Artificial intelligence and music: What influence AI will take on the music in the future - and now takes

The exploring the possibilities and dangers of artificial intelligence is becoming increasingly important, there are no two opinions about that. The mere fact that Donald Trump, shortly after his renewed introduction, into the office of the US President, the start of the 500 billion expensive AI predictStargatehas postulated, gives a good impression of how important the topic will be in the future. Trump will probably not want to use AI to make music (as is well known, nasty people have no songs, as Johann Gottfried Seume once wrote), but it will beArtificial intelligencealsoin musicplay an ever greater role.

In order to understand in which areas of music artificial intelligence can play a role in the future, you first have to understand what AI can do. Basically, artificial intelligence can be divided into four different areas, the first type a rather weak one, the fourth is a stronger artificial intelligence - and can even be stronger in the future than the human brain.

It is important to know that so far there is only the first two AI technologies. But type 3 and 4, research is certain, will come.

This type can be understood as a basic type of artificial intelligence. It is programmed to perform certain tasks, but has neither memory nor an understanding of the context. An example of this would be the chess computer, for example. This can calculate the best possible train using AI algorithms-but nothing more.

The second type has limited storage capacity and is therefore not limited to a single task. However, one should not be fooled by the misleading Wording: This AI can process large amounts of data and develop procedures based on this, that is, in contrast to Type 1, type 2 can use past experiences via machine learning to make better decisions. An example of this are smartphone language assistants like Siri. Such assistants can research data from the Internet, but also manage other apps or organize the calendar. Another example is self -driving cars that can master the basic traffic rules, but can also react to unexpected situations. AI-supported applications such as Chatgpt also belong to Type 2.

While artificial intelligences of Type 1 can only react and do not plan ahead, artificial intelligences of Type 2 are able to develop strategies.

This category of artificial intelligence, which has not yet existed, should not only be able to process information (as type 2), but can also independently get an idea of ​​the world, that is: develop an almost human -like understanding, knowledge and awareness.

Such artificial intelligence would be at least equal to that of man, would have a comprehensive understanding of itself and could develop independently and grow beyond themselves-so it would be what is known from various utopian science fiction films. If you would miss this AI a human appearance, it would hardly be distinguished from real people.

So now is the question of how and in what form artificial intelligence influences the production, consumption and usability of music. And, as is so often the case with such complex topics, there are of course different facets. We present the most important to you here.

AI is already used regularly in music production. There are various AI programs such as Suno, Udio or Amper that are already successfully used for various areas of music production: for the analysis of musical features such as melody and rhythm, for creating mashups, for creating entire pieces of music-the possibilities are varied. Singing such programs can also generate what producers can help if they compose a song for singers in XY. Then with the help of artificial intelligence, this: With the help of artificial intelligence, the vocal line with the voice of the later interpreter can be composed to show how the song sounds roughly with its/his voice. This is a very helpful feature that is already used regularly in music production.

But there are also other prominent examples: the (remaining)Some time ago with the help of such programs of the Machine Audio Learning John Lennon's voice of old demos to separate withNow And ThenPublishing one last "real" song by the Beatles - without a question a great way to use AI in music production. There is also with NirvanaDrowned in the Suna song that is through artificial intelligence and the use of a-Coversanger has arisen. And it will certainly continue to create songs that were not completed from existing pictures of old, deceased artists. There are already examples in Germany. Last year, for example, a new song by pop singer Udo Jürgens was completed and published:When I went away.

The first AI-generated song that made it into the German charts was in the middle of last yearCrush in a valleyFrom the producer butterbro. He used the music generator website Udio to have melody and singing created for his self-written text. The result: at least 48 in the German single charts, between Beyoncé and Taylor Swift.

And the song is not the only one. For example, the piece is in the NetherlandsSummerchart, who also has a hitandwar.

However, there are not only songs based on AI, but also entire artists. One of them is Ben Gaya, who is nothing more than an avatar created by the German advertising agency. It not only looks good, but also presents itself as if he would write his songs at home on the guitar himself and therefore really brings everything you need as an emerging star on social media. His song Sunshine Soul alone was clicked 80,000 times on Instagram.

The group itself, after all, the most important and most used in the world is open to dealing with AI. Personalized playlists are played, for example, with the help of AI-based machine learning programs-and that since 2018. However, Spotify insists that such recommendations are always based on cultural trends and human flair. And Spotify is proud of it and states that 81% of Spotify users state that personalized music recommendations are their favorite feature on Spotify.

However, Harper's US magazine recently uncovered that there is a systematic AI program from Spotify behind supposed artists in playlists: Perfect Fit Content. For this purpose, Spotify is supposed to work with a network of production companies that use AI cheap to produce tracks, which then hike in Playlists, from which Spotify benefits financially. If you take a look at Playlists like Viral Top 50, you will find a lot of such songs by AI artists like Antoniovival or Melodai.

One of the major topics within the music industry will be one question this year: How is the copyrights about using AI? Because if songs are written with the help of artificial intelligence, they are fed with existing song material - i.e. with songs and texts that all have human authors. So far, however, these have not been remunerated. The legal situation must therefore be clarified.

The GEMA, i.e. the society for musical performance and mechanical reproduction rights, which manages the rights of use from the copyright of over 90,000 members (composer, textipers and music publisher) as well as by almost two million rights holders from all over the world, Has already filed a lawsuit - both against Openai, the company that is also behind Chatgpt, also against Suno. The accusation: The US AI music provider is said to have trained his system on protected recordings of world-famous songs, on the basis of which the generative artificial intelligence then "wrote" new songs-without paying a remuneration. In the United States, too, the large record companies brought together a few months ago and complained together - but so far without any result.

Composers are therefore rightly concerned about their financial future. Because in particular in the area of ​​so -called usage music, i.e. for advertising, series, films, it is quite likely that KI will be used for this in the future - the monetary damage of human musicians.

Such complaints are sample procedures with which basic legal questions are to be clarified. Although it must be underlined that GEMA is not about banning AI services, but about developing a sensible license model. However, not only the national, but international law, are usually affected, which makes persecution, clarification and legal implementation of the whole more than difficult.