The soft (low amplitude) song is a poorly understood acoustic communication phenomena despite the fact that it seems to be common and widespread in birds [1, 2]. Little is known because it has a very short range of action, which makes observation and recording difficult. Several hypotheses were raised in order to explain soft song functionality [3-6], but there is no study where all of them were tested on a single species and the topic is under hot debate [7-8, 2]. We present results from a project design to test all working hypotheses on soft song functionality in the ortolan bunting (Emberiza hortulana), a small migratory passerine species from Eurasia, which males use soft song during breeding season. The first experiment was designed to test hypothesis that males use soft song during territorial intrusions to avoid being predated [3]. We used microphone array allowing for amplitude measurements in the field and simulated territorial intrusion into ortolan bunting territories by a stranger male (playback + model presented) in a paired design experiment (N=27 males tested twice). In random order, we presented (treatment) or did not presented (control) a predator (Falco tinnuculus) and its calls before simulated intrusion of a non-neighbour male. We found that ortolan bunting males responded to conspecific playback in a typical way by approaching and producing songs and calls. In 41% of experimental treatments and in 44% of controls the responding males produced also soft songs. However, we found no significant relationship between presence of a predator and response of males to playback. In particular, we found no significant difference in producing soft songs between treatment and control experiments. However, the presence of other conspecific males in vicinity affected response. We found no support for the predator avoidance hypothesis. (Financial Support: National Science Center, Poland, grant no. 2013/09/B/NZ8/03275).