diff --git a/export_meta.py b/export_meta.py
index 920ca64..aa6a661 100644
--- a/export_meta.py
+++ b/export_meta.py
@@ -33,14 +33,14 @@ def export_forward(
     textnorm_query = self.embed(textnorm.to(speech.device)).unsqueeze(1)
     print(textnorm_query.shape, speech.shape)
     speech = torch.cat((textnorm_query, speech), dim=1)
-    speech_lengths += 1
+    speech_lengths = speech_lengths + 1
     
     event_emo_query = self.embed(torch.LongTensor([[1, 2]]).to(speech.device)).repeat(
         speech.size(0), 1, 1
     )
     input_query = torch.cat((language_query, event_emo_query), dim=1)
     speech = torch.cat((input_query, speech), dim=1)
-    speech_lengths += 3
+    speech_lengths = speech_lengths + 3
     
     encoder_out, encoder_out_lens = self.encoder(speech, speech_lengths)
     if isinstance(encoder_out, tuple):