Update app.py

app.py

@@ -90,11 +90,13 @@ def generate(input, instruction=DEFAULT_INPUT, sampling=False, temperature=0.7,
     elif filetype == "Video":
         frames = encode_video(input)
         content.extend(frames)
-        audio, _ = librosa.load(input, sr=16000, mono=True)
-        content.append(audio)
+        audio_np, sample_rate = librosa.load(input, sr=16000, mono=True)
+        audio_tensor = torch.from_numpy(audio_np).float().to(DEVICE)
+        content.append({"array": audio_tensor, "sampling_rate": sample_rate})
     elif filetype == "Audio":
-        audio, _ = librosa.load(input, sr=16000, mono=True)
-        content.append(audio)
+        audio_np, sample_rate = librosa.load(input, sr=16000, mono=True)
+        audio_tensor = torch.from_numpy(audio_np).float().to(DEVICE)
+        content.append({"array": audio_tensor, "sampling_rate": sample_rate})
     else:
         return "Unsupported file type."