The owner of Bilibili used AI analysis to find out that these singers lip-synced during the New Year’s Eve concert.

Singing skills and timbre are important criteria for measuring a singer's ability, but audiences have different opinions on singing skills, not only because their preferences are not objective enough, but also because they encounter tuning and lip-synching from time to time. Therefore, for a long time, when it comes to comparing the singing skills of different singers, it will always cause huge controversy.

During a live concert, no one can sing "perfectly" due to being separated from the perfect environment in the recording studio. However, with the advancement of technology, on some occasions, singers He knows how to tune and lip-sync, and since then millions of tuners have ridiculed him for comparing him to a human CD player.

Since tuning uses more and more advanced technical means, singing levels should actually be quantified and compared using technical means. Those "pretends" that are inaudible to amateur ears are pale in front of data.

Yesterday, the topic of identifying lip-syncing at New Year’s Eve concerts ranked first on Zhihu’s hot search list. Someone used this idea to do a thorough research.

On January 7, the owner of Station B Up, "Coder Gao Tian", a programmer working at Microsoft analyzed cross-border Check the pitch of the singers at the annual concert to see who was out of tune. By the way, I also found some people who were suspected of lip-synching.

##Link: https://www.bilibili.com/video/BV123411m77L/

As the master of technical Up, programmer Gao Tian mainly did not evaluate the level of singers, but introduced the analysis method.

Who is lip-syncing?

Let’s take a look at UP’s method of identifying “lip-syncing”.

First, intercept a short piece of the original sound sung by the singer. Here, UP takes Zhang Jie's "Reverse War" as an example, using software to separate human voices to avoid interference from musical instruments. The spleeter is used here, and behind it is a neural network built with the TensorFlow framework.

## Project address: https://github.com/deezer/spleeter

Then use the Python speech library Parselmouth to perform pitch analysis on the human voice.

The audio is visualized as a scatter plot:

UP Based on the key, add the natural scale lines of the song to the picture. The frequency corresponding to the line is the so-called accurate pronunciation. Most people who have not received professional training are actually not that sensitive to pitch (only Jay Chou and Leehom Wang have absolute pitch), which means that without digital interpretation, it is difficult to tell whether the singer is singing accurately or not. After visualization, the conclusion is clear.

Zhang Jie’s intonation is still very good

But on the other hand , people are not machines, they always have ups and downs, and it is difficult to achieve perfection in live singing (even if some singers are evaluated as "eating CDs"). Zhang Jie can already be considered to be at the top level of the Chinese music scene. Also at the same level are Deng Ziqi, Sun Nan (no need to analyze the big guys), Tan Weiwei, Zhou Shen, etc.

Deng Ziqi's "Countdown"

##Sun Nan's "See You or Be Parted"

## Tan Weiwei "Mountains and Seas"

## Zhou Shen "Light" Hearing these names and seeing their pitch analysis charts, you may also be able to understand why these people can represent the top level of the Chinese music scene.

Following the same method, the UP host analyzed Joker Xue’s scene and presented a “scatter diagram”:

This is "The Lonely Warrior" performed by Joker Xue this year, which has been criticized by all kinds of people...

Why in so many New Year's Eve concerts? Among them, his singing is the worst to listen to?

The reason is simple, because people who sing worse than him seem to be "lip-syncing"...

This is like a In the class exam, Xue Zhiqian's level was originally in the middle-to-upper range, but because all the students except the real top students "cheated", he could only get the "last" ranking.

Of course, this conclusion is not groundless, and the UP owner also gave evidence.

Lip-syncing can be completely distinguished from the sound. First of all, if a person does not make a living from singing, but his live performance is comparable to or even better than that of a professional singer, then it is very likely that he is lip-syncing. For example:

It's also "The Lonely Warrior", and every note of a certain artist falls on the line perfectly and accurately.

Then an interesting question arises. For a person who does not make a living by singing, what is the effect of singing?

Such materials should be very precious. Fortunately, we can still find some, such as this:

Trigger Controversy

Analyzing a singer with many fans at a concert will naturally cause controversy. On January 11, coder Gao Tian stated in the second video that he had received another letter from a lawyer protesting.

## Video: https://www.bilibili.com/video/BV12Y411y7Hr/

The author said that perfect pitch cannot accurately reflect the fact that the scene is lip-syncing, because there is still room to modify the sound between the non-live broadcast scene and the video that people see. Although some people have corrected the sound, some people have not. .

This sound may have been recorded and revised in the studio, and the lip-syncing on the spot was lip-syncing; or it may be that the vocal was actually sung live and then revised after the singing.

This also means that there is unfairness. Pitch modification and even lip-synching lower the threshold of singers, making it difficult for ordinary listeners to compare. For us, although everyone has different preferences for singer styles, at least they have the same pursuit of true singing. From this analysis, we can see the differences in singing skills.

No matter what kind of singer you are or what difficulty the song is, it is basically impossible to maintain perfect pitch for 5 minutes in a large stadium. As the phenomenon of "perfect pitch" becomes more and more common, have we also lost more opportunities to appreciate the charm of live?

The above is the detailed content of The owner of Bilibili used AI analysis to find out that these singers lip-synced during the New Year’s Eve concert.. For more information, please follow other related articles on the PHP Chinese website!