Texan talking Obama

« previous post | next post »

One of the things that I acquired over the holiday was a Talking Obama Figure ("Hear His Historic and Inspirational Words") from Gemmy Industries Corp. of Coppell, TX, "the worlds largest provider of all your favorite seasonal decor, animation entertainment and lighting products". This is one of a large number of other Gemmy talking toys, from the "Animated Talking Head Skeleton" and the "Gemmy Talking Dancing Hamster 97 Kurt Busch", to the "Dora the Explorer Talking Christmas Doll in Santa Outfit" and the "Animated Talking Bouncing Van of Love", and  Gemmy's monster hit from 2000-2001, "Big Mouth Billy Bass".

As you press of the red button on its pedestal, the Talking Obama Figure cycles through nine passages from president Obama's speeches. What struck me first about this collection of inspirational oratory was that it's performed by somebody else.

I noticed this because the performer, though he imitates the president's characteristic prosody, has a distinct South Midland accent. Thus the first passage is from the 11/4/2008 Grant Park victory speech:

The road ahead will be long. Our climb will be steep. We may not get there in one year or even one term, but America – I have never been more hopeful than I am tonight that we will get there. I promise you – we as a people will get there.

Listen to the Gemmy figure's pronounciation of get and I in this excerpt:

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

And compare Obama's performance of the same words:

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

This left me with several questions. Why did they use someone else's voice? Was it for obscure technical reasons, or because of IPR issues with voice that are somehow different from text, or what? And once they decided to use another speaker, why did they have him imitate the president's prosody but not his vowels? Was it because they can't hear the difference, or because they thought their customers would prefer a somewhat Texan version of Obama?

And most important of all, what is the connection between NASCAR driver Kurt Busch and a dancing hamster? This seems like a random selection from the cross-product of American memes, a process that might produce future products like the Talking Tiger Woods Lolcat or the Alvin and the Chipmunks Talking Dancing Mitt Romney.


  1. Timm! said,

    January 6, 2010 @ 11:27 am

    Assuming that the recording actually was made in Texas (which is not exactly a given), I'm going with a)technological limitations and b)they couldn't hear the difference.

    Considering how compressed the files would be, and how crappy the speaker is, they probably had fairly specific needs for the file size/bitrate/format/etc to make it sound acceptable and it was easier to do an in-house recording than try and find what they needed on the web.

    If the recording was in house, you know that most people can't "hear" their own accent/dialect so as long as the prosody was more or less Obama's, the recording sounded "correct."

    [(myl) I'm somewhat skeptical of the "file size/bitrate/format" theory — formats and bitrates are easy enough to change. But it's possible that a compression method is used that makes special demands on the initial recording, especially if the chip used is old enough to have very limited memory. That was certainly true back in the days of the 1978 TI Speak&Spell (which I believe used 1 kb/sec LPC coding), and something similar is implied (for a different sort of coding) by this 2001 article:

    Although we'll be downsampling a lot later, starting with high-quality samples is a must. The recording sessions are undertaken in an acoustically "dead" studio environment using 44kHz 16-bit sampling … Sometimes, the audio engineer must "factor" sampled sounds to reduce ROM space usage by merging redundant sounds … Factoring is exhausting work that involves cutting and pasting fragments of syllables, enhancing plosives, testing various combinations of concatenated sounds to make sure they sound like coherent words, and so on. It can take a couple of weeks of full-time effort to get a realistic-sounding result for just a few words, so try to avoid factoring further than necessary.

    These days, I'd expect that on-chip storage for a minute or two of compressed speech at more standard rates (e.g. 32kbits/sec ADPCM = 240kB/minute) would not be a problem, even for very cheap chips. But I could well be wrong.]

  2. John Lawler said,

    January 6, 2010 @ 11:33 am

    I want an "Alvin and the Chipmunks Talking Dancing Mitt Romney", please.

  3. Słowosław said,

    January 6, 2010 @ 11:37 am

    Am I the only person who saw "South Midlands accent" and instinctively thought that they got a someone from Kettering to imitate Obama?

  4. Mark P said,

    January 6, 2010 @ 11:39 am

    I'm not sure whether it's because you specifically mentioned "get" and "I", but the difference seemed very noticeable to me. Whoever did the voice, it was pretty good, at least without Obama's actual speech to compare it to.

    I have wondered about Obama's accent since I heard him refer to getting the promised dog on the Letterman (I think) show. His pronunciation of "dog" was very close to my southern pronunciation, and unlike what I would have expected. I would normally associate that pronunciation of dog with the fake Obama's pronunciation of "get" as "git". It's possible Obama was intentionally drawling "dog" because he and Letterman were joking about his promise to his daughters that he would get a dog for them once they were in the White House.

    On the other hand, is it possible that there is a remnant of southern pronunciation in the northern black population that Obama might have mostly suppressed?

    [(myl) When I first heard the figure's first passage — the one cited above — I thought it was really Obama's voice, because of the context and the prosody. But I thought "Whoa, does he really have monophthongal I and git rhyming with bit? How did I miss that?" So I checked …]

  5. Skullturf Q. Beavispants said,

    January 6, 2010 @ 11:56 am

    I'm not sure if "South Midlands" was a typo for "South Midland", or if both are considered correct. When I read "South Midland", I think Dallas/Tulsa/Wichita, but when I read "South Midlands", I think Northampton/Milton Keynes/Bedford.

    [(myl) Traditional American dialectologists use "Midland" (north and south) as a regional term. The -s was a typo on my part.]

  6. Ginger Yellow said,

    January 6, 2010 @ 11:58 am

    Am I the only person who saw "South Midlands accent" and instinctively thought that they got a someone from Kettering to imitate Obama?


  7. Peter Taylor said,

    January 6, 2010 @ 12:08 pm

    The answer to the first question could be trying to avoid legal issues. When the King of Spain told Hugo Chávez to shut up ("¿Por qué no te callas?") it became a hit ringtone in Spain – but the ringtones used actors to avoid being sued. I forget the details – image rights or something. In the US they probably wouldn't be infringing any rights if they used real recordings, but they might consider the risk that they would be sued too high even if they believe they would ultimately prevail in court.

  8. Rob P. said,

    January 6, 2010 @ 12:12 pm

    I can't do the linguistic analysis, but I can do some IPR analysis. It's possible that they were unable to find an original recording that didn't belong to someone like a news agency. Though the government isn't eligible for copyright protection, and therefore an official recording would be free for public use, a recording made by another party would have some protection. It's also possible that they were working the other direction in that a recording made for them by their voice actor would have protection that a publicly available recording would not. This would make it a bit harder for someone to make a knockoff device by simply copying the chip with the recorded material.

  9. Gordon P. Hemsley said,

    January 6, 2010 @ 12:16 pm

    Nobody has mentioned that in addition to monophthongal vowels, the fake Obama also is less rhotic the real Obama.

    The part of the quote that has all these differences together is:
    "but America – I have never been more hopeful".

    The "I" (or, more specifically, the "I have") and the "more" really stick out.

  10. Tim Silverman said,

    January 6, 2010 @ 12:22 pm

    @Słowosław: I promise you – we as a people will get there, me ducks.

  11. Graeme said,

    January 6, 2010 @ 12:34 pm

    Obama has copyright in the words as written; and also distinctly in the particular delivery. It's easy to prove fair use in say reprinting a chunk of the words for comment. Less so for applying the soundtrack to such a slightly tacky and blatantly commercial purpose, where the risk of being sued is higher.

    Of course the illogic of this is that the 'Obama' doll is both visually and audially off kilter: and hence POTUS's moral rights, such as they are, are even less respected.

  12. Mark P said,

    January 6, 2010 @ 12:35 pm


    Ha! It never really hit me before that the "di" in "diphthong" actually means two, I suppose mainly because I never looked it up before. I feel like my day is not wasted.

  13. Chud said,

    January 6, 2010 @ 12:52 pm

    The recording of actual Obama has crowd noise in it. Maybe there is no recording of this without crowd noise. Maybe it's a lot easier to re-record with an actor than to take out the crowd noise? Presumably the other phrases it says were applause lines also.

    I think that if I had heard this without any hint of accent problems, I would have just assumed at first that it was really Obama, but it sounded funny due to cheap sound-delivery technology. Of course, I grew up and currently live in central Texas.

  14. Brian said,

    January 6, 2010 @ 3:20 pm

    Even after having been primed, I could barely discern the difference. Had I not been looking for them, I certainly would not have heard them.

  15. Catanea said,

    January 6, 2010 @ 3:38 pm

    But surely it's obviously not his VOICE. Accent isn't all there is to a voice. There's a (somebody's going to love this) "tinny" quality to the doll's actor's (or its recording) delivery in comparison to a rich, aged, woody quality in the genuine recording. Sorry. I looked deep into my synesthesiac soul before choosing those words, and decided they had a better rep than any I might pick spontaneously…

    [(myl) I recorded the Gemmy voice as it was played through the tiny, cheap speaker in the base of the figurine. That alone is going to give it a "tinny" quality. There's also clearly been some pretty extreme dynamic range compression applied, and probably some serious spectral EQ to help keep it intelligible through the little speaker on top of whatever coding is being used. So I don't think we can put a lot of stock in the perceived voice quality that results.]

  16. mgh said,

    January 6, 2010 @ 3:42 pm

    I'm not trained to distinguish pronunciation but the first audio clip sounded nothing like Obama to me — hearing it, I would have guessed this was a Bill Clinton doll. (It even has a bit of his rasp — maybe doing Obama voice-overs is one of Clinton's post-presidential side projects?)

  17. Nathan Myers said,

    January 6, 2010 @ 3:56 pm

    mgh: Probably the voice actors who used to do Clinton have had to branch out. Maybe the Midland bits are a hangover from Bush gigs.

  18. Garrett Wollman said,

    January 6, 2010 @ 11:38 pm

    @Graeme: We generally don't do "moral rights" in the U.S., although some states (notably California) do recognize a "right of publicity". Obama being a "public figure", though, all bets are off.

    @MYL: Reading your post reminded me of how off-sounding all of the British attempts at Obama impressions that I've heard are, even the ones from accomplished impressionists like Jon Culshaw. (I assume American impressionists, if there still are any, sound just as off to BrE/OzE ears when attempting to imitate UK/Aus notables.) I wonder about the causes of impression failure: is it simply that the actor lacks the ability to imitate certain accents, or that he is unable to hear the difference between his impression and the real thing (as theorized by another commenter above wrt your subject), or simply that he knows he doesn't have to get it perfect because most of the audience can't tell. I further wonder how accurate impressionists in general really are — I have this nagging feeling that they greatly exaggerate whatever accent features and vocal mannerisms are perceptible and/or humorous to them.

    [(myl) With respect to your question about trans-Atlantic impressionists, we should ask a professional dialect coach like Amy Stoller, but it makes sense that (other things equal) it's harder to imitate a someone who speaks a language or dialect that you're not familiar with. That's what makes Adrian Celentano's Elvis re-synthesis so impressive…

    As for the question of what it is that good impressionists do, it's clearly analogous to what visual caricaturists do — it's not so much an accurate representation of how someone looks or talks, as an effective exaggeration of selected features that are somehow characteristic. I don't think that there's a very good explanation (in the scientific sense) of what that really means, in either case.]

  19. Mark F. said,

    January 7, 2010 @ 12:40 am

    I'd say that the American you're least likely to be sued by is the president. The political cost isn't worth whatever he could gain from winning. The real copyright issue is with the people who made the recordings.

    [(myl) It's certainly true that if there were any copyright concern in this case, it would be with the broadcasters and the like. But I suspect that this is unlikely to be the reason for re-performing the quotes, because the processing involved in putting the recording on the device would make it impossible to determine which of the many alternative sources it came from.]

  20. arc said,

    January 7, 2010 @ 12:43 am

    I'm used to 'Midlands' referring to the English ones, so I was all wtf? over this at first.


  21. Garrett Wollman said,

    January 7, 2010 @ 2:14 am

    In any case, I doubt that the news organizations actually have any copyright in their recordings of the speech, as distinct from their employees' commentary about it; it seems to me this would clearly fall under the "slavish reproduction" doctrine. I'm not so sure that Obama wouldn't have any copyright in the speech, however, since it was made before he legally became so much as president-elect-apparent. Does the Presidential Records Act apply to candidates, even if successful? (This is probably a more appropriate question for a legal blog than a language blog.)

    [(myl) With respect to your first point, your argument is a logical one, but I believe that it's not how the American interpretation of copyright law now stands. For example, CSPAN claims copyright over its recordings, as I understand it, even the recordings of legislative sessions and the like. This came up because we at LDC wanted to collect and distribute some of this material for research purposes.

    With respect to your second point, I'm certain that political campaign speeches are not in the public domain; but as a practical matter, I suspect that their owners usually want to encourage maximal distribution rather than to derive any revenue. It might be possible in principle for a politician to try to use copyright to restrict distribution of a speech that turned out to be embarrassing, but I can't imagine that this would work as a practical matter.]

  22. Bill A said,

    January 7, 2010 @ 8:17 am

    re: To extend this thought: "This seems like a random selection from the cross-product of American memes, a process that might produce future products like the Talking Tiger Woods Lolcat or the Alvin and the Chipmunks Talking Dancing Mitt Romney."

    It has produced the film "Avatar".

  23. Fetcher said,

    January 7, 2010 @ 1:10 pm

    Mark P:
    "Ha! It never really hit me before that the "di" in "diphthong" actually means two, I suppose mainly because I never looked it up before. I feel like my day is not wasted."

    My epiphany came when I was looking at the word "digraph" and thought…ZOMG! The "di" in dipthong is a prefix! It all has to do with pronunciation (I say dip-thong, not di-phtong), so that sneaky "di" hides itself from us.

    As for the post, I hear the difference after the word get…the "there" sounds off to me.

  24. Mark P said,

    January 7, 2010 @ 1:30 pm

    @Fetcher: When you become familiar with words from hearing them, it sometimes hides things. That happened to me with "hy-DRO-genated". It didn't occur to me for a long time that they were talking about hydrogen-ating things, or, as my chemist friends might explain it, getting rid of double bonds and filling all the available bonds with hydrogen. If only all the margarine commercials had said "no hydrogen-ated oils".

  25. Nathan Myers said,

    January 7, 2010 @ 2:18 pm

    Fetcher,Mark: The classic case of this is helicopter, helico- (screw) -pter (wing). One could say the Brits have a lot to answer for, but I think they were just trying to keep word coiners honest. I doubt it helped.

    Speaking of coinages, there's a new word up at http://urbandictionary.com/define.php?term=wanklage

  26. Mano said,

    January 11, 2010 @ 4:52 pm

    I think the decision to re-record was probably an issue with the reverb and crowd noise present in the original recording (as Chud also mentioned). There may be IPR issues as well, but the crowd noise would be enough for me to do the same thing if I were in their shoes (if someone was pointing a gun at my head and demanding that I make the best talking Obama figurine possible, for instance). Linguistic authenticity was probably not the central concern in its design either. It is recognisably an imitation of Obama's voice, and a slightly hyper-real imitation of Obama is perhaps more in tune with the figurine's appearance.

  27. Steve Ross said,

    December 30, 2011 @ 5:52 pm

    I'm the guy that did Obama's voice. I'm a product designer at Gemmy (I illustrated/designed the figurine as well).
    At work I have a reputation for impressions (here are my Presidents JFKthruObama):
    Thus, I was saddled with doing the voice in our studio. At the time I had never really done him. So I listened to him on YouTube for a while and just jumped in and did it.

RSS feed for comments on this post