fix bug of pad_sequence in u2,test=asr (#1153)

paddlespeech/s2t/models/u2/u2.py

@@ -579,10 +579,15 @@ class U2BaseModel(ASRInterface, nn.Layer):
             num_decoding_left_chunks, simulate_streaming)
         assert len(hyps) == beam_size
 
-        hyps_pad = pad_sequence([
-            paddle.to_tensor(hyp[0], place=device, dtype=paddle.long)
-            for hyp in hyps
-        ], True, self.ignore_id)  # (beam_size, max_hyps_len)
+        hyp_list = []
+        for hyp in hyps:
+            hyp_content = hyp[0]
+            # Prevent the hyp is empty
+            if len(hyp_content) == 0:
+                hyp_content = (self.ctc.blank_id,)
+            hyp_content = paddle.to_tensor(hyp_content, place=device, dtype=paddle.long)
+            hyp_list.append(hyp_content)
+        hyps_pad = pad_sequence(hyp_list, True, self.ignore_id)
         hyps_lens = paddle.to_tensor(
             [len(hyp[0]) for hyp in hyps], place=device,
             dtype=paddle.long)  # (beam_size,)