They sampled sound pressure levels directly from recording for one second before the randomly selected song and looked at the dB when the song was present and when it was not present. They looked at the difference in pressure to assess whether the noise had an effect on song structure. To look at the relationship between vegetation and structure on song structure, they calculated the proportion of area around the males that contained structures of vegetation. The structure variables contain the percent of vegetation and reflective surfaces of concentric circles extending from the location of the bird.
They found that noise levels varied considerably with the proportion of urban and vegetative structure, but noise levels were not correlated with urban structure. They found that males from group one produced significantly higher maximum frequencies, broader bandwidths, and significantly slower trill rates than the group 2 males. Minimum frequency and bandwidth were the only traits that showed group specific responses to structure and noise. In males from both groups, peak frequency decreased, and the time between songs increased as urban structure increased. In general, vocal performance declined across all males with increasing noise and increasing urban structure. Peak frequency decreased with increasing urban structure; males put more energy into low frequency songs so it is more expensive to create low frequencies. This shows that unless it is beneficial to produce low frequency songs, like in the case of increased urban structure, males will produce higher frequencies because it requires less energy.