The Hymn of Acxiom
June 10, 2013 8:42 PM   Subscribe

Singer/songwriter Vienna Teng has released a demo for The Hymn of Acxiom, a haunting, vocoded choral composition written from the perspective of a marketing database.

Teng is no stranger to songs written from unusual perspectives, and has a bit of an unusual perspective herself. At age 23, she quit a software development job with Cisco Systems to become a full-time musician. Recently, she completed a graduate degree in environmental studies at the University of Michigan (a move perhaps presaged in one of her songs), while continuing to write and perform music.

In an interview with Ploughshares, Vienna Teng discusses creativity, grad school, environmentalism, and Big Data.
posted by Alterscape (17 comments total) 51 users marked this as a favorite
I <3 Vienna Teng. Great song, but then again I expect no less from one of the most fascinating and thoughtful artists I've ever heard!
posted by jph at 8:50 PM on June 10, 2013 [2 favorites]

That's not a vocoder, really. The soundcloud page says she used one of these, which is more of a keyboard-controlled harmonizer:

Very cool track.
posted by jonathanhughes at 8:50 PM on June 10, 2013 [1 favorite]

Tomorrow I'll drive by Acxiom's downtown office, on my way to my office, half a block from one of their data centers. And I'll plop down in a war room my development team has set up, intended to last only a month but looking to be somewhat permanent. I just emailed this link to one of the two architects on my team, both of whom used to work there and got a glimpse of how they do what they do.

He has surrendered himself to the inevitability of big data. The other one seems to be on the fence. I'll be interested to see their reactions.

Regardless, thank you. That was very good.
posted by middleclasstool at 9:04 PM on June 10, 2013

This was inspired by listening to Morten Lauridsen

posted by weston at 9:20 PM on June 10, 2013 [1 favorite]

After a few listens, I still admire what she's done in a lot of ways, but think I actually regret the subversion of sacred-sounding emotive music in the piece.

I understand that it reflects an earnest and maybe even damn-near holy hope we've been projecting on to data processing and technology -- in particular if you listen to some voices from silicon valley.

I also understand that sometimes near-pitch delivery that flirts with the absurd is one way of delivering criticism.

But some things I just don't want delivered satirically, ironically, sarcastically. Apparently kisses and some choral idioms are among them.
posted by weston at 9:51 PM on June 10, 2013 [2 favorites]

Theano is a deep learning library by the LISA lab, under the auspices of that Yoshua Bengio-led lab. There is a tutorial for it, with some material based on a paper they presented at the 29th International Conference on Machine Learning: here.

Scroll down and listen to the unpreprocessed polyphonal generative fugue: sounds like a 6-year-old at the organ. At the very end, there is a mention of many different techniques they used for a better result, and a sample with preprocessed sound and other easy tricks. That sounds like a 10-year-old. That is, it seems that you would not necessarily be surprised if someone told you that a 10-year-old did this. At least, I wouldn't.

Is music where we are to build the Metropolis? Or merely our Metropolis-Hastings algorithms?
posted by curuinor at 9:54 PM on June 10, 2013 [1 favorite]

Let us all be grateful, that one day the amazon marketing database will one day resurrect us, extrapolating the minutia of long rotten conciousnesses from the data crumbs of our purchasing decisions.
posted by compound eye at 10:15 PM on June 10, 2013 [1 favorite]

This is wonderful. The sacred nature of the music and the faint unearthly sound of her voice really work together in the context.
posted by Gilgongo at 10:19 PM on June 10, 2013

Is it bad that I kept waiting for the drop?
posted by Doleful Creature at 12:09 AM on June 11, 2013 [4 favorites]

I think "marketing database" may just qualify for understatement of the year.
posted by wierdo at 2:20 AM on June 11, 2013

This reminds me of Imogen Heap's "Hide and Seek" quite a lot. If you like the sound of this you'll like that as well.

I can imagine that was the effect that she was going for but couldn't quite do with her chosen effects machine. Imogen's track sounds like it's done with a vocoder, with a vocal sample as the "carrier sound", while Vienna's machine uses some sort of advanced pitch-shifting technique.
posted by svenni at 2:53 AM on June 11, 2013

Oh, and she cites this guy as her main musical influence for this song.
posted by svenni at 3:02 AM on June 11, 2013

that imogen heap track was actually made with the tc-helicon voice live, which I think is a slightly different unit that does the same thing.
posted by jonbro at 3:36 AM on June 11, 2013

that imogen heap track was actually made with the tc-helicon voice live, which I think is a slightly different unit that does the same thing.

She uses that for live performance. For the recording, she used a DigiTech Vocalist Workstation. From
Containing only her vocoded voice, “Hide and Seek” is an unlikely first single born in a flash of inspiration. “My favorite computer blew up on me,” Heap explains. “But I didn't want to leave the studio without having done anything that day. I saw the [DigiTech Vocalist Workstation] on a shelf and just plugged it into my little 4-track MiniDisc with my mic and my keyboard and pressed Record. The first thing that I sang was those first few lines, ‘Where are we? What the hell is going on?’ I set the vocalist to a four-note polyphony, so even if I play 10 notes on the keyboard, it will only choose four of them. It's quite nicely surprising when it comes back with a strange combination. When it gets really high in the second chorus, that's a result of it choosing higher rather than low notes, so I ended up going even higher to compensate, above the chord. I recorded it in, like, four-and-a-half minutes, and it ended up on the album in exactly the structure of how it came out of me then. I love it because it doesn't feel like my song. It just came out of nowhere, and I'm not questioning that one at all.”
So, for all intents, it was made using a vocoder, except the original vocoders were purely analog.

Vocoding was (and probably still is) surprisingly common in telephony. Basically, *everything* going over a long distance line is (or at least, was) vocoded into an 8kbps digital stream and reconstructed with remarkable fidelity, and you could drop to just above 5kbps with acceptable quality. Below that, it fell apart pretty rapidly, but given that a convention B channel on a trunk is 64kbps, vocoding gave you 8 circuits per B channel, or 192 on a T1 and 240 on an E1. Back when the main intercity US trunks were T1/T3 and the microwave equivalents, this was critical to handling the load.

A vocoders without an actual voice input was called the voder, and was one of the first voice synthesizers. The problem was it was amazingly hard to use -- you had fourteen keys, a wrist bar, and a pedal to manipulate, as well as a pitch preset knob acting as a starting point. But basically, it was a vocoder, but you directly controlled the filter inputs.
posted by eriko at 4:20 AM on June 11, 2013

I've loved VT for years now, and very pleased that she's still working through a time when music isn't her main focus. She did a series of StageIt "concerts" last year where she was showing off the vocoder-like toy. It was a little intimidating to see her casually layering sounds with that in realtime.
posted by tyllwin at 7:45 PM on June 11, 2013

Chilling... and prescient.

I am liking this far more than I should.
posted by PROD_TPSL at 2:49 AM on June 12, 2013

« Older How A War Hero Became A Serial Bank Robber   |   Peter Sellers documentary 1969 Newer »

This thread has been archived and is closed to new comments