diff --git a/paddlespeech/s2t/models/whisper/whisper.py b/paddlespeech/s2t/models/whisper/whisper.py
index d20cc04b6..fdd3a6974 100644
--- a/paddlespeech/s2t/models/whisper/whisper.py
+++ b/paddlespeech/s2t/models/whisper/whisper.py
@@ -835,8 +835,14 @@ class BeamSearchDecoder(TokenDecoder):
                 logprob, token = paddle.topk(
                     logprobs[idx], k=self.beam_size + 1)
                 for logprob, token in zip(logprob, token):
-                    new_logprob = (sum_logprobs[idx] + logprob).tolist()[0]
-                    sequence = tuple(prefix + [token.tolist()[0]])
+                    # after Paddle 3.0, tolist in 0-D tensor will return a float/int value instead of a list
+                    new_logprob = (sum_logprobs[idx] + logprob).tolist()
+                    new_logprob = new_logprob if isinstance(
+                        new_logprob, float) else new_logprob[0]
+                    new_token = token.tolist()
+                    new_token = new_token if isinstance(new_token,
+                                                        int) else new_token[0]
+                    sequence = tuple(prefix + [new_token])
                     scores[sequence] = new_logprob
                     sources[sequence] = idx