Baidu Chief Scientist Wu Enda delivered a speech at the third anniversary of the Baidu Voice Open Platform. “I think in the era of artificial intelligence, Chinese people should be more confident.†Baidu chief scientist Wu Enda said at the third anniversary of Baidu’s voice open platform.
On November 22, the third anniversary of the launch of Baidu's voice open platform, Baidu announced the opening of four new voice technology interfaces to the public, namely emotional synthesis, far-field solution, wake-up phase II technology and long-voice solution.
As the head of Baidu's brain for Baidu's artificial intelligence products, Wu Enda said that a very important part of Baidu's brain project is to export the most advanced technology to developers and users.
Talking about the position of China's artificial intelligence in the world, Wu Enda believes that a lot of progress has been made by the Chinese. "Some people are a bit worried about the technology (competition) with the United States. The Chinese should be more confident. In the era of artificial intelligence, there are many (Artificial Intelligence) Technology In China, there are also many (artificial intelligence) technologies in the United States."
Wu Enda said that he could not evaluate the advantages and disadvantages of Chinese and American technology companies in artificial intelligence. "But the specific example I have seen is that the discovery of artificial intelligence technology is more in China."
Wu Enda reaffirmed the core technology components of Baidu's brain, namely speech technology, image processing, natural language, user portraits, and machine learning. He is full of confidence in the future of artificial intelligence. "I hope that in the future we will have (based on artificial intelligence) companion robots, personalized personal education, music composition, robot doctors, etc."
Baidu chief scientist Wu Enda explained the development track of Baidu voice technology recognition technology. It is reported that Baidu's mobile phone Baidu, Baidu map, Baidu input method and other products have already supported voice input. The "Secret" robot has been used in KFC's first concept store "Original+". Customers can not only talk to the secret, but also complete the whole process from order to payment by voice.
Wu Enda said that voice technology will have a good opportunity in many application scenarios in the future, which will bring great changes to the field of human-computer interaction. Wu Enda believes that people can use speech and TV, remote control, speakers, curtains and other "speech" in the future. In addition, Wu Enda believes that the application of voice technology in intelligent customer service centers, automobiles and other fields is also an important human-computer interaction scenario.
Baidu announced on November 23 that since the launch of the Baidu voice open platform in October 2013, the daily online voice recognition requirements have risen from 5 million in 2013 to today's 140 million, and online voice synthesis daily requests have been reached. 200 million, the number of developers exceeds 140,000. In terms of technical indicators, Baidu's speech recognition accuracy rate has now reached 97%.
In February 2016, Baidu's deep speech recognition system Deep Speech2 was selected as the MIT “2016 Top Ten Breakthrough Technologyâ€.
Wu Enda also gave an example of awakening the second-phase speech technology application to the journalist. This technology can be applied in the self-timer scene. By setting a custom wake-up word, people can easily complete the photo by simply calling “1, 2, 3â€. The wake-up rate is 95%. The application scenarios of the voice wake-up function are very extensive. Wu Enda believes that developers have many opportunities to use Baidu voice wake-up technology to achieve diversified product functions.
Wu Enda told 澎湃News that Baidu voice technology, in addition to its application to Baidu’s maps and other products, also hopes to support third-party companies. “We put voice technology into our own products and got very good results in the products, let users When using our products, you can use voice input or voice synthesis... We hope to support third-party companies and put these technologies into their products, so that they can get the same effective results."
The following is a speech by Wu Enda on the third anniversary of the launch of Baidu's voice open platform:
Hello everyone, I am very happy that everyone is here today to participate in the third anniversary of our voice open platform.
I am Wu Enda, the project leader of Baidu Brain. I have a very important job in the Baidu brain project, which is to export the most advanced technology to everyone. On Baidu's brain open platform (ai.baidu.com), we not only output artificial intelligence technology, but also many training materials about artificial intelligence technology.
So, what exactly is Baidu’s brain? Its core technical parts are: voice, image, natural language processing, user portrait, machine learning platform. I have found that many people use Baidu's voice technology for two reasons. The first reason is that Baidu has a very advanced technology to provide a better experience for their users. The second reason, I personally done for decades on labor. Smart projects, found that when you first started a project, you only need voice technology. After a few months, you find that you need image technology or natural language processing technology. If you are using Baidu's brain open platform, you can easily select the technology part you really need to put it in and integrate it to achieve the best results for you. For example, if you want to serve a smart customer, you only need voice recognition and speech synthesis technology at first, but after a few months, you find that you need the most advanced natural language processing technology. If you are using the Baidu brain open platform, I hope that you can easily get these technologies into your own products.
I personally have confidence in the future of artificial intelligence. I hope that in the future we will have accompanying robots, personalized personal education, music composition, robot doctors and so on. These years, artificial intelligence technology has developed very fast, which gives us a lot of new opportunities. There is no way to explore so many very potential opportunities in Baidu, so our goal is to export our artificial intelligence technology. For everyone, I hope you can support your exploration on very promising projects and support you to get very good results in these projects.
The theme of today's conference is voice technology. As you may already know, in many of the most important Baidu products, we already support voice input, including mobile phone Baidu, Baidu map, Baidu input method. If you haven't tried Baidu's input method yet, I hope you give it a try. I like to use Baidu input method when I input information. There is also a secret, recently we put the secret into all kinds of hardware, such as small robots. The secret system can order food for customers, integrate secrets and robots, and give new users a new experience.
Over the past few years, our team has been continually optimizing the speech recognition system. We started using the DNN model in 2012. Later, we have better features. After that, we started using the Sequence DiscriminaTIve Training and started using the LSTM model. Plus CTC, this year our The team developed the DeepCNN model and the results are constantly improving. This is our speech recognition system.
In addition, our speech synthesis model has become better and better. In recent years, we have made great breakthroughs in several technologies, and our speech synthesis effects have become better and better. Now Baidu's ability to synthesize speech in China is far more than that of other companies. Our voice team leader will share more details with you this afternoon.
Because our voice technology, natural language processing technology, and our human-computer interaction technology have progressed so fast, in February 2016, MIT Technology Review listed our technology as one of the “Top Ten Breakthrough Technologies in 2016â€. I feel that these technologies have great potential to change the efficiency and methods of human-computer interaction.
Future voice technology has a good opportunity in many application scenarios, bringing huge changes to human-computer interaction. For example, in the smart home scene, we hope that you can use your voice to talk to your TV, remote control, speakers, and curtains in the future. For example, I hope you can say to your light in the future, "Please turn on the lights." It understands your commands, or speaks to the remote control, it can easily meet your needs. In fact, smart home is just an application scenario. I think that in many application scenarios, voice technology can bring about very big changes in human-computer interaction, including mobile phones, speakers, TVs, and so on. I hope that voice can make you a better intelligent customer service center. In the automotive field, voice is also a very important human-computer interaction scenario, and many of our partners are using our voice technology to explore these scenarios.
Finally, today we know that, especially on the mobile side, we know that voice input is faster, more convenient, and more accurate than keyboard typing. The next speaker is my good friend and Professor James Landay of Stanford University. He will also share research on this aspect, using Baidu's speech recognition system. If you use this system, you can enter it on the mobile phone. Faster, more convenient, and more natural.
Voice input is so convenient, I hope that voice input will become more popular in the future. Think about it, if you want to communicate with me, the most natural way is to tell me a word, or I can tell you a word, so the most natural way to communicate with others is voice. In the past few years, the technological progress of speech and natural language processing and dexterity has been so fast. In many scenes today, the most natural way of human-computer interaction is voice. Human-computer interaction is not just about talking to your computer or to your device. Your computer and device also need to tell you what it is, so our voice technology can also let these devices tell you what it is.
Finally, because artificial intelligence, especially voice technology, has grown rapidly in recent years. There are several different artificial intelligence technologies in Baidu's brain, among which the more mature one is our voice technology. So voice technology has been given to us, to everyone, to you, to your company, to developers a lot of new opportunities. Therefore, I hope that we can output voice technology, I hope it can provide very good service for everyone, and hope that we can use our technology to support you to explore these new opportunities. I hope that you can use these technologies to support you in these new opportunities. Very good results.
thank you all!
Must-have Feature:
Glass-lid
Drainage Spouts
Locking Lid.
Non-Stick Coating.
Adjustable Steam Vent
Heat-Resistant Handles.
Electric Skillet
Shaoxing Haoda Electrical Appliance Co.,Ltd , https://www.zjhaoda.com