Skip to content

Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.

update raw_text

shixian.shi0ee4565c8ec3
diff --git a/funasr/auto/auto_model.py b/funasr/auto/auto_model.py
index 0641f06a8..1c7288e26 100644
--- a/funasr/auto/auto_model.py
+++ b/funasr/auto/auto_model.py
@@ -406,7 +406,7 @@ class AutoModel:
for res, vadsegment in zip(restored_data, vadsegments):
sentence_list.append({"start": vadsegment[0],\
"end": vadsegment[1],
- "sentence": res['text'],
+ "sentence": res['raw_text'],
"timestamp": res['timestamp']})
else: # punc_segment
sentence_list = timestamp_sentence(punc_res[0]['punc_array'], \
diff --git a/funasr/models/seaco_paraformer/model.py b/funasr/models/seaco_paraformer/model.py
index 1867bbfef..e3cedb474 100644
--- a/funasr/models/seaco_paraformer/model.py
+++ b/funasr/models/seaco_paraformer/model.py
@@ -420,7 +420,7 @@ class SeacoParaformer(BiCifParaformer, Paraformer):
if ibest_writer is not None:
ibest_writer["token"][key[i]] = " ".join(token)
- # ibest_writer["text"][key[i]] = text
+ ibest_writer["raw_text"][key[i]] = text
ibest_writer["timestamp"][key[i]] = time_stamp_postprocessed
ibest_writer["text"][key[i]] = text_postprocessed
else: