fix

training/eval.py

@@ -50,7 +50,7 @@ def wer(asr_model_name_or_path, prompts, audios, device, per_device_eval_batch_s
         normalizer = english_normalizer if hasattr(pred, "language") and pred["language"] == "english" else basic_normalizer
         norm_ref = normalizer(ref)
         if len(norm_ref) > 0:
-            norm_pred = normalizer(pred)
+            norm_pred = normalizer(pred["text"])
             normalized_predictions.append(norm_pred)
             normalized_references.append(norm_pred)