95 lines
2.8 KiB
Python
95 lines
2.8 KiB
Python
# -*- 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)
|