OpenAI, a non-profit research company based in San Francisco, has developed a new artificial intelligence (AI) called Jukebox that can compose music in various styles and genres. The neural network can create both instrumental and vocal music, with only a style or artist as a reference point. Additionally, Jukebox can even complete unfinished songs. The AI was trained using approximately 1.2 million music pieces, including their lyrics and metadata such as artist names, genre, mood, and release date. The developers of Jukebox claim that the software has made significant progress in terms of musical quality, coherence, length of the sample, and the ability to orient itself to artists, genres, and lyrics.
Despite the impressive results, the process of creating a two-minute song still requires approximately 18 hours of computing time. Furthermore, Jukebox still has limitations compared to human composers, as it lacks the ability to create musical structures such as choruses. However, the developers believe that Jukebox has the potential to bring benefits to society and democratize access to music creation, rather than being limited to profit-oriented companies. OpenAI is funded by companies such as Microsoft and Tesla CEO Elon Musk.
Jukebox’s ability to create music in various genres and styles is a significant breakthrough in the field of AI. The software’s potential to democratize music creation and provide access to a wider audience is a promising development. However, the limitations of Jukebox compared to human composers highlight the importance of human creativity and the need for a balance between technology and art. As AI continues to advance, it will be interesting to see how it can be used to enhance human creativity and bring benefits to society.
