Update app.py

app.py

@@ -1,53 +1,79 @@
-from transformers import pipeline
-import torch
+import os
+import subprocess
 import soundfile as sf
+import librosa
+from transformers import pipeline
 from datetime import datetime
-import requests
 
-# Initialize the classifier pipeline
+# Initialize Hugging Face pipeline
 classifier = pipeline(
     "audio-classification",
     model="padmalcom/wav2vec2-large-nonverbalvocalization-classification",
 )
 
+def convert_audio(input_path, output_path="input.wav"):
+    """Convert audio to 16kHz mono WAV using ffmpeg."""
+    if not os.path.isfile(input_path):
+        raise FileNotFoundError(f"File not found: {input_path}")
+    
+    cmd = [
+        "ffmpeg", "-i", input_path,
+        "-acodec", "pcm_s16le",
+        "-ar", "16000",
+        "-ac", "1",
+        output_path,
+        "-y"
+    ]
+
+    try:
+        subprocess.run(cmd, check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+        return output_path
+    except subprocess.CalledProcessError as e:
+        raise RuntimeError("FFmpeg conversion failed: " + e.stderr.decode())
+
 def detect_scream(audio_path: str):
+    """Run scream detection on a WAV file."""
+    if not os.path.isfile(audio_path):
+        raise FileNotFoundError(f"Audio file not found: {audio_path}")
+
     audio, sr = sf.read(audio_path)
-    # Resample to expected sampling rate if needed
+    # Resample if needed
     if sr != classifier.feature_extractor.sampling_rate:
-        import librosa
         audio = librosa.resample(audio, orig_sr=sr, target_sr=classifier.feature_extractor.sampling_rate)
+
     results = classifier(audio)
     top = results[0]
     label = top["label"]
-    score = float(top["score"]) * 100  # as percentage
-    # Map to your Detection_Result and Alert_Level
+    score = float(top["score"]) * 100
+
     if label.lower() == "scream" and score > 80:
         alert = "High-Risk"
     elif label.lower() == "scream" and score > 50:
         alert = "Medium-Risk"
     else:
         alert = "None"
+
     return label, score, alert
 
-def log_to_salesforce(sf_instance, audio_url, label, score, alert):
-    # Create Scream_Detection__c record
-    sf_instance.Scream_Detection__c.create({
-        "Audio_File_URL__c": audio_url,
-        "Detection_Result__c": label,
-        "Confidence_Score__c": score,
-        "Alert_Level__c": alert,
-        "Timestamp__c": datetime.utcnow().isoformat(),
-        # add User__c if available
-    })
-    # trigger Salesforce alert automation (email/SMS/in‑app)
+def log_to_salesforce(audio_url, label, score, alert):
+    """Placeholder for Salesforce integration."""
+    print("Logging to Salesforce...")
+    print(f"Audio URL: {audio_url}")
+    print(f"Result: {label}, Score: {score:.1f}%, Alert Level: {alert}")
+    # Integration with Salesforce via simple-salesforce or REST API goes here
 
 def main():
-    audio_path = "input.wav"
-    audio_url = "https://my.blob/storage/input.wav"
-    label, score, alert = detect_scream(audio_path)
-    print(f"Detected: {label}, {score:.1f}%, Level: {alert}")
-    # Optional: Push to Salesforce using simple-salesforce, requests, etc.
-    # log_to_salesforce(sf, audio_url, label, score, alert)
+    # Input file path (any audio format)
+    raw_input_path = "my_audio.mp3"  # Change to your input file
+    audio_url = "https://yourstorage.com/path/to/audio"  # Simulated URL
+
+    try:
+        wav_path = convert_audio(raw_input_path)
+        label, score, alert = detect_scream(wav_path)
+        print(f"Detected: {label} ({score:.1f}%) — Alert Level: {alert}")
+        log_to_salesforce(audio_url, label, score, alert)
+    except Exception as e:
+        print("Error:", str(e))
 
 if __name__ == "__main__":
     main()