from fastapi import FastAPI, UploadFile, File
from fastapi.middleware.cors import CORSMiddleware
import torchaudio
import torch
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
import uvicorn
import io

app = FastAPI()

# Allow requests from Flutter (localhost or any domain)
app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

# Load model and processor once at startup
processor = Wav2Vec2Processor.from_pretrained("Mustafaa4a/ASR-Somali")
model = Wav2Vec2ForCTC.from_pretrained("Mustafaa4a/ASR-Somali")

@app.post("/transcribe")
async def transcribe_audio(file: UploadFile = File(...)):
    contents = await file.read()
    audio_bytes = io.BytesIO(contents)

    waveform, sample_rate = torchaudio.load(audio_bytes)

    if sample_rate != 16000:
        resampler = torchaudio.transforms.Resample(orig_freq=sample_rate, new_freq=16000)
        waveform = resampler(waveform)

    inputs = processor(waveform.squeeze(), sampling_rate=16000, return_tensors="pt")
    with torch.no_grad():
        logits = model(**inputs).logits

    predicted_ids = torch.argmax(logits, dim=-1)
    transcription = processor.decode(predicted_ids[0])
    return {"transcription": transcription}