app.py

from flask import Flask, request, jsonify
import torch
import librosa
import numpy as np
from transformers import AutoModelForAudioClassification, Wav2Vec2FeatureExtractor
import os

app = Flask(__name__)

# Model setup
model_name = 'amiriparian/ExHuBERT'
feature_extractor = Wav2Vec2FeatureExtractor.from_pretrained("facebook/hubert-base-ls960")
model = AutoModelForAudioClassification.from_pretrained(model_name, trust_remote_code=True)
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model = model.to(device)

# Labels for emotion mapping
labels = ['disgust', 'neutral', 'kind', 'anger', 'surprise', 'joy']

@app.route('/detect_scream', methods=['POST'])
def detect_scream():
    try:
        # Check if audio file is provided
        if 'file' not in request.files:
            return jsonify({'error': 'No audio file provided'}), 400
        
        audio_file = request.files['file']
        
        # Validate file type
        if not audio_file.filename.endswith(('.wav', '.mp3')):
            return jsonify({'error': 'Unsupported file format. Use WAV or MP3'}), 400
        
        # Save audio file temporarily
        temp_path = f"/tmp/{audio_file.filename}"
        audio_file.save(temp_path)
        
        # Load and preprocess audio
        waveform, sr = librosa.load(temp_path, sr=16000)
        inputs = feature_extractor(
            waveform,
            sampling_rate=16000,
            padding="max_length",
            max_length=48000,
            return_tensors="pt"
        )
        inputs = inputs['input_values'].to(device)
        
        # Perform inference
        with torch.no_grad():
            outputs = model(inputs).logits
            probabilities = torch.nn.functional.softmax(outputs, dim=1)
            confidence, predicted = torch.max(probabilities, 1)
        
        # Get result
        result = {
            'label': labels[predicted.item()],
            'confidence': float(confidence.item()),
            'alert_level': 'High-Risk' if confidence.item() > 0.8 else ('Medium-Risk' if confidence.item() > 0.5 else 'None')
        }
        
        # Clean up temporary file
        os.remove(temp_path)
        
        return jsonify(result), 200
    
    except Exception as e:
        return jsonify({'error': str(e)}), 500

@app.route('/health', methods=['GET'])
def health_check():
    return jsonify({'status': 'healthy'}), 200

if __name__ == '__main__':
    app.run(host='0.0.0.0', port=7860)