In 1961, the first computer to model a human voice sang a song and it was terrifying

In 1961, the first computer to model a human voice sang a song and it was terrifying
In 1961, the first computer to model a human voice sang a song and it was terrifying

Before the IBM 7094, computers didn’t talk. In 1961, a major turning point took place and was embodied in a song, “Daisy Bell” (also called “A Bicycle Built for Two”). Computing found its voice in Murray Hill, New Jersey, at Bell Laboratories, on a computer then considered the most powerful in the world.

Two scientists, Carol Lochbaum and John Kelly, were working on a voice synthesis system, today called the “Kelly-Lochbaum voice model”. To test the progress of their research, the two men programmed the British song “Daisy Bell” which dates from 1892. We can assume that the choice of the song is linked to its title recalling the company which employed them (Bell Laboratories). , but nothing has ever been confirmed. The IBM 7094 did not sing a cappella, since it was accompanied by an instrumental track, also computer-generated, this time thanks to an engineer (and amateur violinist) named Max Mathews.

In the age of artificial intelligence (AI) and voice assistants, what might seem relatively mundane was at the time a major challenge. The power of the processors greatly limited the room for maneuver of researchers, reports the British online media IFLScience. Since audio cannot be processed in real time, Max Mathews had to record an hour of music and speed up the soundtrack to get seventeen seconds of playable melody.

Subscribe for free to the korii newsletter!Don’t miss any korii articles thanks to this daily selection, directly in your inbox.

Here is computer music

Added in 2009 to the National Recording Registry of the United States Library of Congress, the song is considered a historical document. The first example of digital music, a precursor to all developments in speech synthesis and electronic music that followed, the recording is also described as a technical feat.

“By today’s standards, the IBM 7094 version of “Daisy Bell” seems quite primitivenoted the writer Cary O’Dell, in an essay for the American Library of Congress. The sound is as flat as a telephone dial tone. It is, after all, literally the voice of a robot.”
However, he adds:
“Nevertheless, this meeting between music and machine remains a bold and unique step forward, a great leap towards a new, better world.”

British writer Arthur C. Clarke was able to listen to this particular version of “Daisy Bell.” Touched by this revolutionary demonstration, he drew from it an emblematic scene from his science fiction novel 2001, A Space Odysseyin which the on-board computer HAL 9000 (or Carl 500 in the French version), deactivated by David Bowman, begins to sing this song, before seeing its functions and its speech gradually disintegrate. This passage also appears in the eponymous film by Stanley Kubrick, produced in parallel and also released in 1968.

Perhaps the more sensitive among you will see this as a poignant version, a song for the story, while others may find it passable and annoying. But this short audio production nonetheless remains one of the most important in our fledgling technological culture.

-

-

PREV 60 euros reduction on the Bluetooth speaker that makes your walls (and your neighbors!) shake
NEXT The five best video games of 2024