xche_audio / app.py
Yarik
update 2 app.py
065c08d
raw
history blame
4.76 kB
import argparse
import ctypes
import gc
import os
import sys
import tempfile
import threading
import time
import uvicorn
from fastapi import FastAPI, Depends, HTTPException
from fastapi.responses import FileResponse, JSONResponse
from fastapi.security import OAuth2PasswordBearer, OAuth2PasswordRequestForm
from huggingface_hub import hf_hub_download
from passlib.context import CryptContext
from pydantic import BaseModel
from torch import no_grad, package
from accentor import accentification, stress_replace_and_shift
app = FastAPI(docs_url=None, redoc_url=None)
# Set environment variable for Hugging Face cache directory and Stanza resources directory
os.environ["HF_HOME"] = "/app/.cache"
os.environ["STANZA_RESOURCES_DIR"] = "/app/stanza_resources"
os.environ["MODEL_DIR"] = "/app/model"
# Create necessary directories if they don't exist
os.makedirs(os.environ["STANZA_RESOURCES_DIR"], exist_ok=True)
os.makedirs(os.environ["MODEL_DIR"], exist_ok=True)
# Download the Ukrainian models for Stanza
import stanza
stanza.download('uk', model_dir=os.environ["STANZA_RESOURCES_DIR"])
tts_kwargs = {
"speaker_name": "uk",
"language_name": "uk",
}
oauth2_scheme = OAuth2PasswordBearer(tokenUrl="token")
class Auth(BaseModel):
api_key: str
password: str
# Password hashing context
pwd_context = CryptContext(schemes=["bcrypt"], deprecated="auto")
def get_password_hash(password):
return pwd_context.hash(password)
def verify_password(plain_password, hashed_password):
return pwd_context.verify(plain_password, hashed_password)
api_key = os.getenv("XCHE_API_KEY")
password = os.getenv("XCHE_PASSWORD")
fake_data_db = {
api_key: {
"api_key": api_key,
"password": get_password_hash(password) # Pre-hashed password
}
}
def get_api_key(db, api_key: str):
if api_key in db:
api_dict = db[api_key]
return Auth(**api_dict)
def authenticate(fake_db, api_key: str, password: str):
api_data = get_api_key(fake_db, api_key)
if not api_data:
return False
if not verify_password(password, api_data.password):
return False
return api_data
@app.post("/token")
async def login(form_data: OAuth2PasswordRequestForm = Depends()):
api_data = authenticate(fake_data_db, form_data.username, form_data.password)
if not api_data:
raise HTTPException(
status_code=400,
detail="Incorrect API KEY or Password",
headers={"WWW-Authenticate": "Bearer"},
)
return {"access_token": api_data.api_key, "token_type": "bearer"}
def check_api_token(token: str = Depends(oauth2_scheme)):
api_data = get_api_key(fake_data_db, token)
if not api_data:
raise HTTPException(status_code=403, detail="Invalid or missing API Key")
return api_data
def trim_memory():
libc = ctypes.CDLL("libc.so.6")
libc.malloc_trim(0)
gc.collect()
def init_models():
models = {}
model_path = hf_hub_download("theodotus/tts-vits-lada-uk", "model.pt", cache_dir=os.environ["MODEL_DIR"])
importer = package.PackageImporter(model_path)
models["lada"] = importer.load_pickle("tts_models", "model")
return models
@app.post("/create_audio")
async def tts(request: str, api_data: Auth = Depends(check_api_token)):
accented_text = accentification(request, "vocab")
plussed_text = stress_replace_and_shift(accented_text)
synt = models["lada"]
with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as wav_fp:
with no_grad():
wav_data = synt.tts(plussed_text, **tts_kwargs)
synt.save_wav(wav_data, wav_fp.name)
threading.Thread(target=delete_file_after_delay, args=(wav_fp.name, 300)).start()
return JSONResponse(content={"audio_url": f"https://pro100sata-xche-audio.hf.space/download_audio?audio_path={wav_fp.name}"})
@app.get("/download_audio")
async def download_audio(audio_path: str):
return FileResponse(audio_path, media_type='audio/wav')
models = init_models()
def delete_file_after_delay(file_path: str, delay: int):
time.sleep(delay)
if os.path.exists(file_path):
os.remove(file_path)
class ArgParser(argparse.ArgumentParser):
def __init__(self, *args, **kwargs):
super(ArgParser, self).__init__(*args, **kwargs)
self.add_argument(
"-s", "--server", type=str, default="0.0.0.0",
help="Server IP for HF LLM Chat API",
)
self.add_argument(
"-p", "--port", type=int, default=7860,
help="Server Port for HF LLM Chat API",
)
self.args = self.parse_args(sys.argv[1:])
if __name__ == "__main__":
args = ArgParser().args
uvicorn.run(app, host=args.server, port=args.port, reload=False)