When AI starts singing, can human singing still be heard?

Beijing, April 14 (Reporter Wang Shiyao) "Every note, every breath is sung by AI. "Although AI singing is not new, the official release of AI songs by professional singers still shocks many people. This also makes us wonder, when the AI starts to sing, can the human song still be heard?

A while ago, Taiwanese singer Chen Shanni released her latest single "Teach Me How to Be Your Lover", but a few days later, she announced that the song was actually sung by her own AI model.

At the same time, she also announced her original intention on social media, "In the current hot discussion of AI development, I hope to think with all those who care about creation through this song - if the era of AI is bound to come, what creators should care about may not be 'whether we will be replaced', but 'what else can we do'." ”

Singer and producer Chen Shanni. Photo courtesy of interviewee

Simulation and reality

From the moment the machine is built, "Can robots replace humans?" "It has always been a topic of debate, and then robots have a new name - artificial intelligence (AI).

With the rapid development of the times, some traditional industries gradually transform into the direction of science and technology, industrialization and intelligence, and some of the manpower is indeed replaced by AI. However, as a carrier of human emotions, whether art can be calculated and reproduced, and how receptive the audience is, is still the focus of public attention.

Novels, paintings... In these art fields, mature AI products have been launched, such as Microsoft "Xiaoice", an artificial intelligence poetry writing system, which published a collection of modern Chinese poetry "The Sunlight Lost the Glass Window" in May 2017; In October 2018, an AI-created portrait, Edmond Belamy, was successfully auctioned for $430,000 at Christie's in New York.

Data map: The robot's live interpretation of Peking Opera attracts attention. Photo by Tang Yanjun

Nowadays, AI songs produced by professional singers and producers have also begun to be officially released, and its "simulation" degree can even be fake and real, deceiving many professionals and true love fans. How is this done? Chen Shanni said in an interview with Chinanews.com that the premise of realizing the fake truth is that she provides a large amount of reference data for AI. "Because the AI training materials I provided spanned works from 2004 to 2022, AI vocal made this possible by learning to sing."

However, even the singing of the same person may present different effects due to differences in sound status. In the comments at the beginning of the song's release, we found a few comments mentioning that Chen Shanni's voice seemed to be "rejuvenated". Chen Shanni also replied: "Many of my works tend to be electronic style, but this new song is a relatively rare type of music since "Love Song" in 2004, so the sound learned by AI is placed in this song, and it is indeed possible to present such a result." ”

So, will a more perfect singing voice move the audience more? Fan Li Yu told reporters that as a music fan, he actually felt that there is a difference between AI singers and real people at the beginning of listening to songs. Listening to it after the answer is revealed is a state where the novelty gradually becomes cold.

Another fan of Chen Shanni, Xingxing, said that whether she knew the "truth" of the song did not change the work itself, but what changed her attitude as a listener, and she may rarely click on it later.

Infographic: The music library not only provides original genuine CDs, but also 5TB of high-quality digital music resources. Photo by Zhang Yao

How to reinterpret the singer's dream?

In Chen Shanni's view, the transformation of the music industry has always been ongoing. Around the turn of the millennium, when the entire pop music industry began to go digital, there were many good sound engineers who did not agree with this trend and thought that the sound quality had deteriorated. "So far, I still don't fully agree with the so-called deterioration of sound quality, I think it is a disagreement about the beauty of the sound texture, and of course, the sound specifications of the streaming will change later." She said.

Chen Shanni, who had become a producer at that time, also accepted this change in the music industry, her idea was very pure, "If we can pay less time cost and lower the threshold for participating in pop music, can we allow the pop music industry to accommodate more different ideas and imaginations?" ”

"Through AI, everyone can realize their dream of being a singer in the future", this is Chen Shanni's thinking and prospect for the music industry. However, is the lowering of the threshold for singers a good thing for the long-term development of the industry?

Infographic: Tape. Photo by Yin Liqin, reporter of China News Agency

"I can't say it's a good thing." She pointed out that in fact, the threshold for singers has been lowered, which is not a situation only for AI songs. Human learning of singing skills and sound discrimination is rapidly decreasing. In the past, singing skills that people thought could be determined through live performances can later be adjusted through technology.

Chen Shanni added: "I saw that some record companies have purchased more convenient Auto-tune equipment, so that the singer's voice during the performance can more easily satisfy the listener's ears, and the singer will naturally shift his energy to the entertainment of the performance and other things such as community management. ”

If the process of lowering the threshold of singers is irreversible, then how to reinterpret the dream of singers is another possibility that as a producer and creator in the industry, you need to look for.

Will it not be replaced by AI?

AI singers may impact the singers themselves, but producers will not be replaced by AI. Chen Shanni said that because AI vocal needs professional music producers, guiding AI with human aesthetics and experience still reflects the producer's ability and his imagination of music.

At this point, Jay Chou has also made similar remarks. Earlier, he posted the MV of the song "In the Name of the Father" released in 2003 on social media, with the accompanying text: "Although AI can do a lot of things, it cannot replace my beauty in music creation." So I don't need to worry, I know this since before AI appeared. ”

Jay Chou social media screenshot

It is worth noting that while we are discussing whether singers and producers will be replaced by AI, there is another group of voices that need to be paid attention to, they are the audience of music products - fans.

In Li Yu's view, real singers will never be replaced by AI singers, and those who can be replaced may not be called singers, "they are just commodities launched by capital, and commodities will certainly be replaced by new commodities."

As for the singer's attitude towards officially releasing AI songs, Li Yu feels that this is just a singer who keeps up with the trend and tries a new production mode. It's as if an urban bitter singer tries to sing dance, or a pop singer suddenly wants to try jazz, it won't make much difference to the listener.

Xingxing took a wait-and-see attitude, "If this is the inevitable trend of technological development, it is not a problem that I will not accept." "For me, music is explaining emotions that words can't accurately convey, and if I can accept that AI will replace singers, it must be because I can find the carrier of emotions in the vocal of AI." (End)