The AI cover is on fire.
At present, the hottest is AI Wang Xinling's cover of Mei Yi Liya's "Pure Land of Bliss", the combination of two otaku gods and goddesses, completed the dream linkage in the hearts of some people, and even some people listened to the songs covered by AI and began to imagine whether they could really see them working on the same stage one day.
There is also AI Sun Yanzi, who has always sparked heated discussions, and is the top of station B. Click on station B, you are very likely to find that "AI Sun Yanzi" "cover" a series of songs such as "Mercury", "Hair Like Snow", "Melody", and even "Kill That Shijiazhuang Man" in a prominent position on the home page, and search for "Sun Yanzi" on station B, and you will also find that "AI Sun Yanzi" has been mixed into it, and the songs of "cover" are often more than one million views.
Sun Yanzi, who has not released a music album for six years, was ridiculed by netizens that she would be unemployed if she did not release an album. Although it is a joke, but the rapid development of AI technology, does sound the alarm for many practitioners, once AI technology becomes more mature to the extent of fake real, singers, will it really be replaced?
Why listen to AI sing?
AI singer cover, as the name suggests, is to use the singer's voice material to "feed" and train the AI model, and then replace the original song you want to "cover", and after post-processing, a "cover" is formed.
In addition to Sun Yanzi, the most popular AI top stream on the Internet, Jay Chou, Faye Wong, Xu Song, including A sister, Lady Gaga, Taylor Swift and other foreign singers, have also been made into AI singer covers by netizens.
If you listen carefully to the songs covered by these so-called AI singers now, they are not ten-year-old fans, and it is difficult to quickly identify whether they are AI or real people, because after the "feeding" and training of a large number of materials, their timbre is very close to real people. If it is a ten-year-old fan, although it is not indistinguishable, it is very likely to tell from the AI's bite which period of material it is "fed", which proves that the AI sound is highly close to the real person.
And these singers can be imitated, because on the one hand, they are well known to the public, many songs have been out of the circle for many years, on the other hand, their timbre is very recognizable, and their singing skills can be maintained at a high level, so that both listeners and AI can quickly find characteristics from them, complete the combination, and the look and feel is not considered a violation.
The explosion of AI singers' covers is the epitome of today's "fast food" era, and some people have a curious mentality to produce and listen to music that combines singers and songs across regions and genres. For example, A sister covers Han Hong's "Heavenly Road", LadyGaga's cover of "Standing on the Grassland Looking at Beijing" and other groups, just like the ghost animal songs that were once spliced together at station B, the more unexpected they are, the more they can meet the needs of curiosity.
There are also some people who do not care about music itself, for them, music is just a simple entertainment that is the easiest to obtain in a tired life, and many times just "listen to a sound". So whether it is a singer or an AI singer cover, as long as the timbre is not flawed, the atmosphere created is in place, and it makes no difference whether it is sung by myself or by a machine.
It is precisely for the above reasons that the basic listening needs of some listeners are met by AI singers, and there is a market for AI singers, and it is logical that AI singers are popular for covers. But at the same time, this situation also makes the singer himself "alienated" to a certain extent, and the value of the person as a "singer" itself is weakened.
Can AI replace singers?
Recently, AI has made the topic of Sun Yanzi's career very popular, but can AI really replace singers? Or in other words, these singers at the level of heavenly kings and queens who have been popular in the Chinese music scene for decades led by Sun Yanzi can be loved by countless people for many years, just because the timbre is recognizable and the singing skills continue to be online?
Of course not, the charm of a singer is not only the charm of being a singer himself, but also his own unique personality charm. Just as many singers have imitators who are extremely similar to their timbre and appearance, but these imitators can only stay in the stage of imitating them, the connection between people who like singers and the singer himself cannot be replaced.
Even sometimes I like a singer, not only his songs, but also the joys and pains that he experiences together on this road because of fate, as well as important moments in each other's lives. This kind of thing that even imitators cannot do, and AI without vitality cannot do it.
And even if the AI cover can restore the timbre, it cannot restore the emotions injected by the singer into it. Because when a singer sings a song, he will be affected by the current mood, state, and life experience. Choking, forgetting words, changing words, and even inevitably deteriorating in voice conditions as we age, those possible imperfections are precisely what AI perfection cannot replace.
Andy Lau's concert was re-screened last year, and the choking when singing "Seventeen" touched countless people, which was his retrospective of decades of acting career standing at that moment, and it was also a testimony of the audience walking through youth with him. When Andy Lau sang "Like me, don't cover your face, let passers-by find out", the emotional concentration injected has gone beyond the song itself.
Returning to songs, concepts and production, which are also irreplaceable by AI. Chen Shanni launched the song "Teach Me How to Be Your Lover" sung by herself in March this year, according to her, the singing priorities and breathing pauses of this song are all generated by AI. Because the singer himself tuned, the timbre and singing texture were so similar that the sound engineer was deceived.
Many people feel that this is the threat of AI to singers, but Chen Shanni herself feels that the ingenuity in the music production process belongs only to people, and the aesthetic experience and thinking cognition of people as "creators" are the core characteristics that will not be replaced in the forward wave.
Therefore, the emergence of AI singer covers, the real replacement is not Sun Yanzi, but an assembly line singer with empty technology, timbre, and no thinking as a creator. However, challenges are also opportunities, as Chen Shanni gave a thinking question, "When AI really learns to sing, it is time for creators and singers to re-understand their own value."
Inevitable pitfalls
At present, AI is still in the development stage, and most of its functions are controlled by humans, so there are certain problems that need to be solved, whether legally or morally.
The good and evil of human nature is only a thought, and AI can be given the temperature of human nature. Someone on station B used AI to simulate Yao Beina singing "Let Her Land" and Cui Xueli singing "Ditto", although it could not completely restore the timbre and the processing of some paragraphs, but it allowed the two singers to live in people's hearts in another way, and moved people who liked them to reunite.
AI will also become a wounding sword because of the evil of human nature. AI face changing was once all the rage on the Internet, originally just netizens entertaining themselves to replace actor episode clips, but today some people have begun to use such technical means to transfer flowers, fabricate facts, insult and slander stars and even amateurs.
It is hard to imagine that when the sound technology of AI is not limited to the fake reality of the song cover, but can directly simulate or replace people's daily conversations, will there be people with ulterior motives, using the dual technology of changing faces and changing voices, completely fictionalizing the lives of real people, causing indelible damage to them.
In terms of law, although the singer's voice is not protected by the Copyright Law, the process of making and releasing AI songs will still infringe the singer's voice rights and violate the relevant provisions of the Copyright Law.
The most typical case of copyright is that not long ago, foreign netizens used AI training to say that the voices of Drake and The Weeknd were made into the original song "Heart on My Sleeve", which was clicked and played more than 8.5 million times on the TikTok platform, bringing considerable benefits to netizens. The popularity of this song made the copyright owner Universal Music quickly remove it. Since then, other record labels and the Recording Industry Association of America have publicly stated that covers of songs using AI cannot be copyrighted.
Michael Nash, chief digital officer of Universal Music Group, elaborated on the impact of AI on the living environment of creators such as independent musicians: "State-of-the-art AI can now generate derivative products that closely mimic the unique style of the original creator. Often, it simply creates a whole bunch of imitations — diluting the market, making original works harder to detect, and infringing on the artist's legal right to compensation from his work. ”
On November 25, 2022, the Cyberspace Administration of China, the Ministry of Industry and Information Technology, and the Ministry of Public Security jointly issued the Provisions on the Administration of Deep Synthesis of Internet Information Services, which supervises and administers technologies that use deep learning, virtual reality and other generative synthesis algorithms to produce network information such as text, images, audio, video, and virtual scenes.
In April this year, the Cyberspace Administration of China issued the Measures for the Administration of Generative Artificial Intelligence Services (Draft for Comments), which proposes that organizations and individuals that use generative AI products to provide services such as chat and text, image, and voice generation should assume the responsibility of producers of content generated by the products; Respect the legitimate interests of others, prevent harm to the physical and mental health of others, damage portrait rights, reputation rights and personal privacy, and infringe intellectual property rights. It is prohibited to illegally acquire, disclose, and use personal information, privacy, and trade secrets.
epilogue
The explosion of AI singers' covers is a popular carnival based on existing technical means, but under the carnival, there are still many problems that need to be faced or urgently solved.
Whether you are a creator or a receiver, you are inevitably in it. At present, instead of passively discussing "whether it will be replaced", it is better to actively think about "what else can be done", as Chen Shanni said when explaining the concept of the AI song "Teach Me How to Be Your Lover", "What the future will change is decided by everyone who is moving forward now, and we must be careful." ”