In the Future, the Spring Festival Gala May Not Need a Live Host.

By Super Neuro

At the 2019 Online Spring Festival Gala on January 28, an AI virtual host took the stage to host the show. Sa Beining, as one of the hosts on the same stage with the AI stand-in, announced on the spot that he was ready to "retire". Has the time really come for AI to replace hosts?

This year's online Spring Festival Gala saw the debut of four virtual hosts, based on Sa Beining, Zhu Xun, Gao Bo and Long Yang. The four hosts hosted the show with their virtual hosts, which really added a lot of fun to the gala.

Jump to watch: Four virtual hosts

As can be seen from the video, there is a great deal of similarity between the virtual AI host and the real one. In addition to being as good as humans in answering questions and making witty comments, they also have corresponding body language, and each virtual host also has different expression characteristics.

What is a Virtual Host?

It is reported that the virtual host who appeared on the Internet Spring Festival Gala this time was PAI (Personal AI) created by ObEN, which can create a virtual mirror model based on a person's image.

The core technologies of this product are mainly in three aspects:

First, the visual image is established. Using the host's photo, 3D scanner, 3DS MAX, MAYA and other 3D modeling tools, AI builds an algorithm model, which understands the relationship between the color distribution and structural depth of the photo, and finally constructs the face and body shape.

Secondly, it is the synthesis of sound. The AI voice technology they use does not require the collection of a large number of voice libraries. It only needs a dozen voice recordings to achieve the establishment of a sound model through feature parameter extraction, transfer learning and other methods.

Finally, to make the virtual host as realistic as possible, it is necessary not only to match the visuals with their voices, but also to give them personalization. Through sensors and motion tracking devices, combined with AI and motion capture training, the virtual host can simulate oral expressions, facial expressions, gestures, body movements, and scene interactions according to the corresponding body.

Although this is the first time that AI has crossed over into hosting, there have been quite a few appearances of AI virtual anchors in the world.

The first AI news anchor

At the Fifth Internet Conference held in November 2018, Xinhua News Agency showcased a new reporter, the first AI virtual news anchor. Qiu Hao, the prototype of the AI, said: "The image is my image, and the sound is also suitable for my voice, but I have never said these words in the broadcast..."

Jump to watch, AI virtual anchor

In this video, we can see the anchor in the video introducing himself in his accent, and his face and lips move accordingly to match the voice. I have to say, there are surprises, but it is still a bit disappointing. Compared with real anchors, people can tell at a glance that it is a mechanized product. Its working method still requires humans to write press releases. But

This kind of technical support comes from Sogou's "avatar technology". The key behind this technology is speech synthesis and image generation.

Speech synthesis technology can use a small amount of audio data to allow the machine model to learn the speaking characteristics of the input object, grasp the timbre, rhythm, emotion and other aspects, and ultimately realize the audio information of the input text.

Image generation uses learning and construction in aspects such as face recognition, 3D face reconstruction, and expression modeling, ultimately achieving the correspondence between output audio and output visual information.

Putting aside those criticisms, the biggest promotional point of this technology is that it can realize the on-screen clone function, which means that you can use TV, tablet and mobile phone to see one person broadcasting three different contents at the same time.

Earlier AI anchor

There were reports in Japan earlier that they had also made an AI anchor.

In April last year, the host of a program called "NEWS CHECK 11" on Japan's NHK TV station was a cute cartoon AI anchor - "News Anchor Yomiko".

  This anchor was created using CG technology. It learned to split into phonemes from a large number of recordings, then learned to recognize and read text, and finally used it to read the news.

In addition, robots have also been used as news anchors.

The beautiful robot "Elika" developed by Osaka University and Kyoto University also served as a news anchor for Japan's NNN TV station in April 2018.

The character of "Elika" is a 23-year-old beautiful girl. She is given a standard female face and her speaking voice is synthesized based on the voice actor's recording, which can be described as very natural.

In addition, it also has an advanced dialogue system. When talking to people, it collects information through microphones and sensors, perceives the other party's voice and movements, and then turns to the other party and has a fluent conversation.

Its eyes, mouth, neck and 19 other parts can move through air pressure to show a variety of expressions, and can also make some simple movements, which are very lifelike.

When will the host be replaced?

Back to the online Spring Festival Gala, the appearance of the four naughty virtual hosts received a lot of attention, and from the audience's attitude it can be seen that they like them quite a lot.

So will this be an opportunity for the host to be laid off? Maybe Xiao Sa and Xiao Xiao Sa will not agree.

As for the tone and awkward facial expressions of AI news broadcasts, there is still room for improvement. Japan’s once popular news broadcasting robots have not replaced the local journalism industry on a large scale, and at most they are still at the level of assistance and novelty.

So from this perspective, technological advances do bring us new and wonderful experiences, but perhaps in an era like this, gimmicks go faster than reality.

Maybe the day of being replaced will come eventually, but it is definitely not today. We should believe that when that day comes, humans may have already figured out how to get along with AI. During the Spring Festival at that time, we will see not only AI hosting the Spring Festival Gala, but also AI performing shows. We can just open our mouths and wait for AI to feed us. 

Maybe the day when AI is replaced will come eventually, but it is definitely not today. We should believe that when that day comes, humans may have already figured out how to get along with AI. During the Spring Festival at that time, we will see not only AI hosting the Spring Festival Gala, but also AI performing shows.

As for us, all we have to do is open our mouths and wait for AI to feed us.