Prepare audio recordings
Contributions to the Macaulay Library should be an accurate copy of an original field recording. This means doing a minimal amount of editing to sound files. Follow these steps to prepare your sound recordings before uploading them to an eBird checklist.
These general guidelines are applicable to any sound editing program, but four popular editing programs are Adobe Audition, Ocenaudio, Audacity, and WavePad. Step-by-step instructions for editing in these programs can be found lower on the page. Sound analysis programs, such as Raven, should not be used for sound editing since they are not built for editing and preserving sound files.
Steps for preparing sound files
Many sound editing programs are “destructive,” meaning that if you edit a sound file in the program, your original sound file is permanently changed. Therefore, it is extremely important to keep copies of your original field recordings that are not modified in any editing program. The Macaulay Library recommends keeping two copies of your original sound files—for example, one on your computer, and a second on an external hard drive—and using a third copy of the files for editing and submitting to the archive. We also recommend that you save the final edited versions of your files for future reference.
WAV is the standard audio format used at the Macaulay Library, the Library of Congress, and other sound archives dedicated to the long-term preservation of audio. WAV is an uncompressed audio format that provides an accurate copy of wildlife sounds. By recording in the WAV format, you will maximize the usefulness of your recordings for research and conservation both today and in the future. The eBird upload tool can handle large audio files—up to 250 MB in size!—so if your original field recordings are .WAV sound files, be sure to upload them as .WAV files (not compressed .mp3 files) to your eBird checklist. If you are not currently recording in an uncompressed format like .WAV, consider switching. If your recorder supports it, a sample rate of at least 48 kHz and a bit depth of 24 bits is also recommended.
The beginning of a sound file is often the messiest part of the recording. Recordists sometimes make noise as they start a recording and get situated, and then there can be a significant amount of waiting time before a bird actually sings. When reviewing your sound file, try to leave a lead-in of up to 3 seconds of background sound before the first target sound. If there is handling noise or a loud background species during these 3 seconds, shorten your lead-in to remove the unwanted noise. At the end of the recording, after the last target sound, apply the same approach of including up to 3 seconds of background sound before the end if it is free of louder sounds. If you do not have much background audio to work with, that’s fine too–don’t add blank audio or manufacture background sound.
Even when you are close to a bird and have made a great recording, chances are that the bird sound could still benefit from being boosted a few decibels. We ask that you adjust the level of each recording so that the loudest sound reaches -3 dB, otherwise known as “normalizing.” This makes it easier to hear the target sound in a recording and creates a consistent listening experience between recordings across the archive. It also makes it easier to assess the quality of a recording. Leaving a recording at a low volume does not reduce background noise, it only makes the entire recording harder to hear.
Human voice announcements, on the other hand, tend to sound quite loud compared to bird song, so we suggest keeping the peak level for those announcements at -10 dB. Editing programs make it easy to adjust level of different segments, usually using a “Normalize” command, so please do so with your recordings before uploading. Individual segments in the same uploaded file should each be normalized separately.
On those rare occasions when a bird is cooperative and you’re able to make multiple recordings of the same individual bird, prepare each sound file as described above, and then group the files into a single file with 1 second of silence between each recording. The new file is the one that should be uploaded to eBird. This grouping of files should be done only when all of the files are from the same individual. Alternatively, if you record one sound file from one individual, and then record a second sound file from a different individual of the same species, you should submit each of those sound files separately, and not group them together.
If a recording contains multiple species that are loud and prominent, or particularly rare or otherwise interesting, the same recording can be uploaded for multiple species. However, the recording should be trimmed and normalized differently depending on the primary subject. If additional species occur in the background but are not particularly prominent or important, they can be listed in the comments field as background species.
If you have made a long, continuous recording of a bird that vocalizes infrequently (such as an owl), be sure to keep the entire recording. There is valuable information in the silence between songs—information that is lost if a continuous recording is chopped up into many short segments.
However, if you have made a continuous recording with interruptions such as walking or extreme handling noise, those sections can be removed and replaced with 1 second of silence.
We encourage recordists to make voice announcements in the field and append these announcements to their bird recordings. Making a voice announcement in the field is a great way to capture important information about the target species, date, time, location, and information that is important to the context of the recording, such as the animal’s behavior, habitat, weather. If you make a bird recording in one sound file, and then make a voice announcement in another sound file, merge the two together in a single file before uploading to eBird. In other words, please don’t upload a sound file that only contains a voice announcement.
Sound editing software can be a very powerful tool, allowing you to “erase” background birds, loud insects, raindrops, stomach growls, etc. While it can be tempting to remove these types of sounds from your recordings, we encourage you to give us recordings that include what you heard and recorded in the field, even if it’s not all pretty. Manipulating your recordings in this way can make them unusable for scientific research, as well as having the potential to make them sound unnatural. Because recordings in the Macaulay Library archive may be used for a wide variety of purposes, it is important to keep them as natural as possible.
Similar, apply filters only when necessary—if a strong low frequency sound (such as wind or distant road noise) is masking the target sound, apply a gentle low-cut (high-pass) filter at the minimum frequency possible (250 Hz or less). If the target bird is the loudest sound in a recording, there is no need to filter the recording at all. And again, noise reduction and other extreme editing techniques should never be used when editing your recordings for upload to the archive.
In the example below, you can see a recording of the song of a Dark-eyed Junco. Notice the loud handling noise at the beginning and end of the recording and the relatively low level of the bird song.
After editing, the clunks are removed and the bird song is normalized to -3 dB, resulting in a louder and much more prominent target sound.
Adobe Audition has the widest suite of features available and is available to try for free, but subscriptions must be purchased on a monthly or yearly basis for long-term use.
Ocenaudio is available for no charge and includes dual viewing windows of waveforms and spectrograms.
Audacity is available for no charge but lacks several features found in Audition and Ocenaudio, such as dual viewing of waveforms and spectrograms.