This master thesis is part of the PASSAGE project, which aims to develop a system to automatically generate Standard German subtitles for spoken Swiss German in collaboration with the « Schweizer Radio and Fernsehen » (SRF). For this task, a normalized transcription from speech recognition needs to be post-edited for the training of an automatic post-editing system. The goal of the thesis is to develop and evaluate a rule-based automatic post-editing system, which is intended to assist post-editors in generating corrected segments. The system exploits regular expressions and rules using the natural language processing library spaCy. The developed rules cover errors due to spoken language and specific Swiss German phenomenons. The system detected errors (performed modifications) in approx. 5% (4%) of the normalized transcription input. A human evaluation confirmed the necessity (correctness) of the modifications in 97% (88%) of the cases.