Several researchers have identified the need for sound event recognition in soundscape analysis. Especially in attention and restoration research not only the sound level is important, but also the sound event. Recognition techniques suitable for this task face some extra challenges compared to existing techniques. The important differences between operation environments of current techniques and the environments of soundscape research are the lack of control over the environment and the open character. We will discuss the demands that preprocessing and recognition techniques have to adhere to. These demands include dealing with changing transmission effects, with unseen sources and with multiple concurrent sources.