Fix flat audio inputs

app.py

@@ -30,10 +30,11 @@ def clone(audio, *args):
     if wav.dtype == torch.int16:
         wav = wav.float() / 32767.0
 
-    if wav.shape[0] == 2:  # Stereo to mono if needed
-        wav = wav.mean(0, keepdim=True)
-    if wav.shape[1] == 2:
-        wav = wav.mean(1, keepdim=False).unsqueeze(-1)
+    if len(wav.shape) == 2:
+        if wav.shape[0] == 2:  # Stereo to mono if needed
+            wav = wav.mean(0, keepdim=True)
+        if wav.shape[1] == 2:
+            wav = wav.mean(1, keepdim=False).unsqueeze(-1)
 
     wav = wav[-int(sr*20):]  # Take only the last 20 seconds