Passive acoustic monitoring (PAM) is a powerful tool for answering questions in both basic and applied ecology. However, it has its challenges. Here we focus on two questions to suggest what we should consider when making sound recordings: 1) what causes individual or species differences in detectability, and 2) what may be the effects of non-target sounds. One of the key aspects of a signal that determines detectability is the amplitude of the signal. Species or individuals with a loud signal can be detected at a long distance from the signaller, whereas a species or individuals with a soft signal cannot reach far and can be easily masked by noise. Star finches (Bathilda ruficauda) sing to attract mates, and the main part of their song is so soft that it is only audible to human ears at a distance of a few metres. We analysed their songs recorded in the laboratory and found that the acoustic structure of the song component affects the song amplitude. Interestingly, we found negative correlations between amplitude and several acoustic parameters related to song complexity, suggesting physical constraints on the song amplitude. Given the use of song in this species, the song amplitude may have evolved to increase the signal complexity at the expense of amplitude. For a long time, the amplitude of sound signals has been neglected due to technical difficulties, especially in soft signallers. We suggest that more attention be paid to this, which may make the results of acoustic monitoring programmes more informative for behavioural ecology and increase the applicability of the sound data for other PAM studies. Secondly, the most problematic effect of the non-target signal, both for the observer and the signaller, is the masking of the target signal. Animals have several strategies to avoid this. For instance, they vocalise louder in noisy conditions or shift their timing out of the noise. We recorded domestic canaries (Serinus canaria) under intermittent white noise conditions, expecting them to shift their singing activity out of the noise bursts. Surprisingly, however, they preferred to start singing during the noise bursts, in other words, the white noise triggered the birds to sing. Our results, together with findings from the current literature, indicate marked species differences in noise-induced song plasticity in birds. Overall, we need to be aware of the ecology of the target species when interpreting the meaning of the signalling behaviour.