diff --git a/color_syncnet_train.py b/color_syncnet_train.py
index c3b99f2..8ae099e 100644
--- a/color_syncnet_train.py
+++ b/color_syncnet_train.py
@@ -55,7 +55,7 @@ class Dataset(object):
 
     def crop_audio_window(self, spec, start_frame):
         # num_frames = (T x hop_size * fps) / sample_rate
-        start_frame_num = self.get_frame_id(start_frame) + 1 # 0-indexing ---> 1-indexing
+        start_frame_num = self.get_frame_id(start_frame)
         start_idx = int(80. * (start_frame_num / float(hparams.fps)))
 
         end_idx = start_idx + syncnet_mel_step_size
@@ -278,4 +278,4 @@ if __name__ == "__main__":
     train(device, model, train_data_loader, test_data_loader, optimizer,
           checkpoint_dir=checkpoint_dir,
           checkpoint_interval=hparams.syncnet_checkpoint_interval,
-          nepochs=hparams.nepochs)
\ No newline at end of file
+          nepochs=hparams.nepochs)