diff --git a/codes/models/audio/music/music_gen_fill_gaps.py b/codes/models/audio/music/music_gen_fill_gaps.py
index 7b53ca95..14eaf563 100644
--- a/codes/models/audio/music/music_gen_fill_gaps.py
+++ b/codes/models/audio/music/music_gen_fill_gaps.py
@@ -172,7 +172,7 @@ class MusicGenerator(nn.Module):
     def do_masking(self, truth):
         b, c, s = truth.shape
         mask = torch.ones_like(truth)
-        if self.frequency_mask_percent_mask > 0:
+        if self.random() > .5:
             # Frequency mask
             cs = random.randint(0, c-10)
             ce = min(c-1, cs+random.randint(1, int(self.frequency_mask_percent_mask*c)))