Patrick Feaster, an audio historian with a photographic memory for phonographic history, dropped a bomb at the Association for Recorded Sound Collections meeting on May 15, 2026. His claim: a working sound recording machine predated Thomas Edison's phonograph by nearly a century—built in the 1780s by a German medical and mechanical enthusiast named Georg Theodor Jacob Müller.
Feaster's investigation started over 20 years ago when he stumbled on a German article mentioning a "Müller" who exhibited a talking machine in Erlangen, Germany, in 1788. The article dismissed it as a hoax. But Feaster kept digging and found a 1788 book describing the device, plus two eyewitness accounts that agreed on details. The machine stood 3.5 feet wide by 2.5 feet high, flanked by life-size male and female figures, and contained 34 speech mechanisms resembling organ pipes, along with levers, rollers, cylinders, clockwork, and 10 bellows.
The Critical Twist: Two Müllers
In January 2026, Feaster made a key discovery: there were two Müllers demonstrating speaking machines in 1780s Germany. One, Laurentius Müller, used a puppet that was indeed a hoax. The other—Georg Theodor Jacob Müller—was a serious devotee of medicine and mechanical sciences. Multiple contemporary accounts, including a testimonial from physicist Johann Tobias Mayer, described his machine producing distinct male and female voices by passing sound through tubes from the cabinet up through the figures' arms into their mouths.
Mayer noted that removing the figures and pressing an ear directly to the cabinet's top made the speech clearer. The machine's repertoire included answers to 12 riddles, book passages, laughing, crying, kissing sounds, and even sung arias—feats that Edison's phonograph would later accomplish by recording and playing back the human voice.
The Echo That Couldn't Be Natural
Two details give Feaster's argument weight. First, Müller's machine used an "artificial ear"—a mechanism that gathered sound from air, similar to a hearing aid from the 1780s. An artificial ear could have been part of a recording device. Second, audience members who spoke three or four words into the figure's ear heard those same words repeated back after a delay—in their own voices. A natural echo with that delay would require a far larger volume than Müller's cabinet. The simplest explanation: the machine recorded and played back the words.
"Even if Müller was a fraud," said Jacob Smith, a media historian at Northwestern University, "Patrick has given us a richer picture of the horizon of imagination surrounding talking machines long before Edison." Feaster himself helped rewrite sound-recording history in 2008, when he and colleagues demonstrated that the phonautograph of the late 1850s likely captured sounds on paper first.
No physical machine survives, and the evidence remains circumstantial. But Feaster's careful sifting of 18th-century accounts—and that impossible echo—suggests the first voice recording may have happened a century earlier than we thought.
Source: Edison may not have been the first to record the human voice, new evidence suggests
Domain: scientificamerican.com
Comments load interactively on the live page.