A transcription system including a transcription device for the automatic transcription of dictated material and additionally employs transcribers who manually transcribe some of the dictated material. The transcription device contains an abortion unit for aborting processing of certain dictated material by the transcription device and retransmitting this dictated material to one of the transcribers for manual transcription if the quality of the dictated audio data (AI) or the quality of the automatically transcribed text data (ATTI) is so poor that the correction of recognition errors in the automatically transcribed text data (ATTI) requires more effort than if said dictated material is transcribed manually from the outset.