Fix import error by using sam-audio library

app.py

@@ -10,7 +10,19 @@ except ImportError:
                 return func
             return decorator
 
-from transformers import AutoProcessor, AutoModelForAudioSegmentation
+import gradio as gr
+import torch
+try:
+    import spaces
+except ImportError:
+    class spaces:
+        @staticmethod
+        def GPU(duration=60):
+            def decorator(func):
+                return func
+            return decorator
+
+from sam_audio import SAMAudio, SAMAudioProcessor
 import numpy as np
 import librosa
 import tempfile
@@ -21,8 +33,8 @@ MODEL_ID = "facebook/sam-audio-small"
 
 print(f"Loading model: {MODEL_ID}...")
 try:
-    processor = AutoProcessor.from_pretrained(MODEL_ID)
-    model = AutoModelForAudioSegmentation.from_pretrained(
+    processor = SAMAudioProcessor.from_pretrained(MODEL_ID)
+    model = SAMAudio.from_pretrained(
         MODEL_ID,
         device_map="auto",
         torch_dtype=torch.float16
@@ -30,17 +42,13 @@ try:
     print("Model loaded successfully.")
 except Exception as e:
     print(f"Error loading model: {e}")
-    print("Attempting to load with 8-bit quantization...")
+    # Fallback currently not fully supported for SAMAudio custom class in bitsandbytes via config directly 
+    # unless it inherits correctly. Let's try standard float32 if float16 fails, or keep the error.
+    print("Retrying with default precision...")
     try:
-        from transformers import BitsAndBytesConfig
-        quantization_config = BitsAndBytesConfig(load_in_8bit=True)
-        processor = AutoProcessor.from_pretrained(MODEL_ID)
-        model = AutoModelForAudioSegmentation.from_pretrained(
-            MODEL_ID,
-            quantization_config=quantization_config,
-            device_map="auto"
-        )
-        print("Model loaded with 8-bit quantization.")
+        processor = SAMAudioProcessor.from_pretrained(MODEL_ID)
+        model = SAMAudio.from_pretrained(MODEL_ID, device_map="auto")
+        print("Model loaded with default precision.")
     except Exception as e2:
         print(f"Critical error loading model: {e2}")
         raise e2

requirements.txt

@@ -4,8 +4,12 @@ transformers>=4.38.0
 accelerate>=0.27.0
 bitsandbytes>=0.41.0
 scipy
+scipy
 librosa
 opencv-python-headless
 spaces
+git+https://github.com/facebookresearch/sam-audio.git
+torchaudio
+