diff --git a/TTS/tts/layers/xtts/trainer/dataset.py b/TTS/tts/layers/xtts/trainer/dataset.py
index 41401fd6f8..e0fa671abc 100644
--- a/TTS/tts/layers/xtts/trainer/dataset.py
+++ b/TTS/tts/layers/xtts/trainer/dataset.py
@@ -109,7 +109,7 @@ def check_eval_samples(self):
             try:
                 tseq, _, wav, _, _, _ = self.load_item(sample)
             except:
-                pass
+                continue
             # Basically, this audio file is nonexistent or too long to be supported by the dataset.
             if (
                 wav is None