日本公司推移动语音翻译程序

日本最大的移动运营商NTT Docomo宣布将于11月1日正式推出一款实时移动语音翻译应用程序，可在用户通话时实现日语与英语、中文、韩语之间的翻译转换，之后还将添加法语、德语、意大利语等其他七门语言。该公司在声明中表示，任何智能手机都能通过该应用程序获得快速准确的翻译，因为该程序是在其云处理系统基础上运行的。不过，用户需订购该公司的套餐才能使用此翻译程序。与此同时，阿尔卡特-朗讯以及微软等公司也都在研制类似的语音转换应用程序，不但能实现一对一的语言转换，还能实现十人以内、四种语言的会议实时翻译。虽然发展前景广阔，不过有专家也对这种语言转换程序的可靠度表示担忧，称语音识别和机器翻译终究无法取代语言技能，尤其在商务翻译中，机器软件或许并不是那么可靠。

An app offering real-time translations is to allow people in Japan to speak to foreigners over the phone with both parties using their native tongue.

NTT Docomo - the country's biggest mobile network - will initially convert Japanese to English, Mandarin and Korean, with other languages to follow.

It is the latest in a series of telephone conversation translators to launch in recent months.

The products have the potential to let companies avoid having to use specially trained multilingual staff, helping them cut costs. They could also aid tourism.

However, the software involved cannot offer perfect translations, limiting its use in some situations.

NTT Docomo unveiled its Hanashite Hon'yaku app for Android devices at the Combined Exhibition of Advanced Technologies (Ceatec) show in Japan earlier this month, and plans to launch it on 1 November.

It provides users with voice translations of the other speaker's conversation after a slight pause, as well as providing a text readout.

"French, German, Indonesian, Italian, Portuguese, Spanish and Thai will be added for this application in late November, raising the number of non-Japanese languages to 10," the firm said in a statement.

"Fast and accurate translations are possible with any smartphone, regardless of device specifications, because Hanashite Hon'yaku utilizes Docomo's cloud [remote computer servers] for processing."

The caller must subscribe to one of Docomo's packages to be able to use it.

NTT Docomo will soon face competition from France's Alcatel-Lucent which is developing a rival product, WeTalk. It can handle Japanese and about a dozen other languages including English, French and Arabic.

Alcatel-Lucent uses a patented technology to capture the user's voice and enhance it before applying speech recognition software. The data is then run through translation software before being run through a speech synthesizer.

The firm said all this could be done in less than a second. However, it has opted to wait before the speaker has stopped talking before starting the translation after experiments carried out with workers at insurance company Axa suggested users preferred the experience.

"We are still working on improving the system," Gilles Gerlinger, the product's co-founder, told the BBC.

"You can do conversations with one person, but we want to allow conferences with 10 people and four different languages, and the system would provide translations in every language needed.

"We also have a project called Myvoice which can have a synthetic voice that sounds like your real one."

Mr Gerlinger suggested that his firm would make money from the product by renting servers with the necessary software to big businesses, and charging smaller ones a fee for the amount of time they used the service.

Despite the ambitions of those involved in the nascent sector, one analyst questioned their chances of success

"These kind of real-time technologies have been 'two to three years away' for the past decade," said Benedict Evans, technology expert at Enders Analysis.

"Both speech recognition and machine translation are sort of there if you're not too fussy.

"But they are generally not as good as speaking the language itself, and my suspicion is that they would not reliable enough to use them for business purposes when you need to be really sure about what the other person said."