will be saved.
+out_dir = "./results/"
+
+# Configure various parameters for the inference process:
+kwargs = {
+    'preprocess': 'full',  # Options are 'crop', 'resize', or 'full'
+    'still_mode': True,
+    'use_enhancer': False,
+    'batch_size': 1,
+    'size': 256,  # Image size can be either 256 or 512 pixels
+    'pose_style': 0,
+    'exp_scale': 1,
+    'result_dir': out_dir
+}
+
+# Perform inference to generate the video from the source image and audio.
+video_path = inference(source_image=source_image, driven_audio=driven_audio, **kwargs)
+
+# Print the path of the generated video file.
+print(f"==>> video_path: {video_path}")