Humour

Talk to your computer

Speech recognition technologies are coming of age. By T A Balasubramanian

We are back in the domain of Doodh Byramji, known to his close friends as plain Doodh. As we know, Byramji is the hyperactive IT research engineer from Baffle Technologies, called Baff-Tech for short. He is encouraged to probe into the odds and ends of the computer world by his CEO, Baidyanath Baffle, the founder and owner of Baff-Tech.

Presently, Doodh is seated at a table in Quick Sip, the trendy restaurant frequented by geeks and jaunty business honchos carrying laptops and wearing headphones. In conversation with Doodh is the sagacious-looking Groucho Goose, Manager, Slinky Marketing Strategy for Confusing Clients, from Duckbill & Goose.

Dear Diary (writes Byramji, putting down his astute observations as a spicy narrative that he plans to make into a book later):

Groucho has been telling me recently about what he calls ‘speaking naturally,’ and that’s made me quite curious.

“Well Groucho, you seem quite excited about something that you wanted to meet me about.”

“Ah, Doodh, indeed I am excited,” says Groucho, ordering espresso for himself and special tea for me. “Most people speak over 120 words per minute but type less than 40 words per minute. What if you could create email, documents and spreadsheets simply by speaking to your computer—or your mobile device? What if you could control your gadget merely by talking to it, starting programs, using menus, surfing the Web?”

“You are, of course, referring to speech recognition?”

“Of course I am. Do you recollect an earlier chat we had a few years ago, focused on the limits of miniaturization?”

“Hmm, yes I do. That’s when I said that I was certain that, after a certain point, computers will never become any smaller, because we still need a screen and some way to input information?”

“That’s right, Doodh. And I suggested that these are easily solvable problems.”

“Ah, I recall it. We will no longer need screens on our computers, you said, because we could all be sporting goggles that project an enormous virtual monitor before our eyes.”

“You have a good memory, Doodh. I said we could do away with keyboards too, because we would have perfect speech recognition software. We could all just dictate text into our computers, palmtops and watches.”

“Well, Groucho, I still have my doubts about the goggle screens. Sure, these personal monitor glasses exist, but you don’t exactly see people wearing them in the office. I have tried these things out at trade shows, and found them to be pretty annoying compared with the standard big flat-panel screen. I am not saying that they will never become common appendages, like your iPods today. I am just saying that it is not a sure fire thing.”

“You may be right. But I will, however, bravely stick my neck out to say this: speech recognition will soon replace the keyboard.”

“And why are you certain now?”

“The software gets smarter with each new version. Thanks to a nasty hand problem, I do most of my writing using Pen Dragon, which I am addicted to, and I have been using this pet steno since version 3. I have been correcting each transcription error along the way, and my Pen Dragon has dutifully mastered every quaver of my voice. After all these years, with version 10, I get 99% accuracy.”

“I think that is amazing, but I beg to differ, Groucho. Dictation software will never reach 100% accuracy. We will always need a keyboard to correct typos, but not because the software is not good enough.”

“Oh, but then what is the glitch?’

“Ah, in my opinion, it is because in the English language, too many words—and phrases—sound alike. Let me give you a few instances that my own little experiments with Pen Dragon have exposed. ‘A case we summarily dismissed’ became ‘a case we so merrily dismissed’. ‘I might add’ became ‘I, my dad’. ‘A nameless feeling’ turned into ‘an aimless feeling.’ You begin to get the impression that you are interacting with a somewhat retarded steno, though, of course, there is an improvement over many years.”

“That’s true. You must admit, however, that when spoken at normal conversational speed, ‘an aimless’ and ‘a nameless’ sound identical even to us humans, eh? No handbook of body language or context decoding would ever reveal the correct interpretation. But I still say that speech recognition, like the wheels of justice, grinds on slowly, but it grinds exceedingly fine. It will eventually become perfect, as Pen Dragon becomes sophisticated enough to ‘understand’ the context and inflection of my speaking—and I would excuse the rare slip that would even have you and me fooled.”

“Ah, but with all the grinding, we do need the humble keyboard to make those corrections.”

“Yes, I suppose so,” says Groucho with a sigh. “But mark my words, smart speech technology is going to show up at every place where speech happens. Of course, this doesn’t mean that people will always choose to speak. For example, I am unlikely to want to use my voice to handle my finances, at least in public spaces.”

“It would work well in the living room. I can see the day when I yell ‘News!’ at the television set when I walk in, and my favorite news channel will appear on the screen.”

“I know some people who yell at their PCs already,” says Groucho. “Especially when they break down without warning.”