Update model.py

model.py

@@ -1,56 +1,20 @@
-"""
-🔥 TEST YOUR AUDIO MODEL FROM HUGGING FACE
-Simma7/audio_model
-"""
-
+from transformers import AutoModelForAudioClassification, AutoProcessor
 import torch
-import librosa
-from transformers import AutoProcessor, AutoModel
-from huggingface_hub import snapshot_download
-
-DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
-
-# ================= DOWNLOAD MODEL =================
-print("⬇️ Downloading model from Hugging Face...")
-
-model_dir = snapshot_download("Simma7/audio_model")
-
-print("✅ Download complete")
-
-# ================= LOAD MODEL =================
-print("🧠 Loading model...")
-
-processor = AutoProcessor.from_pretrained(model_dir)
-model = AutoModel.from_pretrained(model_dir).to(DEVICE)
-
-print("✅ Model loaded")
-
-# ================= PREDICT =================
-def predict(audio_path):
-    audio, sr = librosa.load(audio_path, sr=16000)
-
-    inputs = processor(audio, sampling_rate=16000, return_tensors="pt", padding=True)
-
-    with torch.no_grad():
-        outputs = model(**inputs.to(DEVICE))
-
-    # embedding-based score
-    embedding = outputs.last_hidden_state.mean(dim=1)
-    prob = torch.sigmoid(embedding.mean()).item()
 
-    return prob
+model_name = "Simma7/audio_model"
 
-# ================= MAIN =================
-if __name__ == "__main__":
-    audio_path = "test.wav"   # 🔥 put your audio file
+print("[INFO] Loading model...")
 
-    print("\n🔍 Analyzing audio...")
+model = AutoModelForAudioClassification.from_pretrained(
+    model_name,
+    trust_remote_code=True
+)
 
-    prob = predict(audio_path)
+processor = AutoProcessor.from_pretrained(
+    model_name,
+    trust_remote_code=True
+)
 
-    if prob > 0.5:
-        print("\n🔴 FAKE AUDIO")
-    else:
-        print("\n🟢 REAL AUDIO")
+model.eval()
 
-    print(f"📊 Confidence: {prob:.4f}")
+print("[SUCCESS] Model loaded successfully ✅")