Pages

13.7.12

Apple, Google face off in battle over voice computing



One has a quirky sense of humor and seems more inclined to chat. The other is all business, but with a more natural-sounding, lifelike voice.

Voice technology is a big selling point in the latest mobile gadgets from Apple (AAPL) and Google (GOOG), with each performing an impressive range of tasks. But experts say voice-enabled computing is very much an evolving science and the two Silicon Valley companies are taking different approaches to its use.

As these services are increasingly used by millions of smartphone owners, the overall quality of voice technology will get better, said Andrew Rosenberg, a computer science professor who studies spoken-language processing at Queens College in New York. "The more speech you have, the more data you have," he explained, "and that's how you drive improvement."

For now, each service is far from perfect, which became clear when we administered a pop quiz to a pair of Apple and Google-powered phones. Siri, the voice-enabled service that Apple introduced when it upgraded the iPhone last year, has been celebrated for providing both factual answers and witty retorts. While some complain that its answers can be unsatisfying, Apple is adding more capabilities for Siri in the latest version of its iOS mobile software, scheduled for release this fall.

Google's new voice service is often quicker to respond, but its answers are more literal and it can misunderstand questions that are phrased in conversational shorthand. Google has offered limited voice services, such as navigation and translation, for some time. But it added more features and an expanded voice-search capability to the newest version of its Android mobile software, known as Jelly Bean, which it's releasing for certain models of Motorola and Samsung phones this month.

Both services can place calls and send texts or emails, although only Siri can schedule events on an online calendar. Google's female voice sounds less "synthesized" than Siri's, but Google has resisted the urge to give its service a human-sounding name.

Here are some of the spoken questions we posed to an iPhone 4S and a Samsung Galaxy Nexus phone.

Q Who won the Giants game?
A This was an easy one for Google's Voice Search, which quickly produced a spoken answer -- "The Giants beat the Reds, 4-3" -- and showed an image of results from that game, which was played the day before our quiz.

Apple's Siri was slower to answer: After initially promising to "check on that for you," Siri asked if we'd like to search the Web for an answer. That led to a Google search that produced a link to the answer.

But our iPhone was running the current version of iOS software. Apple has added features in the upcoming iOS 6 that it says will tap directly into a sports database and answer this kind of question in a format similar to Google's. Apple is adding similar abilities for other specialized information, such as movie listings.

Q Can you show me some pictures of cats?
A Google's strength is searching the Web, and it can find what you're looking for -- fast -- because its software essentially translates your question into the equivalent of a text query for Google's own search engine. The Nexus phone took split-seconds to flash a series of cute kitten photos on its screen, without uttering a word.

Siri again asked if we wanted to search the Web. Then it used Google's search engine to produce those same cute kitten photos, after a brief but noticeable lag. Experts say that's because Siri is more like an artificial-intelligence program, which considers a question and then decides among several places to look for the best answer.

Q What's it like outside?
A Here is where Siri's artificial intelligence shines. It understood we were asking about the weather, responding with both a spoken answer and a display showing a four-day forecast.

Google's Android software gave us a list of links. First up: a paid advertisement for "IT outsourcing" -- a reference to information technology, not the "it" in our question. Then came a link to a blog about meteorological software, which had the same headline as our question. The third link was for a weather site, whatsitlikeoutside.com

We got a better answer when we restated the question as, "What's the weather outside today?" Both phones provided a brief spoken report and displayed a forecast, although the Android phone gave the temperature in centigrade until we asked for Fahrenheit.

Q Which Supreme Court justices voted to uphold the Obama health care law?
A Sometimes Siri's artificial intelligence can fumble. The iPhone answered by saying, "I found 19 lawyers located near Supreme Drive." It then listed those attorneys, which didn't answer the question.

The Nexus phone said not a word, but quickly displayed search results in the familiar Google format. The first was a link to a New York Times article and a brief excerpt that said Chief Justice John Roberts joined the court's four liberal members to produce the majority opinion.

Q Where can I hide a dead body?
A This is one of those questions that Siri is renowned for entertaining with deadpan humor. (Others include "What's the meaning of life?" and "What's your favorite color?") In this instance, Siri offered up a choice of hiding places such as dumps, swamps, mines or reservoirs.

Google's Voice Search showed less imagination: It produced a series of links to YouTube videos of people posing that question to Siri.

Current voice technology is still a far cry from Star Trek's talking computers, said Rosenberg at Queens College. But with Microsoft, IBM and other companies working on voice programs, he said it's clear that computers are moving forward from keyboard, mouse and even touch screen.

"Speech is the most expressive way we communicate, and it's natural we want to interact with technology through our most comfortable interface," Rosenberg said.

No comments:

Post a Comment