Starting January 2026, any audio uploaded to eBird and the Macaulay Library will be automatically normalized to a standard volume level. This means that the peak volume (the loudest part of the file) will be increased—or decreased—so that it reaches -3dB (3 decibels below the maximum volume for a file). This automated step follows audio best practices that we already encourage, helps keep the archive uniform so that files can be compared, helps quieter recordings be more audible.
What is normalization?
Normalization is the process of increasing or decreasing the volume of a sound file so that the peak volume (the loudest part of the file) reaches a standardized point. We use -3dB as an archival standard for normalization. So if your uploaded file peaks at -30dB, it will be increased by 27 decibels, and will sound much louder. This is different from compression or limiting, which change how audio behaves at different volume levels. It's really more like just turning up the volume on the existing file.
Why are we doing this?
Normalizing helps files be more comparable. It allows quieter recordings to be more audible, fainter recordings to show up better on the spectrograms, and it standardizes recordings so they can be compared more easily, as well as assessed for quality.
Why am I hearing more background noise now?
Turning up the volume doesn't always sound better! Normalizing doesn’t change the quality of a recording, it only makes it more audible by increasing the volume uniformly across the file. Recordings with a lot of background noise (wind, traffic, etc.) may end up being less pleasant to listen to after normalization, but only because both the bird and the other noise sources are now easier to hear. See our recording tips page for strategies to reduce background noise in the field.
Is my original file being altered?
The normalization process only runs on the lower-sized file we stream on the web, as well as the resulting spectrogram. The original file will not be altered.
Do I still need to trim files before upload?
Yes! Trimming will be just as important as always, perhaps even more so because a recording with a loud bump at the beginning will probably normalize to that bump rather than the bird, resulting in a quiet overall recording—the opposite of what we’re trying to achieve with normalization.
I already normalize my files when I prepare them for upload. Does this mean I can skip this step now?
Maybe! While a simple file, once trimmed, should look and sound the same whether you normalize to -3dB yourself or let the system handle it, there are a few cases where it might have a less desirable result, so if you’re in the habit of normalizing already, you may want to keep doing it. A few reasons:
- Normalizing during editing can help you hear the recording better and be able to trim more accurately.
- If you make multiple recordings of the same bird and combine them into one file, per best practices, you should normalize each segment separately. The auto-normalization will not account for this, and will normalize the file as a whole.
- If you are recording in 32 bits and go over 0dB, you should normalize before uploading to recover this information. Because the auto-normalization only runs on the lower-resolution proxies, it will not recover 32-bit recordings that are over 0dB.
- You may want to normalize first so that your originals are also normalized. We are only auto-normalizing the web copy, so if you download your original, it may still be at a lower volume.
Will this apply to recordings before January 2026?
At this time, we are not running this process over previously uploaded recordings, only those going forward. We may decide to go back and retroactively regenerate normalized proxies in the future.