USV Call Annotator
A browser-based tool for annotating ultrasonic vocalizations (USVs). Load an audio recording, view its spectrogram, and mark each call as a time–frequency box, then export clean annotations for analysis.
What it does
Mice communicate with ultrasonic vocalizations well above the range of human hearing. To study them, researchers first have to find and label each call in a recording, a slow, manual task. The USV Call Annotator makes it fast and visual: it turns an audio file into a spectrogram (a time-versus-frequency picture of the sound), lets you draw a box around each call, give it a label, and play it back to confirm. Annotations export as CSV and JSON for downstream analysis. Everything runs in your browser, and your audio never leaves your computer.
How to use it
- Click Launch USV Annotator, then drag in a .wav recording (or set a folder). The tool computes the spectrogram automatically.
- Scan the spectrogram and drag a box around each call to capture its start/end time and frequency range.
- Give each box a label (e.g. "USV" or your own call type), and play it back to verify.
- Zoom and pan to inspect closely; adjust the display threshold to bring faint calls out of the background.
- Export your annotations as CSV (and a per-file JSON) when you're done. They save right next to your recording.
Highlights
Quick start
Open the tool, drag in a .wav file, and start boxing the calls you see on the spectrogram. Label each one, then export to CSV. Questions or feature ideas? Reach out at guangwei.zhang@vcuhealth.org.