← Atlas catalogue V3 · Sonnet

The second you can’t hear

Bats at dusk

A bat emits a short pulse and waits for the echo. Two synthetic rhythms, sparse (A) and dense (C), on one time axis. Move the obstacle and check how many pulses are emitted before the echo of the first one returns.

Pattern Asparse rhythm · IPI 90 ms · pulse 5 msecho after 17.6 ms

Sonogram of pattern A: two pulses and a modelled echo Window 0 to 200 milliseconds, frequency from 10 to 90 kilohertz. Two sloping pulse traces of pattern A at 20 and 110 milliseconds, each descending from 48 to 38 kilohertz. The violet dashed trace is the modelled echo of the first pulse; its delay is given by the label above the chart and the figures below.

Pattern Cdense rhythm · IPI 7 ms · pulse 1.2 msecho after 17.6 ms

Sonogram of pattern C: 26 pulses and a modelled echo Window 0 to 200 milliseconds, frequency from 10 to 90 kilohertz. Twenty-six short, steep traces every 7 milliseconds, starting at 20 milliseconds, each descending from 55 to 24 kilohertz. The violet dashed trace is the modelled echo of the first pulse; its delay is given by the label above the chart and the figures below.
  • pulse emitted
  • echo of the first pulse (model)
  • horizontal axis: time, ms · vertical axis: frequency, kHz · diagram with synthetic parameters, not a recording
3.0 m

Keyboard: arrow keys change d by 0.1 m, Home and End jump to 0.5 m and 18 m.

Round-trip distance (2d)
6.0 m
Echo return time (t = 2d / v)
17.6 ms
Speed of sound (model assumption)
340 m/s

At d = 3.0 m the sound travels 6.0 m (there and back), and the echo returns after 17.6 ms.

Without JavaScript you see the initial state (d = 3.0 m). The slider, scale switch and playback need JavaScript; explanations and sources are below.

Pattern A IPI 90 ms

0pulses emitted in the meantime, before the echo of the first one returns

The echo returns before the next pulse: one response corresponds to one pulse.

Limit in the model: d = 340 · 0.090 s / 2 = 15.3 m. From there the echo coincides with the next pulse; up to approx. 16.15 m it begins while that pulse is being emitted (90–95 ms), and beyond that it falls after it.

Pattern C IPI 7 ms

2pulses emitted in the meantime, before the echo of the first one returns

Before the echo returned, the pattern emitted 2 more pulses (nos 2–3). The echo falls between pulses nos 3 and 4, so its position alone does not say which pulse it answers.

Limit in the model: d = 340 · 0.007 s / 2 ≈ 1.2 m (precisely 1.19 m). Up to approx. 1.39 m the echo begins during the next pulse (7–8.2 ms), and beyond that it falls after it.

Scale 2 · 6 ms window

One pulse up close

A single pulse lasts from a fraction of a millisecond to a few milliseconds, and its frequency falls during it. This is exactly the sloping trace on the sonogram. Both windows have the same scale (0–6 ms, 10–70 kHz), so the difference in length and steepness can be seen without converting. Both patterns are synthetic: they do not describe any species.

Axis: ×1 · real

Real scale (×1): ultrasound, which cannot be heard without a detector. The switch changes only the description of the axes and figures; the shape of the trace stays the same.

Pattern Aone long, gentle pulse
A single pulse of pattern A in a 6 ms window The trace descends gently from 48 to 38 kilohertz over 5 milliseconds. Window: 0 to 6 ms, 10 to 70 kHz, the same scale as in the window for pattern C.
Pulse duration
5 ms
Bandwidth (FM sweep)
48 → 38 kHz
Inter-pulse interval (IPI)
90 ms
Repetition rate
11.1 Hz
Pattern Ca short, steep pulse
A single pulse of pattern C in a 6 ms window The trace descends steeply from 55 to 24 kilohertz over 1.2 milliseconds. Window: 0 to 6 ms, 10 to 70 kHz, the same scale as in the window for pattern A.
Pulse duration
1.2 ms
Bandwidth (FM sweep)
55 → 24 kHz
Inter-pulse interval (IPI)
7 ms
Repetition rate
142.9 Hz

Axes of both windows: horizontally, time from the start of the window in milliseconds; vertically, frequency in kilohertz (10–70 kHz). That is why the wide, long trace of A and the narrow, steep trace of C can be compared directly. Time expansion plays the recording back more slowly; according to the Bat Conservation Trust, a tenfold slowdown lowers 50 kHz to 5 kHz [3].

Playback on demand · synthetic, ×10

You start the sound yourself and can stop it. It plays 200 ms of the chart stretched to 2 seconds: the first pulse, its modelled echo (at the current d from the slider) and the following pulses of the pattern. These are tones generated in the browser, not a recording of a bat.

Silence. You start playback yourself; the volume is limited.

The same without sound

After a ×10 slowdown, pattern A is a series of single short ticks every 0.9 s, each at a pitch of about 4–5 kHz. Pattern C is a rapid series of short, falling squeaks, about 14 per second. The echo of the first pulse is heard as a separate, quieter tick after a time 10 times longer than t.

The playback is always processed, whichever window and axis scale is chosen.

How to read this

Signal, echo and playback are three different things

Emitted pulse
The sound sent out by the animal. On this page it is only a synthetic shape with the stated parameters, not a field recording.
Echo (calculated in the model)
The reflection of the same pulse after a time t = 2d / 340 m/s. The path is 2d because the sound travels to the obstacle and back. The echo depends on distance, not on the emission rhythm.
Processed playback
The same waveform played back 10 times more slowly, until the frequency falls into the audible range. This is how time-expansion detectors work [3]. It is not the sound that a bat hears, nor that a person hears without a detector.

What must not be concluded from this page

  1. The species. A sonogram is not a taxonomic stamp. The same bat changes its bandwidth and pulse length depending on whether it flies over water or among branches, and the frequency ranges of many species partly overlap [1] [2]. Patterns A and C are neither species nor a particular evening.
  2. The distance to prey. The formula d = 340 · IPI / 2 says only what obstacle distance would give an echo coinciding exactly with the next pulse in this model. A simple product does not prove the real distance of prey, and the animal changes its rhythm in flight.
  3. The principles of perception. When the echo falls after the next pulse, the model shows only that the position of the echo on the axis is ambiguous. It does not describe how a living animal copes with this.
  4. A constant speed. 340 m/s is a model assumption (taken over from V1). The speed of sound depends on air temperature.
  5. A detector recording. A heterodyne detector listens to a narrow, tuned band; frequency division loses amplitude dynamics; a full sonogram is given by time expansion or broadband recording [2] [3]. The recording and its context are worth discussing with a chiropterologist.

Sources and status

The links were carried over from version V2. In V3 they were not reopened, so their content needs to be checked by the editors and a chiropterologist. The charts on this page are our own diagrams with synthetic parameters, not a map or a field record.

  1. [1] Bat Conservation Trust: flight, food and echolocationLink from V2; to be rechecked.
  2. [2] Bat Conservation Trust: introduction to bat detectorsLink from V2; to be rechecked.
  3. [3] Bat Conservation Trust: time expansionLink from V2; to be rechecked.
  4. [4] Nadleśnictwo Oleśnica Śląska: spotkanie z nietoperzami [Oleśnica Śląska Forest District: a meeting with bats]Link from V2; to be rechecked.
  5. [5] Illustrative photo of a common pipistrelle from V2, Gilles San Martin: Wikimedia Commons, CC BY-SA 2.0Link only: the image is not embedded (the page works offline). Attribution and licence to be checked.
  6. [6] Design inspiration, not a source of facts: Bartosz Ciechanowski, SoundDescription checked when the package was prepared; it was not opened in this session. Nothing was copied from it.