Some people are using AI to continuously generate memes, and some people are using AI to generate music; the same is AIGC to generate music, some are imitating humans, and some are replicating classics. AIGC is a tool that produces different values ​​in the hands of different people, and this number may reach trillions of dollars.
As 2022 draws to a close, AIGC has become one of the most-watched technology keywords this year. Only a few words of description are needed to allow AI to associate and draw, and through the most direct and fastest visual feedback, it will have an impact on human beings' conventional understanding of art.
If we look forward to the future, as Gartner calls "one of the five most influential technologies in 2022", by 2025, the data created by AIGC can account for 10% of all produced data.
Although the current AIGC is more fully visually competitive, developing faster, and more eye-catching, music is an important carrier of human culture and emotion, and AI is actually transforming the industry just more subtly.
Tags, algorithms, neural language recognition, and other functions have almost become the standard configuration of music apps. They can automatically generate song lists and push them to users. Now you open QQ Music, there should be many people who have directly given the choice of 30 songs per day and personal recommendation".
If anyone wants to experience the feeling of being ridiculed for music taste, then AI can also provide intimate services. In late 2020, the online magazine The Pudding launched an artificial intelligence app, How Bad Is Your Spotify?, which entered its database with millions of Pitchfork comments, Reddit posts, and record store recommendations At one point, enthusiastic Twitter users even flooded the Pudding site by analyzing users' Spotify playlists, mocking their junior musical tastes or their obsession with an artist "to an uncomfortably high level."
AI's transformation of music doesn't stop at interest analysis or taste appreciation. The American drama "Scorpion (S2E8)" (Scorpion, S2E8) aired in 2014, told a story that was quite sci-fi at the time: a blogger musician wrote a program that could continuously produce "Divine Comedy", but he was killed because of it Disaster - the murderer is another creator who uses programs to write songs for profit.
In the past 8 years, thanks to the efforts of the big platforms, the fantasy of the "Scorpio Project" has quickly become a reality. AI is finally about to take a real look at the music creation process!
In June 2016, the world's first 90-second piano music composition created by artificial intelligence was born. It came from the "Magenta" project jointly launched by Google and CalArts.
Since then, there has been more and more AIGC music, and the underlying technology has been divided into two paths.
One way is automatic composition. In 2017, Sony CSL (Computer Science Laboratory) artificial intelligence program FlowMachines launched Daddy's Car, which is the achievement of AI composition, known as the first complete song composed entirely by artificial intelligence; Microsoft’s AI learned dozens of Christmas carols through analysis and released an AI song called Joyful Time in The City.
Compared with 90-second works, the complete AI automatic composition is a big step forward. But they still need to draw inspiration from the vast historical musical works of human beings, which is an extension of human artistic inspiration. For works like "Daddy's Car", the composition and arrangement are done by AI, and the harmony and lyrics are helped by human composers Done; Happy Hour in the City needed human assistance, too. In fact, this kind of music work cannot be said to be "completely created by artificial intelligence", but more of the self-indulgence of the technology circle, and the actual acceptance of the audience has to be questioned.
Another way is to focus on "AI automatic synthesis of music works". For example, the sound synthesis technology engine "Lingyin" developed by Tencent Music's Apollo Sound Lab, highly restores and reproduces the singer's voice characteristics, and reproduces the artist's classic voice. The first AI singer was released in July this year, based on Yang Chaoyue's voice synthesis, Only about 10 minutes of recording by the singer is needed to reproduce the original sound very well, and even bring "emotion" into the singing. Later mature works include "Letter Not Sent" (reproduced and reproduced Teresa Teng's voice), and "May You Be Treated Gentlely by the World" (reproduced and reproduced Anita Mui's voice).
The difference between "Auto Composer" and "Lingyin Engine" is that "Automatic Composer" imitates music, while "Lingyin Engine" uses a self-designed deep neural network model, plus "synthetic music" with existing sound style elements, the latter does not shy away from using the best works of art in human history and re-enacting these artistic styles.
This is also another development direction of AIGC-like music - instead of letting machines make original music, it is better to let machines assist people to create music, and it can do more than simply output songs, changing the way we experience music in the past.
Douglas Eck, the technical lead of Google's AI music project Magenta, said in an interview with Science that AI was already a chess master, but now it has become a helper for humans to improve their chess skills, and so does music, no matter how fast the machine learning speed is, humans can adapt and use AI technology at a faster speed to assist the creation and play new ideas.
To put it bluntly, AI doesn't have to replace humans, it's good to be a handy tool. And the content generated by AI tools may not be in reality but in the virtual world. That is, how to completely digitize the people in the middle.
AIGC's music has evolved to this day. It is different from the automatic speech synthesis technology in the past but has a higher complex scale. The AI ​​voice rendered by the engine is closer to real people in terms of timbre and tone, and it is also more expressive in singing. "emotion". At present, "Lingyin" technology not only allows AI to sing (officially called "synthesis") existing songs but also allows users to customize synthetic songs written, composed, and sung by AI with one click.
In this environment, the question people face is: how to make the metaverse digital people sing, dance, rap, and play basketball?
This is not easy, because not only music is needed in the video world, but also music (instrumental/human) sounds, mouth shapes, expressions, movements, and other elements are required. In richer scenes, dance, mirror movement, lighting...
If you use the "Music XR Maker System" of Tencent Music Tianqin Laboratory, these requirements can be fully realized, and the application scenarios are diversified. The first virtual human Xiaoqin of the system was launched in October this year. It is driven by AI and based on Music is used to generate dance performance and singing elements, with the purpose of improving operational efficiency and the authenticity and fun of virtual people. This also provides better solutions for scenarios such as conferences, customer service, and news broadcasts. It is understood that soon there will be a "super singer" who has come from afar from the "next generation" and has been selected to debut in TME. Perhaps she will subvert our imagination of singing and become the first year of the "music virtual idol". "The Ziwei star from the sky.
Zhou Wenjiang, Vice President of QQ Music Technology said, different from the large investment in customized head virtual idols, we hope to build Music XR Maker into a platform for high-efficiency and low-cost production of digital humans, and at the same time enable digital humans to have stronger real-time content production capabilities, allowing any music The interactive combination of music, dance, stage, and other elements is possible.
In 2022, Zuckerberg’s Metaverse social platform Horizon Worlds, which was burnt out with US$10 billion, has an impressive retention. On October 16 this year, Meta’s internal documents were exposed, which showed that the number of monthly active users of Horizon Worlds was less than 20. 10,000, which is far from the original expectation of 500,000, and the interest duration is only one month.
It's funny to say that as a virtual world, what the Metaverse lacks the most is people. The most difficult thing is to create good soil to attract people to create together.
In comparison, another metaverse case in China is more pragmatic. On July 16 this year, Tencent Music’s virtual music carnival TMELAND and PepsiCo created an immersive virtual 3D performance. This virtual music performance provided the content and theme for the metaverse experiment, and this commercial cooperation also provided a "standard template" for future virtual scene marketing. This may be the first case in which a large commercial customer pays the bill in the domestic AIGC field. The support of many brand owners to the content co-construction of Metaverse will also enrich the ecology and form a viable marketing scene.
Looking at the music industry, from the early 1.0 recording era, the output of music content was concentrated in the hands of a few professionals with resources; entering the 2.0 streaming media era, more music practitioners participated, and the recommendation and distribution mechanism gave them opportunities to showcase; With the advent of 3.0, AIGC even allows music consumers to have the ability to reverse output and participate in co-creation. To continue to strengthen investment in AI, the future of the music platform may be more fertile soil than imagined.
Enough highly sticky users, people never tire of appreciating music, established payment habits, gradually advancing co-creation tools (not directly involved in the creation and even avoiding copyright disputes), and diversified scene collisions, who said Isn’t the music platform a better soil than the gamified metaverse?
Metaverse is never something that can be achieved overnight. We can use our mature technology to implement a basic version, and then continue to improve various technologies to improve the metaverse experience. More possibilities for music + AI Dong Zhi, head of computer vision at Tianqin Laboratory, said, our core capabilities in AI-driven, they all have one thing in common, which is to generate various basic elements of the virtual world through music and AI..
From AI writing to AI painting, to AI music and video, AIGC has always faced some controversies: If the generated content is not good enough, what is the meaning of its existence? If the content it generates is good enough (and it may be in the future), will it spark a new wave of career replacement scares? Will human spiritual entertainment be taken over by AI?
This may be a series of entangled and eternal topics. Techno-optimists often say that a tool is only as good as the purpose of the user.
In fact, what the Internet has been doing is making it easier for more people to enjoy the fun of technological advancement, and the same is true for the music field. If we open tools such as Lingyin Engine and Music XR Maker to all musicians, production organizations, or content companies in need in China, and then use Music Metaverse and AIGC to help listeners enter the music experience by combining virtual reality as soon as possible, the future may even To become a powerful tool for all people to create music in the field of music so that everyone can make music according to their own imagination and enjoy music, that may be a more promising future.