Voices are recorded by microphones on a MD-recorder and a first tape machine. From this tape machine the tape runs to a second tape machine, which plays the tape after 20 sec. The sound is mixed to the input signal, i.e. the recordings are overlayed. At first this process is not audible, because the loudspeakers are not on. After 5 minutes the loudspeakers are switched on and the MD-recording is stopped. Now the MD is played and the sound is used as material for the tape loop. The MD-player makes possible very short repetitions (Zeitschleifen) and time leaps backwards and forwards. After a certain processing time the memory is erased. Now the MD-recording is played and overlayed in the memory, i.e. the process which was unaudible at the beginning is now repeated audiblely, but 20 minutes delayed.