“Conquering Babel”

“Simultaneous translation by computer is getting closer”
From The Economist, Jan 5th, 2013, Seattle – from the print edition

IN “STAR TREK”, a television series of the 1960s, no matter how far across the universe the Starship Enterprise travelled, any aliens it encountered would converse in fluent Californian English. It was explained that Captain Kirk and his crew wore tiny, computerised Universal Translators that could scan alien brainwaves and simultaneously convert their concepts into appropriate English words.

Science fiction, of course. But the best sci-fi has a habit of presaging fact. Many believe the flip-open communicators also seen in that first “Star Trek” series inspired the design of clamshell mobile phones. And, on a more sinister note, several armies and military-equipment firms are working on high-energy laser weapons that bear a striking resemblance to phasers. How long, then, before automatic simultaneous translation becomes the norm, and all those tedious language lessons at school are declared redundant?

Not, perhaps, as long as language teachers, interpreters and others who make their living from mutual incomprehension might like. A series of announcements over the past few months from sources as varied as mighty Microsoft and string-and-sealing-wax private inventors suggest that workable, if not yet perfect, simultaneous-translation devices are now close at hand.

Over the summer, Will Powell, an inventor in London, demonstrated a system that translates both sides of a conversation between English and Spanish speakers—if they are patient, and speak slowly. Each interlocutor wears a hands-free headset linked to a mobile phone, and sports special goggles that display the translated text like subtitles in a foreign film.

In November, NTT DoCoMo, the largest mobile-phone operator in Japan, introduced a service that translates phone calls between Japanese and English, Chinese or Korean. Each party speaks consecutively, with the firm’s computers eavesdropping and translating his words in a matter of seconds. The result is then spoken in a man’s or woman’s voice, as appropriate.

Microsoft’s contribution is perhaps the most beguiling. When Rick Rashid, the firm’s chief research officer, spoke in English at a conference in Tianjin in October, his peroration was translated live into Mandarin, appearing first as subtitles on overhead video screens, and then as a computer-generated voice. Remarkably, the Chinese version of Mr Rashid’s speech shared the characteristic tones and inflections of his own voice.


Though the three systems are quite different, each faces the same problems. The first challenge is to recognise and digitise speech. In the past, speech-recognition software has parsed what is being said into its constituent sounds, known as phonemes. There are around 25 of these in Mandarin, 40 in English and over 100 in some African languages. Statistical speech models and a probabilistic technique called Gaussian mixture modelling are then used to identify each phoneme, before reconstructing the original word. This is the technology most commonly found in the irritating voice-mail jails of companies’ telephone-answering systems. It works acceptably with a restricted vocabulary, but try anything more free-range and it mistakes at least one word in four.

The translator Mr Rashid demonstrated employs several improvements. For a start, it aims to identify not single phonemes but sequential triplets of them, known as senones. English has more than 9,000 of these. If they can be recognised, though, working out which words they are part of is far easier than would be the case starting with phonemes alone.

Microsoft’s senone identifier relies on deep neural networks, a mathematical technique inspired by the human brain. Such artificial networks are pieces of software composed of virtual neurons. Each neuron weighs the strengths of incoming signals from its neighbours and send outputs based on those to other neighbours, which then do the same thing. Such a network can be trained to match an input to an output by varying the strengths of the links between its component neurons.

One thing known for sure about real brains is that their neurons are arranged in layers. A deep neural network copies this arrangement. Microsoft’s has nine layers. The bottom one learns features of the processed sound waves of speech. The next layer learns combinations of those features, and so on up the stack, with more sophisticated correlations gradually emerging. The top layer makes a guess about which senone it thinks the system has heard. By using recorded libraries of speech with each senone tagged, the correct result can be fed back into the network, in order to improve its performance.

Microsoft’s researchers claim that their deep-neural-network translator makes at least a third fewer errors than traditional systems and in some cases mistakes as few as one word in eight. Google has also started using deep neural networks for speech recognition (although not yet translation) on its Android smartphones, and claims they have reduced errors by over 20%. Nuance, another provider of speech-recognition services, reports similar improvements. Deep neural networks can be computationally demanding, so most speech-recognition and translation software (including that from Microsoft, Google and Nuance) runs in the cloud, on powerful online servers accessible in turn by smartphones or home computers. (…)

Read the entire article here

Windows 8 Physician Rounds app video

Learn how healthcare apps could look in the near future with this demonstration app called Rounds, allowing doctors and nurses to use tablets to connect with each other and access medical records from wherever they’re working.

Rounds application is an example of how Windows 8 can combine the functionality of a full PC within the interface of a tablet. It could help multidisciplinary teams improve the quality, safety, speed and outcomes of care. Rounds incorporates the touch-screen capabilities of Windows 8 tablets such as Microsoft Surface to allow doctors to locate patients and initiate instant-messaging sessions with nurses.

About Microsoft in Health
Microsoft is committed to improving health around the world through software innovation. For over 16 years, Microsoft has been providing a broad portfolio of technologies and collaborating with partners worldwide to deliver solutions that address the challenges of healthcare providers, public health and social services, payers, life sciences organizations, and consumers. Today, Microsoft invests in technology innovation and works with health organizations, communities and over 20,000 partners around the world to make a real impact on the quality of healthcare.

About Microsoft
Founded in 1975, Microsoft (Nasdaq “MSFT”) is the worldwide leader in software, services and solutions that help people and businesses realize their full potential.


Translation Tools Could Save Less-Used Languages

Tom Simonite - Wednesday, June 6, 2012, Technology Review (published by MIT)

Languages that aren’t used online risk being left behind. New translation technology from Google and Microsoft could help them catch up.

Sometimes you may feel like there’s nothing worth reading on the Web, but at least there’s plenty of material you can read and understand. Millions of people around the world, in contrast, speak languages that are still barely represented online, despite widespread Internet access and improving translation technology.

Web giants Microsoft and Google are trying to change that with new translation technology aimed at languages that are being left behind—or perhaps even being actively killed off—by the Web. Although both companies have worked on translation technology for years, they have, until now, focused on such major languages of international trade as English, Spanish, and Chinese.

Microsoft and Google’s existing translation tools, which are free, are a triumph of big data. Instead of learning as a human translator would, by studying the rules of different languages, a translation tool’s algorithms learn how to translate one language into another by statistically comparing thousands or millions of online documents that have been translated by humans.

The two companies have both departed from that formula slightly to serve less popular languages. Google was able to recently launch experimental “alpha” support for a collection of five Indian languages (Bengali, Gujarati, Kannada, Tamil, and Telugu) by giving its software some direct lessons in grammar, while Microsoft has released a service that allows a community to build a translation system for its own language by supplying its own source material.

Google first realized it needed to give its system a grammar lesson when trying to polish its Japanese translations, says Ashish Venugopal, a research scientist working on Google’s translation software. “We were producing sentences with the verb in the middle, but in Japanese, it needs to go at the end,” Venugopal says. The problem stemmed from the system being largely blind to grammar. The fix that the Google team came up with—adding some understanding of grammar—enabled the launch of the five Indic languages, all used by millions on the subcontinent but largely missing from the Web.

Google’s system was trained in grammar by giving it a large collection of sentences in which the grammatical parts had been labeled—more instruction than Google’s translation algorithms typically receive.

Venugopal says that, so far, the system can’t handle the underserved languages as well as Google’s existing translation technology can handle more established languages, such as French and German. But, he says, offering any support at all is important for languages that are relatively rare online. “It’s an important part of our mission to make those other languages available on the Web,” he says. “We don’t want people to have to decide whether to publish their blog in their own language or in English. We want to help the world read your blog.”

Microsoft is also interested in helping languages not in common use online, to prevent those languages from being sidelined and falling from use, says Kristin Tolle, a director at Microsoft Research. Her team recently launched a website that helps anyone to create their own translation software, called Translation Hub. It is intended for communities that wish to ensure their language is used online.

Using Translation Hub involves creating an account and then uploading source materials in the two languages to be translated between. Microsoft’s machine-learning algorithms use that material and can then attempt to translate any text written in the new language. Microsoft piloted that technology in collaboration with leaders of Fresno, California’s large Hmong community, for whose language a machine translation system does not exist.

“Allowing anyone to create their own translation model can help communities save their languages,” says Kristin Tolle, a director at Microsoft Research. Machine translation systems have been developed for roughly 100 of the world’s 7,000 languages, says Tolle.

“There is a lot of truth to what Microsoft is saying,” says Greg Anderson, director of nonprofit Living Tongues, which documents, researches, and tries to support disappearing languages. “Today’s playing field involves a digital online presence whether you are community or a company—if you don’t have a Web presence, you don’t exist, on some level.” Anderson says that sidelined languages making a comeback are usually those from communities that have embraced online life using their language.

Margaret Noori, a lecturer at University of Michigan who works to preserve the Anishinaabemowin or Ojibwe, a native American language, agrees, but adds that preserving a language involves more than the Web. “There is a reason to be online in today’s world, but it absolutely must be balanced by songs sung only aloud and ceremonies never recorded.”

Microsoft’s Translation Hub is also aimed at enabling the translation of specialist technical terms or jargon, which general purpose online translation tools do not handle well. Nonprofits could, for example, use it to translate materials on agricultural techniques, says Tolle, and the technology can also be useful to companies that wish to speed up translation of instruction manuals or other material.

“Companies often want to have their data available to them privately and retain their data—not to provide it to someone else that will train a translation system,” she says. Volvo and Mercedes have expressed an interest in testing Microsoft’s Translation Hub, says Tolle.

Tom Simonite - Wednesday, June 6, 2012,
Source:  Technology Review (published by MIT)