Saturday, January 26, 2013

Were Jonathan Coulton's Actual Audio Tracks Reused by Glee?

No doubt you've heard about the Coulton/Glee controversy by now. If not, you can Google it. It's all over the place: Wired, NPR, CNN, and I won't rehash the details here. I really admire Jonathan Coulton as a musician, and I enjoy the majority of his work. I can safely say that I wouldn't be where I am today without his influence and inspiration. As for Glee, I've never seen it before, and up until this story broke, I had zero opinion of it and completely ignored its existence. I plan to go back to that status quo ASAP, but for the moment, the status is NOT quo.

As a personal disclaimer (and as a warning to more sensitive readers): I need to mention that the song in question, "Baby Got Back" (BGB), is one of two Coulton songs that I generally try to forget the existence of, due to what I consider offensive content (though musically speaking, I'd take his folksy arrangement over the original rap any day). But enough about me. On to the analysis!

MOTIVATION

People are talking about what legal recourse Coulton has, and the consensus seems to be that, due to some kind of legal loophole which I don't fully understand, his arrangement may not be protected by copyright. I am not going to rant about just how broken the system is, to allow such a distinctively unique arrangement to be stolen note-for-note for commercial purposes, without so much as an attribution or an advance notification, while simultaneously getting people sued for using a song as a backing track in a non-commercial setting with full attribution (shouldn't they be thankful for the exposure?).

However people are also saying that Coulton would have a much stronger case if his actual original backing track was used. The tracks sound so similar that Coulton even asked for help to determine if this was the case. Several people have responded using youtubedoubler, or by creating versions placing one track in each ear, or alternating between versions beat by beat. These convincingly demonstrate the versions are extremely similar. At the very least, it is obvious that the arrangement is the same -- aspects like key, tempo, instrumentation, rhythm, minor lyrical changes, and even the exact same notes are virtually indistinguishable. But beyond that, are the recordings actually the same? Can we even determine that?

My friend and fellow Coulton fan, Paul R. Potts, wrote a lengthy blog post where he searched for Coulton's missing "duck quack" sound effect (which Coulton used as a word censor), and tentatively concluded that the tracks looked the same, though admitting it was hard to tell. I found it a very interesting analysis, but I'll admit, I was not completely convinced. So I decided to take a crack at it myself. Rather than look for the quack, I decided to focus on minute performance details that would be easy to overlook and hard to hide.

METHODOLOGY

Unfortunately, trying to prove these two very similar things are identical (except for differences in vocals, mixing, and mastering, of course) is actually very challenging, because what you are really trying to do is prove the absence of any differences. Proving that something doesn't exist is difficult, because absence of evidence is NOT evidence of absence. It's analogous to trying to prove there is no Bigfoot -- just because you can't find evidence doesn't mean it's not hiding out there somewhere under the pines. With that in mind, the approach I took was to try and find evidence to prove that these are different tracks -- at least until I got bored of looking, or ran out time. Or actually did find a difference.

And, unfortunately for Coulton, I may have found a difference. Or several of them, actually. At least in the bass part. But I'll let you decide.

In 2008, Jonathan Coulton released a special album entitled "JoCo Looks Back". This was released as part of a fundraising campaign for Creative Commons. For a donation, you got a USB thumb drive containing source tracks to his 20 most popular songs ("chosen via a complicated algorithm") -- including BGB. I have this disk, and I used the raw, unmixed tracks from it to analyze Coulton's version. (Note: If Glee had these source tracks -- and I'd be very surprised if they aren't available via a torrent somewhere -- then they wouldn't even have to edit out the duck quack.) To analyze the Glee version, I downloaded a copy of the original full track that leaked onto YouTube before the episode aired (since this is just a cover of Coulton, there shouldn't be any issue with that). What actually got aired was apparently cut into a shorter version, but I believe the full version is what is selling on iTunes. And it lines up better with the original.

All the tracks were imported into Reaper, and lined up. Then the spectra were graphically analyzed using the included "gfxspectrograph" plugin (using a "blackman-harris" window with an FFT size of 32768, whatever that means). Originally, I was going to look at the rhythms of the banjo part. I figured the part was complex enough to have timing inconsistencies, and that it would be easy to see or hear if something was off. However, I quickly realized that the bass part was much more visible in the spectrographs. In order to hear the bass part better, I also applied a low-pass EQ filter around 400-500 Hz to the Glee track. To be fair, I applied the same filter to JoCo's bass track. This very well may not sound good on speakers with a weak bass response.

RESULTS

Listed below are a few spots where I found subtle differences between Coulton's bass track and Glee's. Each is accompanied by a description of what's going on, along with a brief mp3 clip, a spectrograph, and a transcription (as I hear it) from both versions. Note that the transcriptions are NOT canon. They are just what I hear and see. It is possible that I'm misinterpreting something, especially in the Glee version, where the part is buried in a larger mix.

The spectrographs below may look like a dayglow-pterodactyl, but here's how you read them: the horizontal (x) axis is time, and the vertical (y) axis is frequency. The bright purple lines represent a strong amount of sound at a certain pitch. The red and yellow are softer amounts of sound, and green and black are essentially silence for our purposes. Due to the way sound is generated, the lowest line is the "fundamental" pitch that you actually hear, while the higher parallel lines are the quieter "overtones" that give a sound its distinctive timbre. We'll ignore them since they are at integer multiples of the fundamental frequency, and don't provide any additional meaningful information. Vertical lines represent unpitched "noise" -- in the solo bass track, this is probably the sound of JoCo plucking the string, or moving his fingers around. In the full Glee track, this is probably the drum beat, or other instruments. In the full Glee track, you can see sounds from all the other instruments playing along, so its a lot busier than Coulton's solo bass track. But the bass line is still fairly easy to pick out.

EXAMPLE 1

The first place we'll look is the chorus. The bass part here has a fair bit of ornamentation, and its not likely to be exactly the same between the two by coincidence. The prechorus ends with the bass on G (the dominant) then moves up a step to A, to begin the chorus in the relative minor. It continues its upward trajectory by sliding up to an E, and ascends through F, G, and the high A before jumping back down an octave. This repeats 3 times before returning to the home key of C to finish and prepare for the next verse. To make the image a bit easier to follow along with, I labeled the main notes in the first appearance of this motif. The other two repetitions are unlabeled, but have the same shape. What is interesting is the little notes in between the main notes. Unfortunately, you'll need to click to view at full size if you want to actually see the details. I identified at least four possible differences.

At (#1) we have a little scale filling the area between A and E. In Coulton's track it is clearly a B and a C, without the D. If you look at the spectrograph, it is clear that there is a "gap" just before the E (where a D would be). Its hard to tell from just listening, but if you look at the graph of Glee's version, the D seems to be present. Instead, the gap is immediately after the A (where the B would be). In both versions, this is the case for all three repetitions.

At (#2) Coulton's version has another little run of notes between the E and the F, clearly consisting of a D and an E. I have listened to this section repeatedly in the Glee version, and hear absolutely no D; they just restrike the E. Checking the graph, I also see no evidence that would indicate that a silent D is present. This seems pretty obviously different to me, in each of the three repetitions.

At (#3) the Glee version seems to have a much more lengthy and prominent slide (glissando), at least in the graph (are they using a fretless bass? I don't know, but I don't think JoCo is). This could very well be a trick of the ear though, and/or an artifact in the image. I'm not convinced this is a good example.

I almost missed it the first time, but (#4) seems fairly definitive as well. After the repeated section completes, JoCo hits the C and holds it for a full two beats, contrary to the following measure (and numerous other places throughout the song) where he hits it a second time after a beat and a half. In the Glee version, I am fairly certain, from both listening and viewing the graph, that this note is restruck the first time.

Coulton:
Coulton's solo bass track - Example 1 (click to enlarge)

















Glee:
Glee's full track, emphasizing bass line - Example 1 (click to enlarge)





EXAMPLE 2
This example is notable because it's a case where there seems to be an extra note in the Glee recording that is obviously not in the Coulton recording (thus something was added rather than removed). This section occurs shortly after where the duck quack would have been, and, oddly, doesn't seem to occur in the first verse. This fact is interesting in and of itself, since JoCo's track seems to liberally copy and paste between verses. But what you can hear (and see) rather prominently in the Glee track is a quick additional E (circled) between the A and the F. This occurs under the phrase "'cause them punks" and is not present in JoCo's version at all. There may be additional differences with the rhythms in the Glee track, but it's hard to tell in the full mix.

Coulton:
Coulton's solo bass track - Example 2


Glee:
Glee's full track, emphasizing bass line - Example 2

CONCLUSIONS

While it is difficult to say for sure, it would appear that, at the very least, the bass part differs ever so slightly in these two recordings. This would seem to indicate that they were recorded independently. While you could theorize that, with enough digital wizardry, you might be able to transform one into the other, there quickly comes a point where its much easier to just re-record the track. This is especially true if you have a half-way decent instrumentalist on hand, and an army of lawyers reminding you that you are within your rights to do this, as long as you don't reuse the original recording.

Of course, I can't say anything about the other tracks, since I didn't analyze them. For all I know, every track except the bass could be reused from Coulton. But if they did use Coulton's other tracks, only with a new bass line, then they would have had to use the individual source track version that I'm using (see update). If this were true, then there would also be no duck quack to find. It also begs the question as to why they would re-record some parts but not others. Certainly, it would be faster (i.e. cheaper) to produce, though riskier if you got caught. But then why bother re-recording the bass part specifically? I would think that if you've got the time and the talent to re-record one part, you may as well do them all.

As much as it pains me to admit it, and despite the near perfect match, I cannot convince myself that this is Coulton's original track. At least not the bass part. Personally, I would have preferred to have found nothing of note; to continue to believe that Coulton's tracks were stolen in such a way that he could at least legally defend himself. But wishing doesn't make it so, and I have to follow what the evidence shows. And in all honesty: it only makes sense that if Glee is smart enough to know what they can get away with, they'd also be smart enough to avoid making a mistake like that. Although stealing was actually a pretty dumb move in the first place.

UPDATE (1/27)
Paul pointed out (in a discussion on Coulton's forums) that, because the bass track is centered, it would be removed if you used the phase inversion technique (often used to remove vocals, which are also centered). This would obviously be a valid reason for the bass track, specifically, to be re-recorded (along with some percussion which was also centered). It would even allow for the possibility that the full song (rather than just the source tracks) was used -- either the version with vocals, or the karaoke version -- but with a new bass track plopped on top. So whether the rest of the track is the original or not is still up in the air, though it is still a definite possibility at this point.

5 comments:

angelastic.com said...

Hmmm, interesting. (Nice Under the Pines and Mandelbrot Set references!)

Paul Potts said...

I can definitely hear the difference you're talking about, but can't really account for it. My best guess at this point is still that they used the karaoke track (hence the duck and identical banjo) _or_ did the vocal-kill trick on the original song, but I don't know how to account for the bass. If I get a chance I will listen to all the different versions a little more today.

Dave Leigh said...

I don't think it was necessary to use the karaoke track. I think they used the source tracks and replaced what they felt like replacing. I don't really hear differences in the tracks that are most likely to be different (such as the banjo). And despite anything they may say to the contrary, GLEE had access to the source tracks. I just did a Google search after reading Caleb's post suggesting it may be available via torrent, and did in fact find at least 12 seeds.

Thus, the individual tracks were available, and if GLEE replaced only some of them, they're still liable for the rest.

Caleb said...

I just added an update -- The vocal kill technique may have also killed the bass, since its centered (thanks Paul!).

zhochaka said...

Several points come to mind.

1: The original music and individual recordings are separately protected by copyright law. There are detectable differences in performance even when cover versions are trying to duplicate an original. It's why orchestra and conductor matter to classical music purchasers.

2: From the reports of what Coulton did--he added a melody for one thing--this is closer to setting a poet's words to music than recording a cover version. From the reports of what Glee did, they used Coulton's music. It almost doesn't matter whether they used the source tracks or not.

3: Hymns sung in church often have a recommended tune, but there are any number of alternative tunes possible. Church Choirs often have fun with the possibilities, such as the carol "While Shepherds Watched" to the hymn tune "Cranbrook". There is solid copyright precedent for splitting tune, arrangement, and words into distinctly-owned entities. And the setting for Glee might almost be made for that sort of game.