L’Artificial intelligence it is now an integral part of various sectors, even in more artistic fields such as music.
It is precisely in this context that it is moving Google DeepMind which recently announced, in collaboration with YouTubethe project Lyria. We are talking about an AI-based music generation model.
In this context, in addition to Dream Track, other tools capable of generating music were also presented, designed with the support of artists, songwriters and producers to create an effective and collaborative system. But how does Lyria work and how will the relationship between musicians and AI change?
Analyzing music from an informational point of view, we can talk about data such as note, beats e vocal harmony. The combination of these factors generates a sequence of sounds, appreciable to the human ear but difficult to decipher (at least so far) by AI models.
If chatbots have to deal with a text prompt, for example, in this case the AI finds itself dealing with multiple instruments and voices at the same time, which must be divided and understood individually. And this is where Lyria comes in, capable of combining instruments and voices adapting it to certain musical styles.
Google DeepMind is working on Lyria to find the perfect balance between artists and AI
Dream Track falls into this context, offering users the possibility of generating, via classic prompts, musical sequences with lyrics and the style of a specific artist.
To balance this use, it must be said, Google is studying a method for the artists involved to obtain a percentage of profit on the views obtained from YouTube videos that “exploit” their vocal style. Lyria also takes advantage of the project Music AI Incubator of YouTube, which aims precisely to find a balance between AI and artists.
The collaboration that arises between developers and those who make a living from music is also leading the AI model to have greater capabilities regarding the creation of music from scratch instrumental pieces. The results obtained so far by Lyria are encouraging and, like many other platforms in the sector, with the data it assimilates in the coming years it will only improve exponentially.