main.py

import os
import torch
import librosa
import numpy as np
import tempfile
from fastapi import FastAPI, UploadFile, File, HTTPException
from transformers import Wav2Vec2Processor, Wav2Vec2ForCTC
from librosa.sequence import dtw
from contextlib import asynccontextmanager

os.environ["NUMBA_CACHE_DIR"] = "/tmp"  # Ensure Numba caching works in container environments


# --- Core Class Definition ---
class QuranRecitationComparer:
    def __init__(self, model_name="jonatasgrosman/wav2vec2-large-xlsr-53-arabic", auth_token=None):
        """
        Initialize the Quran recitation comparer with a specific Wav2Vec2 model.
        """
        self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
        
        if auth_token:
            self.processor = Wav2Vec2Processor.from_pretrained(model_name, token=auth_token)
            self.model = Wav2Vec2ForCTC.from_pretrained(model_name, token=auth_token)
        else:
            self.processor = Wav2Vec2Processor.from_pretrained(model_name)
            self.model = Wav2Vec2ForCTC.from_pretrained(model_name)
        
        self.model = self.model.to(self.device)
        self.model.eval()
        self.embedding_cache = {}

    def load_audio(self, file_path, target_sr=16000, trim_silence=True, normalize=True):
        if not os.path.exists(file_path):
            raise FileNotFoundError(f"Audio file not found: {file_path}")
        y, sr = librosa.load(file_path, sr=target_sr)
        if normalize:
            y = librosa.util.normalize(y)
        if trim_silence:
            y, _ = librosa.effects.trim(y, top_db=30)
        return y

    def get_deep_embedding(self, audio, sr=16000):
        input_values = self.processor(
            audio,
            sampling_rate=sr,
            return_tensors="pt"
        ).input_values.to(self.device)

        with torch.no_grad():
            outputs = self.model(input_values, output_hidden_states=True)

        hidden_states = outputs.hidden_states[-1]
        embedding_seq = hidden_states.squeeze(0).cpu().numpy()
        return embedding_seq

    def compute_dtw_distance(self, features1, features2):
        D, wp = dtw(X=features1, Y=features2, metric='euclidean')
        distance = D[-1, -1]
        normalized_distance = distance / len(wp)
        return normalized_distance

    def interpret_similarity(self, norm_distance):
        if norm_distance == 0:
            result = "The recitations are identical based on the deep embeddings."
            score = 100
        elif norm_distance < 1:
            result = "The recitations are extremely similar."
            score = 95
        elif norm_distance < 5:
            result = "The recitations are very similar with minor differences."
            score = 80
        elif norm_distance < 10:
            result = "The recitations show moderate similarity."
            score = 60
        elif norm_distance < 20:
            result = "The recitations show some noticeable differences."
            score = 40
        else:
            result = "The recitations are quite different."
            score = max(0, 100 - norm_distance)
        return result, score

    def get_embedding_for_file(self, file_path):
        if file_path in self.embedding_cache:
            return self.embedding_cache[file_path]
        audio = self.load_audio(file_path)
        embedding = self.get_deep_embedding(audio)
        self.embedding_cache[file_path] = embedding
        return embedding

    def predict(self, file_path1, file_path2):
        embedding1 = self.get_embedding_for_file(file_path1)
        embedding2 = self.get_embedding_for_file(file_path2)
        norm_distance = self.compute_dtw_distance(embedding1.T, embedding2.T)
        interpretation, similarity_score = self.interpret_similarity(norm_distance)
        print(f"Similarity Score: {similarity_score:.1f}/100")
        print(f"Interpretation: {interpretation}")
        return similarity_score, interpretation

    def clear_cache(self):
        self.embedding_cache = {}

# --- Lifespan Event Handler ---
@asynccontextmanager
async def lifespan(app: FastAPI):
    global comparer
    # Use environment variables or a secure configuration in production
    auth_token = os.environ.get("HF_TOKEN")
    comparer = QuranRecitationComparer(
        model_name="jonatasgrosman/wav2vec2-large-xlsr-53-arabic",
        auth_token=auth_token
    )
    print("Model initialized and ready for predictions!")
    yield
    print("Application shutdown: Cleanup if necessary.")

app = FastAPI(
    title="Quran Recitation Comparer API",
    description="Compares two Quran recitations using a deep wav2vec2 model.",
    version="1.0",
    lifespan=lifespan
)

# --- API Endpoints ---
@app.get("/", summary="Health Check")
async def root():
    return {"message": "Quran Recitation Comparer API is up and running."}

@app.post("/predict", summary="Compare Two Audio Files", response_model=dict)
async def predict(file1: UploadFile = File(...), file2: UploadFile = File(...)):
    tmp1_path = None
    tmp2_path = None
    try:
        suffix1 = os.path.splitext(file1.filename)[1] or ".wav"
        with tempfile.NamedTemporaryFile(delete=False, suffix=suffix1) as tmp1:
            content1 = await file1.read()
            tmp1.write(content1)
            tmp1_path = tmp1.name

        suffix2 = os.path.splitext(file2.filename)[1] or ".wav"
        with tempfile.NamedTemporaryFile(delete=False, suffix=suffix2) as tmp2:
            content2 = await file2.read()
            tmp2.write(content2)
            tmp2_path = tmp2.name

        similarity_score, interpretation = comparer.predict(tmp1_path, tmp2_path)
        return {"similarity_score": similarity_score, "interpretation": interpretation}
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))
    finally:
        if tmp1_path and os.path.exists(tmp1_path):
            os.remove(tmp1_path)
        if tmp2_path and os.path.exists(tmp2_path):
            os.remove(tmp2_path)

@app.post("/clear_cache", summary="Clear Embedding Cache", response_model=dict)
async def clear_cache():
    comparer.clear_cache()
    return {"message": "Cache cleared."}