In the ninth century Persian scholars invented the first known mechanical instrument, a hydropowered organ that played music preprinted onto a rotating cylinder. It would be 1,000 years before inventors cracked the reverse process—printing sounds onto a storage device.

The first machine that could pull music from the air was Édouard-Léon Scott de Martinville’s phonautograph, which he introduced in 1857. The device used a horn to focus sound waves and direct them onto a small diaphragm; attached to the diaphragm was a stylus that scratched a record of the waves onto a soot-stained rotating glass cylinder. The device showed that sound recording was possible, but it remained a historic curiosity for a simple reason: it could not play back the recorded songs. (At least not until last year, when researchers at Lawrence Berkeley National Laboratory deciphered the scratches and played an 1860 recording of a woman singing Au Clair de la Lune.)

De Martinville’s phonoautograph has remained a quaint footnote, but his basic architecture of a horn, diaphragm, stylus and cylinder provided the foundation for all sound recording for the next 70 years. In 1874 Alexander Graham Bell experimented with sound recording using de Martinville’s architecture, except he used a cadaver’s ear. He abandoned his efforts to focus on the telephone, which he introduced in 1876. A year later Thomas A. Edison (right) was experimenting with a way to record sounds made by Bell’s telephone when he shifted efforts to record sounds in the air. His setup was almost identical to de Martinville’s except that Edison used tin foil as his recording surface, which allowed for playback. He brought the phonograph to the offices of Scientific American in December 1877, the same month he patented the device. We wrote, “No matter how familiar a person may be with modern machinery and its wonderful performances... it is impossible to listen to the mechanical speech without his experiencing the idea that his senses are deceiving him.”