My Car Has A Speech Problem: I’m Working On It!

It seems that’s it’s only a matter of time before we’ll spend as much time talking to machine as we do to people – perhaps even more so. As John Malkovich continues his televised love affair with Siri, I’m getting to grips with using Microsoft Sync  in my new car. Yes, not only can I now shout at my car when I get frustrated but it can talk back!

Thanks to the advances in speech recognition and consumer technology, I can now press a button in my car to make a phone call (“Call Bob at home”), find a restaurant “Find Italian Restaurant”), find my way home (“Directions Home, please”), and listen to any of the thousands of songs on my trusty old iPad Classic (“Play Album ‘Mi Bossa Nova'”). I’m pretty sure I’m close to asking my toaster to make me a lightly done whole-wheat bagel or telling my ice dispenser to make me a White Russian [1].

Sadly, it appears that speech recognition is not quite at the same level as human beings just yet. It appears my car is having a few problems with comprehension and expression. I’m not at the point just yet that I can provide a detailed phonological analysis but I intend to do assessment over the next few weeks.

My first hint that my car may have a disorder came after asking it to “Play track, ‘Fugazi.'” This is from the 1984 album Fugazi by the UK Prog Rock band, Marillion. For those unfamiliar with the word, it’s apparently US army slang and an acronym of “Fucked Up, Got Ambushed, Zipped In,” which basically means everything has gone to pot, the shit has hit the fan, and the situation is unapologetically dire. The word is pronounced foo-GAH-zee (/fu’gɑ:zi/) or perhaps foo-GA-zih (/fu’gæzɪ/) but either way, it’s a /fu/ sound for the intial CV.

Unless, you are my car. When I tried to ask it to play the track, it seemed to be unable to understand, and it was only when offered me some alternatives that I worked out why; it asked me if I want to listen to FJOO-gah-zee, with the stress now on the first syllable, and an epenthetic palatal approximant, /j/.

Poor thing! Rather that seeing “fu” as being pronounced like “futon,” “further,” “full,” or “fun,” it seems to think it should be treated as the “fu” in “fuel,” “fuchsia,” “fugitive,” or “fume.” Of course, the fact that the pronunciation of “fu” can vary between /fju/ and /fu/ is what’s likely to be causing the problem. The system’s “brain” presumably contains rules that are applied to text strings, and one of those is something along the lines of [“fu”/fju/] and there may then be a list of “exceptions” – of which “fugazi” isn’t a member.

It’s also having a hard time with hearing Hounds Of Love [2], an album brought out by Kate Bush a year after Fugazi. Instead of Hounds Of Love, I get “Playing ‘Pretzel Logic,'” an admittedly terrific album from Steely Dan but irritatingly not the one I wanted!

So, I’ve decided the car needs therapy. It will doubtless be challenging because I can’t show it pictures and ask it to name things, nor can I give it any test that requires a pointing response (“Show me which one is a cat/hat.”) What I can do is ask it to play a selection of tracks and then when it gets the wrong one, jot down the error. Then, while the erroneous track is playing, I can ask it “What’s this?” and see how it pronounces it – which is basically how I discovered the “fugazi” misarticulation.

What I’m really looking forward to is expanding my practice to include the speech and language needs of non-sentient, pre-sentient, and semi-sentient machines that have issues. Surely as more technology becomes speech enabled, more technology will have problems.

Thinking back to my childhood, I suspect I may have stumbled on this while watching the very early (and very black-and-white) Fireball XL5, a UK puppet-based TV show about the adventures of Mike Zodiac and the crew of his spaceship, the Fireball XL5. Sure, the production values were not quite up to those of Prometheus, but this was a long time ago and as a child, your ability to suspend disbelief is stunningly high. One of the crew was Robert the Robot [2], who was made of see-through plastic, had a head like an upturned bucket, and had a problem with intonation and the habit of adding a /Ə/ vowel to the end of a word. Here’s a link to a YouTube clip where you’ll hear a sample of his speech:
 Maybe the plight of that poor mechanical man planted the seed of my becoming an SLP.

In the meantime, if any of you, dear readers, have a Ford with the Sync option, I’d love to hear from you with any examples of mispronunciation that your car has produced, and any mishearings it has demonstrated (like Pretzel Logic for Hounds of Love.)

Once I have a large enough sample, I’m planning to do an analysis and then offer it to the nice folks at Microsoft. That might get me a polite thank you letter or, if Bill Gates is feeling super kind, a new Windows 8 “Surface” when they release it.

Now all I need is for my toaster to develop a lisp…

[1] One part vodka, one part Kahlua, one part cream. If made in a glass directly, drop a few ice cubes in first, followed by the vodka, then Kahlua, and finally the cream. Stir gently. If made in a cocktail shaker, just be sure to fill it; one is never enough, especially while sitting on the porch on a cool summer’s evening. Oh, and don’t skimp – buy good vodka.

[2] The actual track, Hounds of Love, contains a clip at the very beginning that says, “It’s in the trees. It’s coming!” This is from an excellent 1957 movie from French director Jaques Tournier called Night of the Demon (UK) / Curse of the Demon (US). Described by one critic as “one of the most intelligent and thoughtful horror films ever made,” it’s a black-and-white horror flick about a skeptical American psychologist, Dr. John Holden, who is cursed by an ever-so-English cult leader, Julian Karswell. It’s an amazingly atmospheric film and the fact that it is shot in black and white actually enhances it.

[3] The diminutive form of Robert is, of course, Rob or Robby (or Robbie, as our Scottish friends would insist). Anderson’s choice of name may have been influenced by the 1956 movie Forbidden Planet, a loose remake of Shakespeare’s best play, The Tempest. The movie has the first appearance of Robby the Robot, who was destined to become an iconic cast member for the classic 1965/66 series Lost in Space.

Robby the Robot

Robby the Robot

Isaac Asimov also had a character called Robbie in a story of the same name back in 1940. However, the first use of Robbie is, according to Wikipedia, from a 1935 story by Lester Dent and Walter Ryerson Johnson called The Fantastic Island. Robbie is a mechanical version of the story’s hero, Doc Savage, used by him to confuse his evil foes.


7 responses to “My Car Has A Speech Problem: I’m Working On It!

  1. Bill Binko (@lessonpix)

    One of my biggest laughs ever was in a Toyota Avalon (2006?) on the way from Ft. Myers back to Tampa. My boss, Hans, who was so proud of his voice activated nav system says (clear as day)
    Hans: “Home”
    Car: “Ski resorts will now be shown on screen”
    *chuckles throughout the car*
    Hans: “Home!”
    Car: “Ski resorts will now be shown on screen”
    *hilarity ensues*

    Good luck on your plan 🙂

  2. Actually, Speech Dudes, I think you could be wrong about “fugazi”. It is, to my knowledge, a slang term for a fake diamond. How do I know this amazing fact?!? Why, of course, that font of knowledge, the Ian Fleming novel. In this case, it was Diamonds are Forever and, if I recall correctly, it was pronounced /fugeIzi/. It was 1956 when the novel was released and your car being Generation Y (or Alpha, depending on your viewpoint) probably hasn’t read anything before Nirvana released “Nevermind”…

  3. Oh, and I think it’s spelt/spelled “fugazy”…

    • Thanks for the update on “fugazi,” Alex. I haven’t read “Diamonds Are Forever” so Fleming’s use of it is new to me. If, as you say. it was published in 1956 then it’s likely to have been around before that, and therefore pre-dating the Vietnam War. I aw aware that the Italian word “fughesi” means “fake” or “false,” and so that could easily apply to diamonds. I can also see the phonetic similarities – bearing in mind that there are often slight changes when a word gets borrowed from one language to another – but the link between “fake” and “messed up beyond repair” isn’t very direct. It may be that “fugazi” as used as a slang word is a backronym – a definition invented to fit a word. So both “fughesi” and “fugazi” can exist separately but have different meanings. It may also be that “fughesi” and “fugazi” are false cognates and have nothing in common! Neither the OED nor the Merriam-Webster dictionary include “fugazi,” and so I need to do a little more research when I get home from the ISAAC 2010 conference. If I find any more definitive information, I’ll be sure do a post. But until then, my knowledge of the word is a little fugazi 😉

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s