Australia’s CSIRAC made the first computer-generated melody, but no recordings remain. For other primitive early computer music, catch new strains from the BBC from 1951. Photo by thefunklab.

As several of you noticed, the BBC has discovered 1951 recordings of computer-synthesized music, predating the previous earliest recordings from New Jersey’s Bell Labs in 1957.

‘Oldest’ computer music unveiled [BBC News]

So, who gets the credit for the first digital synthesis? This particular recording doesn’t change much, in that Bell was never recognized as the first computer-created music – they just happened to have the earliest recordings still available.

Here’s the timeline:

Australia: CSIRAC

Australia’s CSIRAC computer is generally recognized as the first computer to play a musical melody. Bonus points: the CSIR Mk1 played music in August 1951, before just about anything else, and it did perform in real-time. But it doesn’t count as the first digital synth, because it just sent raw pulses of data directly to the speaker. (Digital-to-analog converters didn’t actually exist in practice at that point.) And since no one seems to have recordings of the CSIR Mk1 from August, the UK Ferranti Mark 1 recording for BBC wins in the documentation category.

That said, the CSIR Mk1 definitely wins cool points for its unusual sound production methods, and it even got a program for easier, more musical entry of melodies later (1957, via Thomas Cherry.)

More details here:

The Music played by CSIRAC [Melbourne School of Engineering]

CDM loves Australia, so perhaps we can simply say that Australia gets the credit for the first computer-programmed musical materials, just not digital synthesis as they lacked a DAC.

UK: Ferranti Mark 1


This Ferranti box made some of the first musical noises by a computer. From an electronic pr0n-filled page of vintage UK ads, via the wonderful Mike’s Electric Stuff.

The BBC recorded the recording above several weeks too late to beat Australia, but the computer in question is the heir to a very important machine. Baby (Small Scale Experimental Machine) is credited as the first “universal” computer – that is, it had the marked advantage of functioning with a range of custom software rather than needing to be rewired each time you wanted to change tasks. Baby predated Australia’s CSIRAC by about a year. The recording here is of the Ferranti Mark 1, which is a production-ready descendent of the Baby.

The recording itself is very interesting, because like the CSIR Mk1, the software used allowed for basic custom programming of melodies. Likewise, while the primitive music was little more than a novelty, it did illustrate to the public that computers would have broader applications than most had initially imagined.

Like CSIR Mk1, though, this isn’t really “computer music” as we now understand it – though it did involve basic sequencing.

US: IBM 704 and MUSIC

The legacy of Max Mathews and Bell Labs remain safe, though, despite these early discoveries. The key is the content: early computers could make noise, but Mathews’ team at Bell could (ahem) create digital music. It was one small step for sound, one giant leap for software. Whereas the CSIR Mk1 and Ferranti machines involved primitive tools for sequencing melodies, the Music I software allowed for control over synthesis, relatively sophisticated and musical descriptions for scores, and an architecture that’s (incredibly) still widely-used in music applications today. (A direct descendent is even included in the XO Laptop project’s Sugar software suite, though nearly any modular computer synth would recognize Music-N as its grandfather.)

The key moment is 1957’s recording of a 17-second composition by Mathews on an IBM 704. (That also means that Music N and Mathews handily win the first original computer music prize.) Max tells some great stories about that train ride over to Manhattan from New Jersey to feed the 704 music. The 704 wasn’t yet speedy enough to handle real-time sound generation, either – let alone counting the train commute time.

Bell did it again in 1961 with a computer-synthesized singer, as heard here during 2007’s Maxfest – arrangement by Max, voice synthesis technology from a team led by John Kelly:

To me, though, these achievements are related. There’s the initial proof of concept, even one as simple as sending data words to a speaker and arranging a melody – then the all-important architectural breakthrough, one that can actually act as a real foundation for the future. And as much as people talk about rapid technological change, conceptual achievement really led the way. Harry Nyquist imagined the basics of digital signal recording as early as 1928, later proven by Bell and MIT’s digital maestro Claude Shannon. (Indiana University has a great big-picture overview of these developments.)

And while these particular examples from Australia and Manchester aren’t entirely musical, I think it’s equally fascinating that the chiptune, circuit bending, and DIY electronics scenes are all returning to the fundamentals of digital synthesis. To these younger musicians, it seems to be a way to get more intimate with the organic qualities of the technology.

Thanks to all of you who sent this in!

  • gabeknox

    This is cool and all, but let's live in the present. Who will have the first QUANTUM computer making weird noises?

  • That was actually my next question … I wonder if it's feasible to get any vintage machines up and running to recreate music!

  • I was at that Computer History Museum event where that video of Max was shot at, and had an amazingly great time.

    I was also at the event the following night at Recombinant Media Labs where Max told the story about how HAL singing Daisy in 2001 came to be. I believe I have video of it. I'll dig it up.

  • scott flavin

    Hmmm…brings me back to an idea i had last year about turning my grandmother's still-working 1984 IBM PC into a synthesizer. How to do this, i still don't know lol

  • gabeknox

    I would think it unlikely, at least if these labs are anything like my school. It's a pretty big and thankless undertaking to get "obsolete" legacy equipment up and running at universities. We have a Synclavier sitting around in my department that a few of us weirdos vainly attempted to get up and running, and if we couldn't even get that comparatively state-of-the-art machine going, I doubt anyone would bother going through the massively complex undertaking of getting these beasts running for nostalgia/bragging rights purposes…not that I'm bitter or anything…

  • Bodhi

    Surely the CSIRAC one should be considered a digital synth, it just had a software DAC!

    Yes, I'm Australian 😉

  • @Bodhi … ah, see, I hoped this would spawn some discussion. That's a good point. CSIRAC also beats MUSIC I/IBM 704 on the real-time front. I suppose the technically correct thing would be to say the MUSIC I system Max devised (computers underneath aside) easily counts as the first modern digital synth. I think MUSIC-N also sits alongside the Moog modular system (with a nod to Buchla and EMS around the same time) as being one of the music technological inventions whose basic architecture remains in use today.

    PS, I'm not *as* up on the computer history, but I'm surprised Baby gets as much credit as it does as the first universal computer when the CSIRAC is basically neck and neck with it in terms of achievements and timing. Maybe the history is being written by the Brits? (In digital sampling, we have a similar thing with the Fairlight CMI overshadowing the Synclavier, while each is arguably equally important in terms of their actual effect, nitpicking on dates aside…)

  • Christian Haines

    There is a book about CSIRAC written by Paul Doornbusch a number of years ago.

  • danner

    Back in 1979, I was working on IBM mainframes as a systems programmer. We had an IBM CE play 'music' on our huge chain-drive line printer. The printer was really loud and the sounds of various characters printing produced different pitches. He had several 'songs' he would play that would use a combination of printing characters, the warning bells and lights and even open and close the huge printer door. very cool…

  • Bodhi

    From what i've just read about CSIRAC and the MUSIC system, I'd agree that the MUSIC I would be considered to be more of an actual synthesizer, whereas CSIRAC and Baby seem to be more electronic pianolas. But does it really matter who was first? Surely how much of a contribution was made to electronic music is much more important! 😉

    I'm also fascinated that after building a fantastical magic machine, ie. a computer in the '50s, one of the first things they did with it was get it to make some noise…

  • Bodhi

    An addendum: according to Wikipedia, the Synclavier was an order of magnitude more expensive than the Fairlight CMI!

  • If you read the books carefully, CSIRAC played music before the Ferranti Mk 1 (the commercialised Baby) by at least a few months if not a year… But it was never recorded, only meticulously reconstructed in about 2000. So they do have the first _recording_ of a computer playing music.

    But, being the first is not so important. The Ferranti music was a curiosity. With CSIRAC it was an early programming challenge, and it was used politically to try and save the project by popularising the computer. Unfortunately neither had composers involved… That's where Max Matthews and co. got it right. They had a DAC, and did DSP, and then they got composers involved because as engineers they didn't know much about music…

    CSIRAC's sounds are a kind of digital synthesis, just not controllable with timbre, as it was feeding digital pulses from the buss to the speaker. MUSIC I also had a fixed waveform for all synthesis. MUSIC III did the real thing, and that was the real start of computer music.

  • @tomson: Well said. I would still say that, even with a fixed waveform, Max basically had the idea at MUSIC I, even if it took until MUSIC III to flesh the rest out, but yes, MUSIC III finally looked more like Csound and modern digital synths.

    On a separate note — I'm not actually sure about Wikipedia's version of the early history of the digital sampler. For one thing, the Fairlight CMI article claims that the Synclavier team only added sampling when they found out the Fairlight could do it. That *may* be the case, but when I last spoke with Jon Appleton from Dartmouth, I got the sense that they were basically developing the concept at the same time, and the Synclavier actually beat the Fairlight to market with a hard drive option. And regardless, there are some important citations missing there. (Hey, it's Wikipedia; that's the fun of it.) The short version of the story, of course — as with the early computer pianola experiments — is that various people were working toward similar goals, which is why "first" tends not to be meaningful outside record books.

    The Synclavier was wildly expensive, yes. It was more like a custom/modular system than the Fairlight in that you could custom-order what you wanted, but yeah, still an order of magnitude beyond the Fairlight.

  • Pingback: DIY Stompbox: Beavis Board, BodySurf // Wii Balance Board + Audiosurf + Motion Controls, Elektron-Users: Richard Devine interview, New Early Computer Music Discovered; What Was the First Digital Synth?, Exclusive Tim Exile Video,()

  • Peter, yes, Max M did have the idea with MUSIC I, but he also had the technology, with one of the first AD converters… It was an inevitable development really. The table-lookup oscillator, unit-generators and separating the score and instrument specification files were all strokes of genius. But CSIRAC also had a music program which would take a separate score file by 1956 or so.

    I'm not sure about the whole Synclavier and Fairlight thing. I do know that the Fairlight wanted to use additive synthesis, but the quality was so crappy that they decided to record real sounds instead – it just tirned out to be a cool idea. I don't know if Synclavier had the idea independently or subsequently to Fairlight. Many similar things happen at about the same time in different places. Like penicillin being developed simultaneously in different parts of the world. And I thought that while the Synclavier was initially more expensive, eventually both of them were just very very expensive, but I don't know for sure.

  • sasarasa88

    @scott flavin

    Check out MONOTONE, The Tracker For Extremely Simple Sound Devices, at

  • Damon

    And the DIY retro kit version will surely be relased shortly. Get your order in early.

  • Pingback: Disquiet » tangents / 3 months of 4′33, mic’ing Marfa, three Martinez scores …()

  • We need more insights like this in this thaerd.

  • q5JSLS ecbumebpomec

  • rFEUXS tzbkefraylzb