How do you know if the music is a performance?

There’s a new song doing the rounds, and in the immortal words of Kylie Minogue, you can’t get it out of your head.

But what if it was created by a robot, or the artist who himself is a product of artificial intelligence (AI)? Do streaming sites have an obligation to post music as created by AI? And does it matter, if you like what you hear?

A survey published last week suggested that 97% of the respondents can’t find an AI song But there are some tell-tale signs – if you know where to look.

Here is a quick guide.

AI Music became one of the hottest topics of the summer after accusations the band Velvet Sundown was Ai-GRUPATED sent them viral.

The band, without a record label and a small social media presence, quickly gained hundreds of monthly listeners on Spotify in the weeks that remained.

The band initially denied the claims, later defining themselves as a synthetic project “guided by human creative direction, and composed, expressed and viewed with the support of artificial intelligence”.

They claim that the project is an “artistic challenge”, not a trick, but many fans feel it.

Internet sleuths suspect Airbrushed Photos of the Band, with non-descript backgrounds and a warm orange filter.

There is no record of them ever performing live – no glowing reviews from fans posted online, or any concert photos or videos. the The band members did not give interviews and does not appear to have individual social media accounts.

Looking at the real life and social media pressure of an artist can be a helpful indicator of whether they are genuine. But experts told the BBC that increasingly powerful, sophisticated technology means it’s increasingly difficult to tell when a song has been created using AI.

However, while it may be deceptive, they say that there are signs that the audience can be alert to.

When LJ Rich started making music with AI about five years ago, he remembers how it could generate three seconds to a minute of audio.

Now, a whole song can be summoned quickly with a quick, which inspires what industry experts defined as an “explosion” in AI platforms – in streaming platforms.

A Song with a formulaic feel – sweet but without much substance or emotional weight – could be a sign of AI, said the speaker of technology and technology without words.

AI songs tend to stick to generic verse-chorus structures, and usually don’t have a satisfying ending. AI also tends to produce lyrics that follow the correct grammatical structure, as rich, while some of the most beautiful or awkward words written by humans do not always make sense.

Just ask for the keys to Alicia and the concrete jungle where dreams are made of “, or the rocks and their flirtation with the double negative of (I can’t get) satisfaction.

“If it doesn’t feel emotional, it’s a big part,” the BBC’s former click continued. “Does it make tension and resolution a fundamental part of the music we love? Does it have a story within it?”

Another signature story is unrealistic productivity levels. Professor Gina Neff, from the Minderoo center for technology and democracy at the University of Cambridge, describes how an artist was recently believed to be an AI after releasing several sound albums.

Their songs are like a mashup of 80s Rock Bands – like “True Classic Rock Hits Put in a Blender”.

“It’s good for background music for most people,” he continued, “but it doesn’t work for creating today’s future superstars with something new out of it.”

Sometimes, what could be a song is a song that is almost perfect, lacking small mistakes and differences.

This can be explained by the lack of strain on the vocals, and overly polished production, says Tony Rigg, music industry manager at University Industry Management at the University of Lancashire.

He adds that odd references, unnatural emotional delivery, and lyrics that feel generic or repetitive can also be signs.

“Ai has never felt sorrow… knows the patterns,” he explained. “What makes a human being is not only good but the stories behind it.”

The vocals are also worth paying attention to. The “singers” always sound a little slurred. Consonants and plosives (hard sounds like “P” and “T”) are incorrect. You can hear “ghost” Harmonies, where backing vocals appear and disappear at random.

However, Rigg calls these signs “non-proof signs”, recognizing that it is not easy for the casual listener to detect AI songs.

As well as being used to create entire songs, AI is also a tool that some established artists are using to support their creativity.

Currently there is no obligation – or consistent way – for artists to let fans know how they use AI.

Some are very open: The Beatles, for example, used machine learning To get John Lennon’s voice from a 1970s Cassette Recording to release what they call “the last song”, now and then, in 2023.

And artists including Hogen Heap and Timbaland have created AI Personas and released singles under their names.

Last month, Heap released the song on the arrival of his AI Model Ai.mogen, who was trained with his voice.

He created the voice model as a chatbot – a “desperate attempt” to keep up with a flood of messages and requests that are accompanied by time constraints.

While “I know something different if you know my voice”, he said that he put a lot of work into the creation of the AI ​​version and did not think that the listeners could talk to the listeners.

HEAP is not trying to mislead listeners – AI.MOGEN is listed as a co-cributor on the track.

But he hopes if people feel a human connection to the song, unknowingly about the vocals being sung by his eInceaved AI ideas about AI.

“I hope people listen, don’t know, find peace in that,” he told the BBC.

He said he’s not against using AI to actually make music, but it’s not something he’s ever gotten to do.

Thap believes there should be more transparency around what goes into a song, and how AI is used.

Citing the example of someone reading the label of a microwave-ready meal so they know the ingredients, he says: “We need that for music, and we need that for AI.”

There is currently no legal obligation for streaming platforms to label AI songs with AI calls, despite increasing calls for them to sign up for such tracks.

In January, Streaming Platform Deezer launched an AI tool, followed this summer by a system that Tags Ai-Mindala Music.

Deezer says its detection system can flag tracks made with more advanced AI music creation tools, and is working to expand its ability to detect music created by others. It says the risk of false positives – for example incorrectly flagging a track that someone has made – is very low.

This week, the company said that a third (34%) of content uploaded to its platform is fully produced – about 50,000 tracks a day.

Manuel Moussallam, Deezer’s research director, said his team was surprised by how many tracks were flagged to “convince us we have an issue”.

The tool quickly flagged the music of Velvet Sundown – the band that went viral this summer – as “100% Ai-Wrate”.

Other platforms have recently announced steps towards greater transparency.

In September, Spotify said it would Roll out a new spam filter After this year to identify “bad actors”, and prevent “slop” recommended by listeners. Last year, more than 75 million spam channels were removed.

It also supports a way to tell artists where and how AI is used in a track, through a system through a consortium of industry members called DDEX. This information is included in a track’s metadata and displayed in its app.

Spotify says it’s about recognizing listeners’ desire for more information, as well as “building trust”.

“This is not about punishing artists who use AI responsible ranking tracks for revealing information about how it’s done.”

When you fall for a new artist, does it matter if they or their songs were produced by Ai?

Some believe that the presence of AI is irrelevant – the engagement driven by pleasure, and the people that people love already serve its main purpose.

Some say that music fans should be able to make informed choices about what they listen to.

Artists share deep concerns about the impact of AI, and hundreds of musicians as well Dua Lipa and Sir Elton John have protested the use of their songs to train AI tools.

For LJ Rich, the use of AI in music raises many “weird and beautiful ethical questions that have yet to be answered.

“Like music that makes the hairs on the back of your neck stand up, does it matter if it’s written or not?”