Cutting edge technology

Introduction: On January 16th, Baidu's input method engaged in a press conference called “AI·New Input Full Sensory Input 2.0”, which released its declared “the first real AI input in China”. Law" Baidu input method AI exploration version.

With the continuous development of technology, both the virtual world and the real world are experiencing an unprecedented AI revolution.

On January 16th, Baidu's input method launched a conference called “AI·New Input Full Sensory Input 2.0”, which released its declared “the first real AI in China”. The input method "Baidu input method AI exploration version, Baidu senior vice president Wang Haifeng said, "Baidu input method will play an important role in the communication bridge, allowing people to interact into a new phase."

It can be said that Baidu input method is the current industry benchmark. This is also the first input method product that really thinks about human-computer interaction in the AI ​​era.

Borrowing AI Advantage Voice Interaction to Open "Baidu" New Experience

From the PC era to the mobile Internet era, the input method can be said to perfectly represent the upgrade and change of human-computer interaction, especially AI. The application of technology makes this cumbersome mechanical search tool have a more humane side. This is undoubtedly an upgrade of the experience after the technical entry, and the voice is the most natural since the birth of human beings. The interactive approach, but it has never been perfectly released in this upgrade war, is undoubtedly regrettable for those who believe in "technology changes the world."

Since Baidu’s high-profile AI strategy, Baidu’s products have undergone a series of intelligent changes, and the input method as a perfect bridge between content and human connection has also ushered in such an AI era. Respect, Baidu as the leader of the content industry, not only rely on its own strong content industry closed-loop, more importantly, in the new round of industrial upgrading, Baidu turned the past search engine into "search + information flow" Dual engine products. After the integration of AI, people in the past search engine era are looking for information and turning into information and services. It not only allows users to find information, but also allows information to find users and reconstruct a new information service experience.

According to Wang Haifeng, senior vice president of Baidu, the input method is one of the products closest to users, and it is also the bridge of AI landing."Fortress", Baidu's all-round AI capabilities are constantly empowering input methods, making the input method more "smart." He also gives the "future portrait" of the input method industry, although the current input method products are mainly used in mobile phones. However, the future input method will be applied to the interaction of various smart devices and users, and will go global, providing more input methods for global users.

Multi-style play method radical change new intrinsic input method logic

Since it is a Baidu input method AI exploration version, it will not only stay at the value of AI at the level of voice interaction, otherwise the word "exploration" will become meaningless.

[123 In order to better stimulate the user's enthusiasm for interaction, Baidu Input Method AI Discovery Edition's "New Exploration" launched the volley handwriting, using the finger to capture in the air within the scope of the camera, the input method can identify and generate text. It is a novelty game that uses mobile phone camera and AI technology. In fact, this means that the input method has the opportunity to leave the PC and mobile phone. For example, in the existing case, Baidu input method volley handwriting technology has been It has been applied on the small watch and the small genius phone watch.

Of course, this is just the side of “exploration” value, along with Baidu AI technology such as Baidu voice, vision, natural language processing, knowledge map and so on. Input method, the input of "all senses" becomes a reality. In addition to voice input, Baidu input method AI exploration version also introduces new functions such as shooting live, show, volley handwriting, etc. Whether it is a beloved idol or day and night companionship Pets, "shooting live" function can let users "drive" the other party to make the same expression through their own actions; "show field" function uses image segmentation technology, allowing users to "traverse" into different virtual scenes, technology, Fresh, cyberpunk, a variety of styles are free to choose, expression creation is more scene-like, more life; volley handwriting is like empty space, can also be smoothly recognized in the air writing, the current overall recognition rate has reached the requirements of large-scale application

From the manifestation of the Baidu input method AI exploration version, it can be said that the user's inherent knowledge of the input method is completely broken, except for the keystroke. In addition to the visual sense of the ideology of voice interaction, Baidu's AI Discovery Edition completely pulls users into another product space. The product form after technology plus code not only better attracts users, but more importantly, this kind of The new way of content interaction isIn the later period of Baidu, the Internet of Everything to meet the industrial upgrading has laid the foundation for convergence. While boosting the industrial upgrading, it has also brought more diverse scenes and new experiences to users.

Deeply digging the user's demand for the ultimate experience to help humanity return to the truth

However, as a user-driven product, the game's "everything" is undoubtedly an effective magic weapon to attract users to watch, but how to release users The strength of the post-development is the key to the product's endless life.

and Baidu Input Method AI Discovery Edition as a default input method for full voice input products, will undoubtedly make voice interaction an important highlight of the product, and for voice interaction, its primary problem is How to accurately complete semantic expression through the user's voice output? To this end, Baidu's input method specifically released the "streaming truncated multi-layered attention modeling" (SMLTA) technology for online speech. According to its statement, "the world's first implementation of local attention modeling recognition accuracy exceeds global attention. Model", and "the world's first large-scale deployment of online voice interaction attention model." This technology can solve the "delay" problem of the traditional model, Baidu claimed that it "improved the online speech recognition accuracy by 15%, exceeding the industry's highest level of 15%."

Moreover, in addition to the accuracy of speech recognition, the Baidu Input Method AI Discovery Edition is also optimized for dialects, enabling seamless switching through the “mixed speech input of dialects, dialects and mandarins”. , users can achieve "free to say." Similarly, Chinese and English inclusions can also be switched back and forth. So far, Baidu input method should be the only input method product that realizes high-precision Chinese-English mixed voice input and dialect-free switching voice input.

At the same time, in the field of offline voice recognition, in order to solve the demand for multiple input scenarios of users, Baidu continued to optimize the Deep Peak 2 system, which made the relative accuracy of offline voice input increase again, which is higher than the industry average of 35%. Even if the user is in an offline scene such as a subway, elevator, tunnel or crowded people, you can still use the Baidu input method for accurate voice input.

As of the end of 2018, Baidu's input method has reached 500 million active days, the peak number of voice requests exceeded 550 million, and the number of AR expressions used exceeded 100 million. At the same time, Baidu's overseas input method has more than 100 million installations worldwide.It supports 120 languages, covers 190 countries and regions, and has cooperated with more than 100 globally renowned IPs to achieve a double harvest of technology and market.

Throughout the history of the river, in the PC era, the input method was to let us interact with the machine through the mouse and keyboard; in the era of mobile Internet, the touch screen replaced the mouse and keyboard and became the new carrier of the input method. Imagine, after a series of interactive means such as keyboard, mouse, touch screen, what kind of industrial form will the dialogue artificial intelligence represented by intelligent voice produce?

Perhaps, the era of Baidu will be opened again!

This article was written by the author of the cutting-edge technology. The views represent only the author and do not represent the OFweek position. If you have any infringement or other problems, please contact us.

Hot topic