main.py

from fastapi import FastAPI, File, UploadFile, HTTPException
from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
import torch
import uvicorn
from fastapi.middleware.cors import CORSMiddleware
import os 
# Initialize FastAPI
app = FastAPI()
app.add_middleware(
    CORSMiddleware,
    allow_origins=['*'],
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

# Load the model and processor
device = "cuda:0" if torch.cuda.is_available() else "cpu"
torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32

model_id = "openai/whisper-large-v3"

# Check if model exists locally, otherwise download it
if not os.path.exists(f"./{model_id}"):
    model = AutoModelForSpeechSeq2Seq.from_pretrained(model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=True, use_safetensors=True)
    processor = AutoProcessor.from_pretrained(model_id)
else:
    model = AutoModelForSpeechSeq2Seq.from_pretrained(f"./{model_id}", torch_dtype=torch_dtype, low_cpu_mem_usage=True, use_safetensors=True)
    processor = AutoProcessor.from_pretrained(f"./{model_id}")

model.to(device)

pipe = pipeline(
    "automatic-speech-recognition",
    model=model,
    tokenizer=processor.tokenizer,
    feature_extractor=processor.feature_extractor,
    torch_dtype=torch_dtype,
    device=device,
)

# API endpoint to upload audio and get the transcribed text
@app.post("/transcribe")
async def transcribe_audio(file: UploadFile = File(...)):
    try:
        # Read the audio file bytes directly from the uploaded file
        audio_bytes = await file.read()

        # Pass the raw audio bytes to the pipeline
        result = pipe(audio_bytes)

        # Return the transcribed text
        return {"text": result["text"]}
    
    except Exception as e:
        raise HTTPException(status_code=500, detail=f"Error occurred: {str(e)}")

@app.get("/")
async def root():
    return {"message": "Welcome to the speech-to-text API!"}

# Running FastAPI with Uvicorn
if __name__ == "__main__":
    uvicorn.run(app, host="0.0.0.0", port=8000)