update repo

egs/wenetspeech/conformer/conf/train_asr_conformer.yaml

@@ -90,7 +90,7 @@ specaug_conf:
 
 dataset_conf:
     data_names: speech,text
-    data_types: sound,text
+    data_types: sound,text_nospace
     shuffle: True
     shuffle_conf:
         shuffle_size: 2048

funasr/datasets/large_datasets/dataset.py

@@ -148,6 +148,12 @@ class AudioDataset(IterableDataset):
                         if "key" not in sample_dict:
                             sample_dict["key"] = segs[0]
                         sample_dict['hw_tag'] = 1
+                    elif data_type == "text_nospace":
+                        text = item
+                        segs = text.strip().split(maxsplit=1)
+                        sample_dict[data_name] = [x for x in segs[1]]
+                        if "key" not in sample_dict:
+                            sample_dict["key"] = segs[0]
                     else:
                         text = item
                         segs = text.strip().split()