Reading a wildlife spectrogram
Read time left to right, frequency bottom to top, and intensity through darkness or color; then compare shapes only after checking scale, window settings, sample rate, background noise, and the original audio.
Scope: A practical introduction to reading time-frequency displays of animal recordings, especially bird, frog, insect, and mammal sounds. Spectrogram patterns support comparison, but settings, noise, and individual variation prevent a picture alone from guaranteeing identification. · Last updated

Orient the three dimensions
Move left to right to follow time and bottom to top to follow low through high frequency. Darker, brighter, or warmer-colored pixels—depending on the palette—show greater energy in a time-frequency cell, not necessarily what a listener perceives as equally loud. Read the printed scales first: two spectrograms can display identical sound at very different widths, heights, gains, and frequency ranges. [1][2]

Translate shapes back into sound
A thin horizontal trace suggests a nearly constant pitch; a rising or falling curve indicates a sweep; repeated vertical strokes often mark clicks or short pulses. Parallel bands may be harmonics generated by one source rather than several simultaneous animals. Note spacing, duration, peak frequency, repetition rate, and sequence structure, then listen again—visual pattern and audio interpretation should constrain each other. [1][3]

The window changes what you see
Spectrogram software divides sound into overlapping windows before estimating frequency content. A longer window separates nearby frequencies more clearly but smears rapid timing; a shorter one sharpens timing while broadening frequency bands. This uncertainty tradeoff means screenshots made with different FFT sizes or window functions should not be compared as though every fuzzy edge came from the animal. [2][4]

Treat the display as measured evidence
Wind, insects, traffic, echoes, microphone response, clipping, compression, and filters can add or erase marks. Frequencies above half the recording sample rate cannot be represented faithfully, and a faint call may disappear beneath the noise floor. Match candidate species by several features and context—location, date, habitat, rhythm, and audible quality—because a similar trace is supporting evidence, not proof of identity. [3][4]
Related guides
Identify it and save the field note.
Where this guide comes from
Source-checked editorial guide. Last updated . This guide teaches identification and field skills; it is not a substitute for expert verification when it matters.
- Cornell Lab of Ornithology — How to use spectrograms ↗
- American journal of primatology — Bioacoustic field research: a primer to acoustic analyses and playback experiments with primates ↗
- Ecology and evolution — BASSA: New software tool reveals hidden details in visualisation of low-frequency animal sounds ↗
- Bioacoustics — Spectrographic analysis and the uncertainty principle ↗


