The rapid rise of ChatGPT has many wondering what else artificial intelligence might change. A Google AI research paper published this week will change how you make music.
The paper describes a tool called MusicLM that “can transform whistling and humming melodies according to the style described in the text captions.” It can also generate “high-fidelity music from textual descriptions such as ‘a soothing violin melody backed by a distorted guitar riff’.”
The paper’s website provides an example of the results produced by the tool. In one example, someone sings “Bella Ciao,” his late 19th-century Italian folk song. Based on that hum, the tool then generates music in various styles of instrumentation, including guitar solos, string quartets, and jazz with saxophones.
The Twitter account @bleedingedgeai posted one example.
“Think of MusicLM as the ChatGPT of music.” murmured Entrepreneur Martin Uetz adds, “I can’t wait for this to go mainstream.”
Musicians who have spent decades learning an instrument may not be so enthusiastic in the same way that illustrators and graphic artists are irritated by AI tools that create striking images from mere text prompts. Hmm.
These AI art tools include Midjourney, Stable Diffusion, and DALL-E 2. Impressed, he shared his experience with the tool on his social media, but was surprised by the backlash from the illustrator. And last year, a Midjourney-generated image won an award at an arts festival, pissing artists off.
The problem artists have with such tools is training themselves through vast collections of digitized artwork without their consent. A recent lawsuit filed in San Francisco by a practicing artist described Stable Diffusion and Midjourney as “a collage tool that violates the rights of millions of artists.”
In fact, due to copyright issues, Google AI cannot publish MusicLM. However, start-ups may be more willing to bring such technology into the world.
Big tech isn’t devoting resources to generative AI either
DALL-E is powered by ChatGPT maker OpenAI. Microsoft has invested billions in OpenAI and plans to use the technology in various products, including the Bing search engine. It has now caught fire under Google’s parent company, Alphabet, which is working on a similar tool to answer this challenge.
MusicLM as a tool is far from perfect, but it gives an indication of where things are going. The same goes for ChatGPT itself. As Mark Cuban recently said about AI chatbots, the billionaire said, “Imagine what GPT 10 would be like.”
The Trust Factor is a weekly newsletter that examines what leaders need to succeed. Sign up here.