Submitted by SHNS
Tue, 01/13/2009 - 14:54
* By L.A. JOHNSON, Pittsburgh Post-Gazette
Mia Liu, a Chinese MBA student at the University of British Columbia, wants to speak clear, precise English for business and pleasure.
That's why she's using Carnegie Speech's NativeAccent software to improve her spoken English.
Carnegie Speech, a Carnegie Mellon University spin-off company, in 2004 started marketing its NativeAccent software and Carnegie Speech Assessment software, which pinpoints errors in sound, rhythm and pitch.
"The software can record my own voice, my own pronunciations, and tell me which vowels I couldn't pronounce right," said Ms. Liu, 33, via telephone on break at home in Chongqing, China. "I can compare it with the native (English) speaker's pronunciation and check the differences."
Liu, whose native language is Mandarin, can access the software program from any computer in the world -- at school in Vancouver, B.C. or in China -- practice her spoken English exercises in private, and receive feedback and additional exercises specifically tailored to her problem areas.
"What we're really looking to do is help people become intelligible in their speech, not really eradicate all traces of their native accent," said Angela Kennedy, Carnegie Speech's chief executive officer.
With the Internet and the global economy, the need for people to be understood across the world has spun into a $6 billion-a-year market for these types of products in a variety of sectors, including business outsourcing, U.S. adult education, U.S. government (in the areas of commerce, intelligence, the State Department and armed services) and aviation.
English is the international language of commerce, Kennedy said, and multinational businesses -- especially those that outsource -- use the software in multiple countries around the world to train their globally diversified work forces.
People speaking more than 40 foreign languages -- including Arabic, Bengali, Farsi, French, Greek, Italian, Japanese, Kazak, Khmer, Mandarin, Montenegrin, Punjabi, Russian, Slovak, Spanish, Tagalog, Thai and Turkish -- use Carnegie Speech products, including NativeAccent software, to improve their spoken English.
Strong client bases are in India, the Philippines, Latin America and Eastern Europe, with burgeoning markets in China and Japan.
People can learn a language at any age, but if a new language has a different writing system or more sounds than one's native language, learning the new language can be more difficult, said Maxine Eskenazi, inventor of the NativeAccent pinpointing technology and Carnegie Speech's chief technical officer.
"When Japanese people learn English, they don't have discrimination between 'r' and 'l'; to them that's just one sound," said Eskenazi, an expert in the area of speech recognition technology.
However, when the acoustics between the two sounds are altered enough, people can hear the difference.
"Show them the error and have them listen to the difference," she said. "That's something that is very, very unique to this software."
Often people learning English are taught by non-native English speakers. Consequently, they learn mostly through written exercises and not actually speaking the language.
"The fact that (NativeAccent users) can practice to their heart's content, in front of a computer, without judgment and without losing face, that's really important," Eskenazi said. "That's the great thing about it being Web-based."
With a headset microphone that can be plugged into a computer, users are able to access the NativeAccent software on servers around the world.
"You're able to speak the curriculum, and then it goes over the Internet to the servers that we have set up to host the software and does the analysis," Kennedy said.
Then, the software sends back to a student's computer a highlighted version of the word or sound -- in red/yellow/green -- visually pinpointing which segments of the word or sound the student got wrong, close or right. For example, if a student mispronounced the "t" in cantaloupe, the "t" would be highlighted in red while the rest of the word would be in green, indicating the rest of the pronunciation was correct, Kennedy explained.
Any business or government agency that needs employees to speak intelligible English can use Carnegie Speech products. U.S. businesses that outsource operations to foreign countries, for example some U.S. companies that operate call centers in India for U.S. customers, use NativeAccent.
"In the U.S. government sector, we are training spoken English and training foreign language," Kennedy said. "We've taken the same technology and are teaching Americans to speak Iraqi Arabic, and we're about to produce a spoken Russian project for the intelligence agencies, as well."
This month, Carnegie Speech, together with partner Mayflower College in Plymouth, England, is launching a new productgeared toward international pilots and air traffic controllers, who have to meet new spoken English requirements by March 2011.
Adult education also is a huge market for NativeAccent. Its individualized curriculum offers the most effective training in a minimal amount of training time because it focuses on users' problem areas.
"It's like having a one-on-one tutor who is customizing a curriculum for you at every point in your training," Kennedy said.
With as little as 10 hours of training, clients are improving their speaking skills an average of 30 percent, and after 20 to 40 hours of training, users are improving 50 to 100 percent, she said.