AI beats 95% of famous musicians, forum user says

AI music and the revaluation of live performance

In just one year, music generated with artificial intelligence has gone from sounding like a tin can to sneaking onto the Billboard charts. The leap, according to several participants, is not a subjective impression: models that work directly with audio — not with abstract scores — and very high-quality sample libraries have closed the gap that separated a generated demo from a studio production. What until recently was a toy for the curious now covers an 80s classic in a soul style, imitates the voice of a deceased artist or invoices an entire song with lyrics and chorus. Without a corporate message or an intermediary in between.

The technical leap: from tin can to symphony orchestra

The difference from the first batches is the technology layer. It's no longer about basic melodies or repetitive beats: as one participant explains, current neural models handle breaths, vibratos, accents and even the friction of bows on strings. Anyone who listens to one of these pieces without knowing its origin cannot distinguish it from a commercial recording.

In the last month things have exploded. YouTube has filled with covers of 80s and 90s hits with extreme twists: metal turned into soul, rap taken to jazz, voices that never sang certain repertoires. There are songs, the most enthusiastic claim, that if promoted as a commercial product would be authentic hits without anyone suspecting a thing. The fine material, with links to specific pieces and the adjustments of each version, is what turns this matter into something more than a curiosity.

Not everything is euphoria. The latest update of one of the most used tools, v6, has disappointed: it sounds more canned than v4.5 and the instruments cover the voices, according to early testers. Pogre is not linear and versions step on each other.

The Xania Monet case: two tracks on Billboard with a real creator behind

The episode that has turned the narrative around is that of Xania Monet. As detailed in the thread, behind the alter ego there is a real creator — a designer by profession — who provides the vision, the lyrics and the narrative, and uses AI as an instrument to generate the music and the voice. The result: two tracks placed on Billboard charts. That's no small antiestéticat.

The case dismantles two ideas at once. First, that AI replaces the creator. Here there is a person with a story and criteria behind the product. Second, that the public rejects the synthetic just because it is synthetic. The interview in which the author reveals the process is the piece that turns the phenomenon into a case study and not just another musical deepfake.

Universal and Sony buy the threat

The major record labels have trinc the usual manual. First ridicule the technology, then try to regulate it or sink it with lawsuits and, when they verify that it is unstoppable, buy it and impose their conditions. Suno and Udio, two of the main music generation tools, are already, according to one participant, in the hands of Universal and Sony. The scare was real.

The problem comes later. Whoever generates music with these platforms and accepts their terms cedes rights over what is created, with contracts that this same participant openly describes as leonine. The way out suggested is to use the tool to test and try out, and then take the demo to a program of your own. The unknown is what will happen when tools from other countries allow creating and even collecting royalties from copyright outside the big labels.

Live performance as the last refuge

Here the debate splits. One block maintains that AI clones without soul, that it sounds like a soda can and that it cannot go on stage to convey anything. And the response it receives is devastating: 99% of current music is also tracks and arrangements that are cloned without difficulty. That is, AI does not destroy live music; it revalues it by contrast.

The thesis has potential. After the parenthesis of the record industry, music is slowly returning to what it was before: the concert, the here and now, the only format that cannot be generated in thirty seconds. Hopefully, they say, live performances will return and television, radio and the internet will be left for canned music.

Does democratizing music miccionan making it mediocre?

The most serious reproach is not technical, it is cultural. The democratization of any discipline, they argue, ends up being the mediocritization of the whole: photography did not improve when developing became popular, nor did cinema with digital effects, nor books when anyone can publish. Music will not be different. There will be good music and a tide of junk impossible to filter.

The other side sees just the opposite: the end of the artisanal era of cultural creation opens an uncontrolled explosion of the big media. Artists, they warn, will end up like shoemakers when industrial production arrived. It is too early to know who is right. The most critical already warn that YouTube is filling with compilations and channels that publish without criteria, and that finding something with quality will be an odyssey.

From record to video: the next leap

The question that hovers over everything is how long it takes for the phenomenon to jump format. If with music it has been this fast, the majority bet is that movies and series are next. Small series with AI are already being made, given a theme, some guidelines and an aesthetic, and the director corrects scenes in real time. The implicit promise is the same as always: create without intermediaries imposing the message.

The forecast stays there, as a forecast. The only verifiable thing is that the gap between professional and generated has closed before anyone agreed on what to call it. And that the link that has not yet broken — the stage, the body, the sweat — is precisely the one no one has managed to clone.