Wax Cylinders and Whistlers: the First Recorded Bird Songs

By Bird Nerd · 2026-09-24 · Behavior

The first bird song ever recorded was not recorded in a wood. In 1889, in Frankfurt, an eight-year-old boy named Ludwig Koch was given an Edison phonograph and a box of wax cylinders for his birthday. He pointed the horn at his pet — a White-rumped Shama, a thrush-sized songster from Southeast Asia kept for its voice across the region — and cut the earliest known recording of a bird. The room was a parlour. The bird was in a cage. The machine held about two minutes.

Everything you can play on this site descends from that cylinder, and the early story is as much about people as about birds: the men who carried fragile wax into cold woods at dawn, and the far larger number who had already spent their lives imitating what the machine was only learning to catch.

A horn, a stylus, and two minutes of wax

The Edison phonograph that Koch’s father brought home was the first recording machine sold to the public. Sound was funnelled through a horn onto a stylus, and the stylus cut a groove into the surface of a wax cylinder turning at roughly 160 revolutions per minute; the same machine played it back. A cylinder ran for about two minutes, so every recording was a single take with no editing and no second chance. Heard today they are narrow and noisy — the birds are there, but as if down a long telephone line.

April mornings, 1900

Nothing about the equipment had been designed for the outdoors. Cherry Kearton — who with his brother Richard had spent the 1890s photographing British birds’ nests and eggs, the first catalogue of its kind — bought an Edison Bell wax-cylinder recorder in 1900 for around £50 and took it to a woodland garden at Kenley, in Surrey. Through the early mornings of April and May he sat with nightingales and song thrushes and tried to get a few bars onto wax.

Two problems shaped the result. Cold cracked the cylinders, so Kearton kept them cool through the night to avoid any sudden change in temperature when he carried them out at dawn. Worse, the horn both collected the sound and played it back, so the noise of the stylus cutting the groove came out amplified into the same air the bird was singing in. The bird heard the machine and stopped.

What that effort left behind is a few notes of a Song Thrush and the song of a Common Nightingale — the earliest recordings of a wild bird singing, as far as the record shows. The cylinders themselves are lost. It would be another quarter of a century before microphones separated the listening from the recording, and field recording became something a person could do without disturbing the bird.

Song Thrush via xeno-canto.org/1151579, public domain. Kearton’s first wild subject, and the easy one to recognise by ear: each phrase is delivered two or three times before the bird moves on, with clean pauses in between. Open the sound picture under the player and that structure is visible at a glance.

Common Nightingale via xeno-canto.org/1099301, public domain. The other half of Kearton’s 1900 morning, and the bird behind the first commercial record of a bird (see below). It is also the hardest of these recordings to imitate: it runs on without repeating itself to order.

People were imitating long before machines

Every early recordist was trying to capture something the human ear had been copying for centuries. The cuckoo is the obvious case — two notes and a falling interval, a call most people in Europe and Asia can produce from memory without ever having been taught it. It was in European clocks long before it was on a record.

Imitation was also a trade: hunting lures, turkey calls and whistled garden songs were working knowledge, passed between people who needed them. When the record industry arrived, some of that skill went straight onto disc. The Library of Congress still holds a 1915 disc, Songs of our native birds, no. 2, whose own catalogue metadata reads, simply, “Bird imitations”, as well as bird imitations recorded at the 1938 National Folk Festival and turkey calls from Coal River, West Virginia. Listen to one of those and then to the real bird, and you can hear exactly what the ear keeps — the rhythm and the interval — and what it lets go.

Common Cuckoo via xeno-canto.org/546155, CC BY-SA. Two notes and a drop of roughly a third: the most imitated bird sound in the world. Whistle it back, then open the sound picture. The interval is the part humans copy; the harmonics stacked above each note are the part we cannot.

1910: the first bird record you could buy

In Berlin, Karl Reich kept and trained nightingales — alongside canaries, in the bird-keeping tradition of his city — and in 1910 issued a ten-inch, 78 rpm disc of a nightingale. It is usually named as the first commercially released record of a bird. The electrical microphone arrived in 1926 and changed the discipline again: a microphone could be placed close to a nest and camouflaged, with cables running to a recording plant up to a mile away, where an operator listened through headphones and lowered the cutting stylus at the right moment.

Ludwig Koch, by then a professional, used that equipment in the 1930s for what became the first popular birdsong publications — Songs of Wild Birds and More Songs of Wild Birds — and his archive later formed the foundation of the BBC Natural History Unit’s sound library. It is now held at the British Library. The shy eight-year-old with the parlour shama had become the reason a generation learned birds by ear from a record player.

What imitation keeps, and what it drops

Birds imitate each other too, and better than we do. The Marsh Warbler builds a song out of phrases borrowed from dozens of other species, delivered at speed in one continuous run — an imitation accurate enough that the original singers are sometimes identifiable from the recording. Common Starlings copy other birds, and also machinery, whistles and human speech.

What separates our imitation from theirs is detail. A whistler can reproduce a melody line; what gets left behind is the fine structure — the harmonics above the fundamental, the fast modulated trills, the ultrasonic components no human voice reaches. The two recordings below make that gap visible.

Marsh Warbler via xeno-canto.org/1083487, CC BY-SA. A mimic imitating other birds, at a speed no human imitator attempts: in 45 seconds this bird runs through borrowed phrases from many species, one after another, without a pause for breath.

Common Blackbird via xeno-canto.org/1051538, CC BY. The garden song most people whistle back. Hum it first, then open the sound picture: the repeated phrases and pauses are all there, but the bird’s notes are built from layered overtones that a whistle flattens into a single line.

Try the test yourself

Whistle the cuckoo’s two notes before you play anything. That interval is the whole of the imitation.
Play the cuckoo recording and switch on the sound picture under the player. Look at the second note: the interval is intact, but the note carries harmonics above it, and the bird adds a small break that the ear smooths over.
Play the Song Thrush and count the repeats. Kearton managed “a few bars” of exactly this — the phrase, two or three times, then a pause.
Finish with the Marsh Warbler and the Blackbird: one copying everything it hears, the other working four or five phrases all spring. Human imitation sits between the two.

Every player on this site carries a Spectrogram switch in the row with the play button — the tiny “Tap to see the sound” line beneath it. Nothing is downloaded until you turn it on, and the picture is calculated in your browser from the recording itself, so what you see always belongs to the audio you are hearing. Links such as /blog/first-bird-recordings-wax-cylinders-whistlers.html#spectrogram-2 open a page with one of these pictures already showing.

Why the frozen song still matters

Before 1889 a bird’s song existed only in the moment and in the memory of whoever heard it, which is why imitation mattered so much, and why the cuckoo ended up in clocks. What the wax cylinder did was make comparison possible: a nightingale in Surrey against a nightingale in Berlin, this spring against 1936.

The 1889 shama cylinder is the direct ancestor of every recording on this site, and so are the conditions that come with them: these songs can be heard, shared and reused because the people who made the recordings chose to allow it. The machine is 137 years old. The imitations it caught are far older than that, and you can still perform one this evening.

Sources: the British Library’s account of Ludwig Koch’s 1889 shama recording; John Bevis on Cherry Kearton’s Kenley recordings of 1900; the MIT Press Reader’s history of collecting and imitating birdsong; and Library of Congress items Songs of our native birds, no. 2 (1915) and bird imitations at the 1938 National Folk Festival, with the American Folklife Center’s post For the Birds. Dates, catalogue wording and quotations follow those sources; the listening guidance and English text are written for this site.

← Back to all articles