My study area covers a large part of the slag heap of Buštěhrad and has an area of approximately 0.72 square km. In 2016, I visited my study area 17 times between March 18 and August 2. I only started to visit the most western and northern parts of the study area from May 11 onwards. During my visits, I record all singing Yellowhammers using my photo camera, a Panasonic Lumix DMC-FZ72. This camera has a video function with high quality sound recording. I record video in MP4 format. At home, using my laptop, I extract the sound in MP3 format using the program ffmpeg. I then use the program Audacity to produce sonograms of my recordings. A disadvantage of this program is that it does not have a function for plotting sonograms, so when I need to save a sonogram I use the drawing program Gimp to make a screen shot and then cut out my sonograms. Another disadvantage of Audacity is that the available zoom options depend on the total length of the recording. Initially, this had the result that the time scale of my sonograms varied, which made it hard to compare them. I now zoom all my sonograms to a standard scale where 220 pixels are one second (horizontally) while the vertical height of 100 pixels shows frequencies between 2 and 10 kHz, linearly scaled.

Often, two or more birds can be heard on the same recording. In such cases, I have learned to speak into the recording by saying "left", "behind" or such things to help me distinguish the different birds at home. Initially, my recordings were usually about 1 minute long but I have learned to record at least 3 minutes since some birds are quite slow to change between the different songtypes of their repertoire. At home, I used my memory and the visual picture of the video recording to determine the location of all birds. In future, I plan to take a map with me to the field to record the locations in a more precise and reliable way.

At home, viewing the sonograms of my recordings, I classify all songtypes (ways of singing the opening phrase) by comparing them with my database of previously recorded songtypes. Since songtypes consist of two elements, it turns out to be easier to first look up each element separately in my database of previously recorded elements and then look for known songtypes that contain these elements, or elements that are similar to it. It is important to look carefully at the exact frequency at which an element is sung. Even though there is a bit of variation in how a bird sings, a difference of more than 1 kHz between two otherwise similar looking elements is often significant.

In 2016, after each visit, I made a map which contains the songtypes at the locations where they were recorded. Comparing maps of different visits, it soon becames clear that birds with the same repertoire tend to be recorded at the same locations, or very close to the previous location. Using this sort of information, checking details of sonograms if needed, I gave names to birds that were clearly recorded during more than one visit. At the end of the season, I critically reexamined my assignment of songtypes to birds and managed to attribute some problematic recordings to known or new birds. At present, I can clearly attribute most of the song I have recorded in 2016 to 29 named birds. Of my named birds, I also tried to record the dialect sung by it (if any). This has not been done completely systematically this year, but I plan to improve myself in this respect during the next season.

The method is not without problems. Birds may stop singing or switch to "dzipping" before their whole repertoire has been recorded. I have two birds (Cornelis and Ubele) who each have a repertoire of three songtypes, of which two are the same. Various noises can also have a bad effect on the quality of the recordings. In particular, wind playing in the leaves of trees (rattling poplar trees!) but also song of other birds (those annoying nightingales and grasshopper warblers!) can be problematic. Human sounds are also present, but less of a problem, except near the water treatment plant on the northern edge of the study area. Some birds apparently sing very little. Some songtypes are very common; if no other songtype is simultaneously recorded of the same bird, it may be impossible to identify it.

In view of these and other problems, a part of the recordings cannot be identified. At present, for 2016, I am left with 21 problematic recordings, often of poor quality, cannot with reasonable certainty be assigned to any of the named birds. Sometimes, these birds sang only one songtype which is shared by several birds, or the quality of the recording is so poor that the songtype cannot be recognized. A few are clearly not from any known bird, but are too vague to allow me to give them a new name.

In addition, since I have no way of telling birds apart apart from their recordings, I have no guarantee that what I think is one bird is not really two birds, or that what I think is two birds is really one bird that has changed its repertoire. At present, Tereza Petrusková and Sharina van Boheemen of Charles University are running a project on individual recognition of Yellowhammers that is much more professional than my study; in particular, they ring birds to be able to recognize them visually. Nevertheless, by and large, my simple method seems to yield sensible results that certainly give much more information than can be obtained by listening only.