The invention concerns a method for creating a summary of an audio-visual document while observing both user-defined criteria and criteria assisting understanding of said summary. The document is first edited into sequences, each associated with attribute values and a dependence parameter. The user firstly selects first sequences either manually, or based on their attribute values. The second sequences are then added based on the value of the dependence parameter. Said added sequences enable the thus created summary to be better understood. In accordance with an improvement, the user introduces a maximum duration for the summary, first and/or second sequences are removed from the summary if the duration of the resulting summary exceeds the duration introduced. The invention also concerns an audio-visual apparatus for creating and viewing the summary.