Correcting　historical　corpora　in　digital　version　is　a　crucial　task　for　the　historical　research,　however,　scan　quality,　book　layout,　visual　character　similarity　can　affect　the　quality　of　the　recognizing.　OCR　is　at　the　forefront　of　digitization　projects　for　cultural　heritage　preservation.　The　main　task　is　to　identify　characters　from　their　visual　form　into　their　textual　representation.　In　this　paper,　we　propose　a　model　combining　recurrent　neutral　network(RNN)　and　deep　convolutional　network(DCNN)　to　correct　OCR　transcription　errors.　The　experiment　on　a　historical　book　corpus　in　German　language　shows　that　the　model　is　very　robust　in　capturing　diverse　OCR　transcription　errors　greatly.　©　Published　under　licence　by　IOP　Publishing　Ltd.