diff --git a/apps/application/flow/step_node/speech_to_text_step_node/impl/base_speech_to_text_node.py b/apps/application/flow/step_node/speech_to_text_step_node/impl/base_speech_to_text_node.py
index b4806c537..1660fe1ef 100644
--- a/apps/application/flow/step_node/speech_to_text_step_node/impl/base_speech_to_text_node.py
+++ b/apps/application/flow/step_node/speech_to_text_step_node/impl/base_speech_to_text_node.py
@@ -37,7 +37,7 @@ class BaseSpeechToTextNode(ISpeechToTextNode):
                 temp_mp3_path = temp_amr_file.name
             any_to_mp3(temp_file_path, temp_mp3_path)
             try:
-                return split_and_transcribe(temp_file_path, model)
+                return split_and_transcribe(temp_mp3_path, model)
             finally:
                 os.remove(temp_file_path)
                 os.remove(temp_mp3_path)