Voice interface, the liberation of human hands. Baidu's alarming progress.

China's leading Internet companies have a strong voice technology, leading the development of the Chinese mobile phone market.

Recently, MIT named the "Top Ten Breakthrough Technologies in 2016", including: Immune Engineering, Precision Editing Plant Gene, Voice Interface, Recyclable Rocket, Knowledge Sharing Robot, DNA Application Store, SolarCity Super Factory, Slack Collaboration Communication software, Tesla automatic driving system, air power.

This article describes the Voice Interface technology.

Breakthrough: Combines voice recognition and natural language understanding to create an effective voice interface for the world's largest Internet market.

Reason: Interacting with a computer through typing is time-consuming and frustrating.

The main researchers: Baidu, Google, Apple, Nuance, Facebook

Technology Time to Market: Current

If you walk around Beijing, we can see many people hold the latest mobile phones in their hands, such as Apple, Samsung and Xiaomi. And if you have closer observation, you will find that they are not using the touch screen, but the voice interface feature.

China is very suitable for the development of voice interface. Nowadays, China has 691 million smartphone users, and Baidu has become a common search engine user. Baidu’s voice development will allow people to communicate with the machine more conveniently since Chinese characters are not suitable for touch screen input. Baidu’s spokesman said the voice technology will be very developed, that people will be able to use it without thinking - the most powerful technology is often invisible.

For decades, the development of voice interfaces has always been the dream of technical experts. Not until recent years, the progress of machine learning has made great progress in voice control.

Voice is no longer limited to pre-programmed commands, and you can use voice even in the noisy streets of Beijing. Voice assistants, such as Apple's Siri, Microsoft's Cortana, Google Now are bundled with most of the smart phone, and recently Amazon’s Alexa can find information, songs, music, create a shopping list. These systems are not perfect; they sometimes misunderstand, misjudge and make a joke. But voice technology does give us a glimpse of a better future, because it can save us the trouble of understanding each interface.

Baidu has made great progress in speech technology, its voice recognition accuracy is very high. Baidu was founded in 2000, and it is China's national search engine. 70% of Internet users are using Baidu search, it is the equivalent of Google in the United States, and Google is blocked in China. Baidu is now expanding its business - music, video streaming, banking and insurance and other fields.

Improving the efficiency of mobile phone interface will greatly benefit China. Searching the web or sending messages in Chinese on smartphone has a fairly slow and frustrating pace. There are thousands of Chinese characters in China. Although a software called "Pinyin" was created, and it can translate Latin alphabet into Chinese characters, there are still many people, especially people over 50 years old, who do not know this software or how to use it. Chinese people use WeChat to deal with a lot of different thinks like work, paying bills. Besides, in China, many people from remote areas are illiterate, so the Internet has a big market there.

Andrew Ng, Baidu’s chief scientist and an associate professor at Stanford University says that voice will soon become a main communication tool. Robots and home appliances could be controlled by voice. The company has a research team at its Beijing headquarters and a lab in the Silicon Valley - they are committed to further improving the accuracy of speech recognition.

Jim, a senior research scientist at the Massachusetts Institute of Technology, has been researching speech technology over the past few decades. He believes that the timing of voice control has arrived.

Last November, Baidu announced that its Silicon Valley Laboratory has developed a powerful speech recognition engine "in-depth dialogue 2 generation." It consists of a very large neural network in its database - there are millions of voice dialogue, and it can hear the dialogue with the database for comparison. "In-depth dialogue 2 generation" of speech recognition accuracy is amazingly high, and it is sometimes even more accurate than human recognition.

Baidu's progress is very alarming. Chinese language is very complicated in pronunciation, because changing a tone will affect the expression of its meaning. An interesting fact about "Generation 2" is that, although the researchers do not speak Chinese, Cantonese or other Chinese dialects, they have developed a software that can recognize Chinese language. In addition, the voice engine is a global voice system, it can also learn English as long as users have enough voice dialogues.

The most frequent voice command among Baidu users is a simple query, such as tomorrow's weather or the pollution levels. Although the system is usually very accurate in this regard, the user's questions also become more complex. Last year, Baidu developed its own voice assistant, DuEr, which helps users find out when a movie was released and where to order it.

Baidu is now facing a greater challenge: how to create an artificial intelligence system to identify more complex voices. In order to achieve this goal, Baidu's Beijing headquarters decided to improve this system in order to identify user’s queries more accurately, which involves neural network technology. In addition, Baidu hired a team to analyze the orders received by DuEr and correct the errors in order to improve the accuracy of the system.

“In the future, I would love for us to be able to talk to all of our devices and have them understand us,” Ng says. “I hope to someday have grandchildren who are mystified at how, back in 2016, if you were to say ‘Hi’ to your microwave oven, it would rudely sit there and ignore you.”

Original article: here