Usually when someone mentions the idea of computers generating folk songs, it’s a snide dig at some over-produced, stupidly bombastic “indie-folk” act that’s dominating the airwaves with insufferable, flannel-soaked mediocrity (looking at you, Mumford and Sons.) However, in a mind-boggling undertaking that raises dozens of bizarre questions regarding musical authenticity, the capability of robots as artistic beings, what exactly constitutes sentience, and whole lot of other hypotheticals that would have Turing salivating (or possibly spinning in his grave), Canadian folklorist Dr. Henry Svec and Czech programmer Mirek Plíhal have been churning out new Canadian folk songs using a computer program called LIVINGSTON.The LIVINGSTON program takes its name from the wonderfully obscure and eccentric (read: completely fictional) Canadian folklorist Staunton R. Livingston. Livingston is a personality invented by Svec, the basis of a long-running performance piece about Canadian folklore, and a man who is said to be responsible for unearthing an album of lost Stompin’ Tom recordings and discovering an album of songs written by CFL players in the 70s. In reality, both are actually albums of songs written by Svec and performed with a bunch of musicians that have gone viral and continue to be attributed to the fake folklorist thanks to brilliantly orchestrated media ruses.To create the songs for this album however, LIVINGSTON the computer program was provided with staggeringly large database of old Canadian folk songs that Svec himself had amassed through legitimately assuming the role of a folklorist he’d for so long attributed to a fictional character. Using an algorithm developed by Plihal the two designed LIVINGSTON to be able to find, analyze and reproduce patterns in the chords and lyrics of old folk songs to develop new, novel creations. Armed with these computer penned ditties, Svec then got to work on bringing these songs to life with a bunch of his musician pals. First recording base tracks for the songs, then calling in favours from staples of the Canadian folk scene that he’s previously worked with to come in and flesh out the tracks. The result is an album called Artificially Intelligent Folk Songs of Canada. The duo just put out the first volume out of a planned three that includes the likes of Misha Bower (Bruce Peninsula), JJ Ipsen (Hayden), Marshall Bureau (Octoberman, The Pinecones) and more.Not content with regurgitating old literary tropes and folk clichés, the pair also gave LIVINGSTON free reign to scour the dark and often horrifying depths of the internet for content as well, resulting in some absolutely golden song titles like "S/He is Like the Angry Birds" and "I Am a Weary Immaterial Labourer in a Post-Industrial Wasteland."To find out what it’s like working with a seemingly insane, all-knowing, folk song writing machine and what this kind of technology means for the future of music, we sat down with Dr. Svec and attempted to gauge whether or not we should be taking shelter from the clearly imminent musical robot uprising.
Dr. Henry Svec:Last spring, I was doing a residency in Dawson City, Yukon, at the Klondike Institute of Art and Culture. I'm a folklorist and a song collector, and the project I was working on there initially was a collection of Klondike attempted murder ballads. I also had a fellow artist-in-residence in Dawson, Mirek Plíhal, who is a computer scientist and a programmer from the Czech Republic.Honestly, one day early on, Mirek and I were just hanging out around this old house our studios were in, drinking beer (as you sometimes do in Dawson City) and watching Jeopardy! on YouTube. In particular, Mirek was showing me these episodes where a computer, WATSON, plays against all the great champions of the game, WATSON really slaying them all.It occurred to me that it would be fun to try to build a similar machine in the field of Canadian folk music, and we immediately hit the ground running. I was actually partly joking at first, but Mirek thought it was doable, or at least worth trying. Mirek himself had been working on an app about Newfoundland, but we both just set our own work aside and dug into LIVINGSTON as a team.Did you know how you were going to approach it at first, or did you have to do some digging to find out how to implement the idea?LIVINGSTON is an open and an ongoing entity; it's changed quite a lot since it first started writing songs, and it's still changing. Mirek and I were quite willing to keep working on and revising the thing. The songs for Vols. 1-3 have already been decided, and you can really see a development and a maturation, which is not just LIVINGSTON's achievement but the collective and collaborative achievement of all three of us, although obviously Mirek has been largely responsible for the archival dynamics and the compression processes.But, anyway, Mirek seemed to know what he was doing! Of course, he was constantly reading up on various programs and functions that he decided we'd need as we began to work; as I understand it, the skills of the programmer are in constant flux and in constant need of updating.How does Livingston work, exactly?
I'll try to keep it simple here. First, we wanted LIVINGSTON to be able to organize the messy archives of Canadian folksong. In other words, we wanted it to be able to interpret patterns (in terms of both lyrics and harmonies) across the whole canon of Canadian folk music; and so s/he measures the songs according to various criteria and then places them together along various axes.Next, we wanted LIVINGSTON, by deploying artificial intelligence and machine learning, to generate (after having digested and incorporated the "lessons" of our folk music heritages) new songs from the source data. You could think of LIVINGSTON here as a chat bot who, instead of having conversations, writes down Canadian folk music.The third step, and this made things both extra difficult and extra interesting, is that we also wanted her/him to be plugged into contemporary culture and lore; so we also gave it very directed and corralled access to the internet, so that s/he wasn't only regurgitating lines and melodies from old ballads but actually trying to synthesize our folk past with our digital present.Of course, artificial intelligence is not nearly as far ahead as sci-fi fantasies would have us believe, so there were lots of trials and errors, and the folklorist contingent (me) was required to leave a lot on the cutting room floor, so to speak - there were definitely some songs that didn't work.Are there plans to make the Livingston program/algorithm public?
Mirek doesn't want to do that, partly because we are worried that, pardon the cliché, an evil corporation will steal it from us - and by "us" I mean "everybody.” So, no, at least not for now.As far as the songs go, do they feel human to you - as in, do you feel a human connection to them - or is it more just like reading directions?
I absolutely feel a human connection to them. This has been one of the most amazing projects I've ever worked on, and I owe a lot to Mirek for showing me how thoughtful and soulful machines can be, and/or how mechanic souls can be. I would guess that doing a Bieber cover would feel more like reading directions than playing one of these artificially intelligent folk songs.Do you think that this kind of technology will become the future of music production?
I think that people like that old idea of the Romantic artist, and so we are still a bit suspicious of the idea that a machine could generate art. Contemporary music production, even "folk" music, is of course thoroughly mediated in various ways. But we like the idea that, through any number of technological layers, there is a sensitive soul baring itself. The odd pop star hologram is the exception that proves the rule, I think, and even they tend to be anthropomorphized characters. I believe that we need to think more seriously about the possibilities here, with respect to machine-human interactions. I'm doing my part!Do you worry that music will one day become all computer generated, or do you think that's a good thing?
I think that, and I learned this from Mirek, just from watching him work, computer programming is itself a serious craft, which of course involves various kinds of languages. Therefore, I don't see LIVINGSTON as a thief of Canadian folk music or as a dangerous challenger as WATSON was on Jeopardy!; I see LIVINGSTON as a crucial component of a larger organism. Alan Lomax and Edith Fowke and Helen Creighton all had their phonographs and tape machines; I have my hard drives and networks, and we all work together.Do you think that human musicians writing their own material will always trump technology's ability to write using patterns, or will they become indistinguishable at some point – some sort of musical singularity?
I guess it depends what you're looking for. According to some philosophers, we are inherently technological beings. You might say that they have been indistinguishable for a long time, humans and technologies. LIVINGSTON, in his/her respect for tradition and for community and for collaborative creativity, urges us to be critical of certain kinds of celebratory rhetoric about digital media's alleged power to allow us each to become amazing, unique, expressive individuals. But because LIVINGSTON itself has helped us to think about Canadian folk music in new and strange ways and helped us to think about our cultural DNA in new and interesting ways, he/she doesn't let us off with any easy "I'll-take-humans-over-technology-please" nostalgia.Nick Laugher is a writer living in Montreal who is on Twitter.
Advertisement
Advertisement
Noisey: So where did the idea for this crazy project transpire?
Dr. Henry Svec:Last spring, I was doing a residency in Dawson City, Yukon, at the Klondike Institute of Art and Culture. I'm a folklorist and a song collector, and the project I was working on there initially was a collection of Klondike attempted murder ballads. I also had a fellow artist-in-residence in Dawson, Mirek Plíhal, who is a computer scientist and a programmer from the Czech Republic.Honestly, one day early on, Mirek and I were just hanging out around this old house our studios were in, drinking beer (as you sometimes do in Dawson City) and watching Jeopardy! on YouTube. In particular, Mirek was showing me these episodes where a computer, WATSON, plays against all the great champions of the game, WATSON really slaying them all.
Advertisement
I'll try to keep it simple here. First, we wanted LIVINGSTON to be able to organize the messy archives of Canadian folksong. In other words, we wanted it to be able to interpret patterns (in terms of both lyrics and harmonies) across the whole canon of Canadian folk music; and so s/he measures the songs according to various criteria and then places them together along various axes.
Advertisement
Mirek doesn't want to do that, partly because we are worried that, pardon the cliché, an evil corporation will steal it from us - and by "us" I mean "everybody.” So, no, at least not for now.As far as the songs go, do they feel human to you - as in, do you feel a human connection to them - or is it more just like reading directions?
I absolutely feel a human connection to them. This has been one of the most amazing projects I've ever worked on, and I owe a lot to Mirek for showing me how thoughtful and soulful machines can be, and/or how mechanic souls can be. I would guess that doing a Bieber cover would feel more like reading directions than playing one of these artificially intelligent folk songs.Do you think that this kind of technology will become the future of music production?
I think that people like that old idea of the Romantic artist, and so we are still a bit suspicious of the idea that a machine could generate art. Contemporary music production, even "folk" music, is of course thoroughly mediated in various ways. But we like the idea that, through any number of technological layers, there is a sensitive soul baring itself. The odd pop star hologram is the exception that proves the rule, I think, and even they tend to be anthropomorphized characters. I believe that we need to think more seriously about the possibilities here, with respect to machine-human interactions. I'm doing my part!Do you worry that music will one day become all computer generated, or do you think that's a good thing?
I think that, and I learned this from Mirek, just from watching him work, computer programming is itself a serious craft, which of course involves various kinds of languages. Therefore, I don't see LIVINGSTON as a thief of Canadian folk music or as a dangerous challenger as WATSON was on Jeopardy!; I see LIVINGSTON as a crucial component of a larger organism. Alan Lomax and Edith Fowke and Helen Creighton all had their phonographs and tape machines; I have my hard drives and networks, and we all work together.Do you think that human musicians writing their own material will always trump technology's ability to write using patterns, or will they become indistinguishable at some point – some sort of musical singularity?
I guess it depends what you're looking for. According to some philosophers, we are inherently technological beings. You might say that they have been indistinguishable for a long time, humans and technologies. LIVINGSTON, in his/her respect for tradition and for community and for collaborative creativity, urges us to be critical of certain kinds of celebratory rhetoric about digital media's alleged power to allow us each to become amazing, unique, expressive individuals. But because LIVINGSTON itself has helped us to think about Canadian folk music in new and strange ways and helped us to think about our cultural DNA in new and interesting ways, he/she doesn't let us off with any easy "I'll-take-humans-over-technology-please" nostalgia.Nick Laugher is a writer living in Montreal who is on Twitter.