Fig 1.
Exemplary spectrograms demonstrating the measured temporal and spectral parameters of the song duration, maximum frequency (Max Freq) and minimum frequency (Min Freq) in nightingale song for a) whistle songs, b) trill songs and c) buzz songs.
Fig 2.
Process of generating recordings of CS (left) and EX (right).
The similarities and differences in previous experience, recording equipment, song recording, song recording duration, data transfer and verification of recordings by experts are pointed out (from top to bottom).
Fig 3.
Temporal distributions of 5,679 citizen science recordings (abbreviated as rec) with the ’Naturblick’ app for 2019.
In red, the times of previous six nightingale studies are shown (2004–2017; [61–64, 79, 80]). a) Representation of the number of recordings in relation to the time of day. b) Presentation of the weekly number of recordings in the course of the breeding season. https://doi.org/10.5281/zenodo.4817236.
Fig 4.
Comparison of defined recording categories for 2019, based on a) 5,679 citizen science recordings (CS) with smartphones via the ’Naturblick’ app and b) six expert recordings (EX) with high-quality microphones. The categories are displayed in different colours (red: one song type was in its entirety recognizable by syllables and elements in the spectrograms, orange: some syllables or elements were not clearly shown in the spectrograms, light blue: other bird species and dark blue: no birds). https://doi.org/10.5281/zenodo.4817236.
Fig 5.
Comparison of the number of citizen science recordings (rec) in the year 2019, based on recordings with the ’Naturblick’ app of 245 one-time users, 361 frequent users and 18 power users (one-time users = 1 recording; multiple users = 2–19 recordings; frequent users ≥ 20 recordings).
In red: one song type was in its entirety to be identified by syllables and elements in the spectrograms, in orange: some syllables or elements were not clearly shown in the spectrograms, in light blue: other bird species and in dark blue: no birds. a) The number of recordings. b) The duration of recordings shown in the boxplots. The median is represented by a solid black line and the mean by a dashed black line within a box. The borders of boxes are 25 and 75 percentiles. The bars above box plots indicate significant differences between two stimulus categories. c) Shows the cumulative number of songs in all user groups for the first category. https://doi.org/10.5281/zenodo.4817236.
Fig 6.
Comparison of citizen science (CS) recordings with the ’Naturblick’ app and expert recordings with equipment using professional microphones (EX) in the year 2018 and 2019.
a) Spectrograms of CS recordings (top) compared to EX recordings (below) for the one example per song categories (whistle, trill and buzz). b) Comparison the signal to noise ratio (SNR) as boxplots for three of these selected song types per each category: whistle (left), trill (middle) and buzz (right). The median of the boxplot is represented by a solid black line and the mean by a dashed black line within a box. The borders of boxes are 25 and 75 percentiles. The bars above box plots indicate significant differences between two stimulus categories. SNRs over 10 dB (red line) were defined in this study as valid quality according to Fitzpatrick and colleagues [58]. The red dotted line indicates a value of 5 dB, a threshold used by Barmatz and colleagues [57, 58] forvalid quality. The black lines indicate significant differences. All recordings were examined and displayed under the same settings (sample rate = 22,050 Hz, FFT = 1024 points, Hamming-Window, overlap 93,75%). https://doi.org/10.5281/zenodo.4817236.
Fig 7.
Comparison of deviations from original playback files of the different recording devices, i.e. smartphone brands (CS) and expert recordings equipment with professional microphones (EX) in a test.
Depicted in boxplots is the deviation in a) the minimum frequency, b) the maximum frequency and c) song duration from playing back a test file of nightingale song. The median is represented by a solid black line within a box. The borders of boxes are 25 and 75 percentiles. The red line shows a zero line. The closer a deviation to the zero line is, the smaller it was. The black lines above boxplots indicate significant differences between the recording devices tested.
Fig 8.
Comparison of citizen science (CS) recordings with the ’Naturblick’ app and expert recordings with equipment using professional microphones (EX) during a playback test.
Spectrograms of CS recordings (top) compared to EX recordings (below) for the one example per song categories (whistle, trill and buzz).