Recently, Wang Zhongyuan, vice president of Kuaishou Technology, was a guest of the "Future Media Interview" program intelligent media feature jointly launched by Sina News and Cover News, sharing the possibility of "AI music + short video", discussing the advantages and shortcomings of AI singers, and looking forward to the future of AI music.
In Wang Zhongyuan's view, the three stages of short video creation, understanding and distribution have integrated a large number of AI technologies, which makes the threshold for creation lower, among which AI music and AI singer projects make music creation enter the inclusive stage, and users can create their own songs in a very convenient way.
In Kuaishou, a national short video APP, behind the rich content ecology, music is an important factor that motivates users to create. According to statistics, 76% of Kuaishou's short video works have a soundtrack, and 90% of Kuaishou users expect short videos to have a soundtrack.
Wang Zhongyuan believes that the strong dependence of Kuaishou users on music when creating short videos once again proves the magical charm of the art of music since its birth. With the development of technology, music has changed from an art form owned by the privileged class to an art form that ordinary users can appreciate.
In terms of music creation, AI can create music and sing songs, making music creation enter a new stage. "It's no longer a unique ability of musicians, but the average user can try to create their own music and sing their own songs." Wang Zhongyuan said.
In addition, AI music has also created new development space for the short video field. Wang Zhongyuan said that Kuaishou has invested a lot of AI engineers, using AI models to help with music creation and research and development of AI singers. "In the research and development of AI singers in the past two years, we have gone through two major stages, from the traditional WORLD sound coder based on signal processing, to now there is a network sound coder, which is based on big data, through repeated learning, so that the AI model can imitate the singer's singing."
Nowadays, the singing level of AI singers is basically close to that of an ordinary live singer, but in Wang Zhongyuan's view, THERE is still an essential difference between AI singers and live singers, and he does not expect AI singers to replace real singers. "It's actually more like a fun form of interaction, helping users interact, appreciate, and sing their own songs with their own tones."
In fact, AI music creation is not much different from traditional music creation in terms of process, but in every link, AI models can help people explore new possibilities, and it is very convenient and fast. This advantage can be better played in the era of short videos, "after integrating INTO AI, a large number of users can create music, and at the same time, through the consumption behavior of other users, select those divine comedies that are really very popular, and popular music and explosive gameplay can emerge faster."
In response to the copyright problem brought about by AI music, Wang Zhongyuan also put forward his own views: "Some of the copyright should belong to the research and development team of the AI model, but on the other hand, if the user creates a song based on AI, there are many unique adaptations and creations of their own, then the copyright issue I think can be further discussed." ”
There is no upper limit to the development of technology, when the real world and the digital world are converging, AI models become larger and larger, gradually breaking through the parameters of tens of billions, hundreds of billions, trillions, and the creative imagination space is becoming more and more extensive. Wang Zhongyuan incorporated the concept of metaversics into the layout of Kuaishou AI technology: "It is conceivable that in the near future, AI singers and live singers will have dialogues in different time and space. ”
It is understood that Kuaishou is about to launch a hyper-realistic virtual singer that combines AI music, AI singers and virtual human technology. According to the relevant product personnel, this virtual singer combines digital modeling technology and content understanding and generation technology to create a new form of musical expression, which can make music creation more universal and more diverse, and is committed to making every user feel the beauty of music.
It is worth mentioning that Kuaishou will also combine virtual human technology and e-commerce business to help merchants sell goods. Kuaishou small shop merchants can realize virtual people with goods live broadcast through the "old iron intelligent live broadcast room". Kuaishou provides a virtual digital human anchor and a virtual human live broadcast operation platform to realize the live broadcast room of its store, so that the virtual human e-commerce anchor can introduce and sell goods and interact with the audience like a live e-commerce anchor. It is understood that the project is in the internal testing stage, and will open cooperation with merchants after going online.