Google has developed a new artificial intelligence (AI) system that can compose music based on text descriptions.
According to Neowin , dubbed MusicLM, Google researchers described in their academic report that this is a model that generates high-fidelity music from text descriptions, such as “mellow violin tone supported by guitar sounds”.
“We demonstrate that MusicLM can be tuned on both text and melody in that it can transform whistling and humming tunes in the style described in text captions,” the paper says. “.
According to the research paper, MusicLM can take cues from user-generated descriptions such as “enchanting jazz song with a saxophone solo and a solo singer” or “90s Berlin engineering with low bass and strong kick”… to get the corresponding results.
The emergence of MusicLM was launched by Google after the rapid development of ChatGPT by OpenAI – an AI-powered natural language processing engine. ChatGPT is so popular that Google’s management is said to have declared “code red” to accelerate its AI projects. Google is expected to introduce more than 20 AI-related projects this year, including an AI-powered project for Google Search.
– based music maker AI MusicLM isn’t the first . Previously, several systems with similar themes were created such as Riffusion, Dance Diffusion, Google’s AudioLM and OpenAI’s Jukebox.
People shouldn’t expect MusicLM to go public in the near future because of concerns about programming biases that could lead to technology glitches, lack of representation, and conflict, according to researchers at Google. with creative content copyright. In fact, in one test, Google researchers found that about 1% of the music MusicLM generates was copied directly from the songs it was trained on.