AI-powered software is making its way into the music industry, sparking debates. Applications like Suno and Udio can generate music based on user prompts, enabling an innovative approach to creating and experiencing music. However, some criticize these AI-created tracks for sounding too similar to the voices and styles of famous artists. But what are the impacts of AI tools for artists and music creators? To discuss how AI is redefining the music landscape, I spoke to Matheus Braz, Beyoncé’s recording engineer.
“I feel like the use of AI in the music industry is at a starting phase,” says Braz, explaining that not much has changed in his routine since the boom of AI tools. “We see these apps generating music, but it still doesn’t represent a risk to established artists due to the level of proficiency in which we work.”
Over the last 6 years, Braz has collaborated with global superstars. His recent jobs include Beyonce’s Renaissance and Cowboy Carter albums as well as Childish Gambino’s Bando Stone and the New World. For him, there is something very unique and specific in the creation of songs, “making it difficult for tech companies to develop software able to automate all these steps involved in professional music creation.”
Nearly 60% of independent artists have been deploying AI in their music projects. On the other hand, only 25% of professional music producers are now using automated tools in their craft. They have major concerns about maintaining a human touch in their creative processes.
Optimizing time
“We never use AI for creative purposes, but there are some types of software that can optimize small editing tasks, especially to tune up the quality of sounds.” Part of Braz’s job is to assist artists during recording sessions, preparing the studio, positioning microphones, and setting up the equipment to capture and manipulate sounds. Further, he is responsible for editing, mixing, and mastering all the recorded tracks.
“When we record in a place with background noise like air conditioning, I usually use Waves or Izotope. This software can automatically identify and remove dissonant sounds, making recordings cleaner and more polished,” he explains. “It’s like having a variety of hammers available, and you have to choose which one works better for each circumstance,” he jokes.
When brainstorming ideas, Braz explains that there are moments when artists want to try samples from different songs, and AI can assist in this process. By using software called Lalal.AI, it is easier to separate the vocals and instruments of audio files. “We gain some speed in the studio, so we get to experiment with more sounds. It’s just a matter of simplifying these tryouts. After we see potential in a certain sample, I contact the specific artist, asking for the raw version of the song.”
A challenging scene for independent artists
Braz sees two ways to look at how generative tools affect small and independent artists. From a positive perspective, AI can assist in creativity, “especially young artists who are uncertain about their sound and are still sonically exploring paths they can follow.” As a means of inspiration, after writing the lyrics, “these artists can use AI to generate several versions of a song, with different beats and genres, so they can understand what works best.”
As a drawback, Braz explains that some people are exploiting generative AI to boost their gains, “creating lots of generic songs, without any effort, and uploading them to streaming platforms.” By doing so, more competition is created, making it challenging and unfair for “small artists that really dedicate their time to compose, mix, and master their songs.”
AI Music and potential concerns
Recently, a US musician was accused of using AI and bots to stream billions of his AI-generated songs, earning millions in royalties. According to Braz, streaming platforms should have a better system to identify and categorize AI music. In this way, people can distinguish what they are listening to. “There’s no problem in enjoying AI-generated music, but there should be tags indicating whether a song is made by AI.”
Further, he emphasizes the need for more regulations over generative AI music apps. Just like any algorithmic-driven tool, these music generators have been trained with a vast amount of data. However, companies behind their creation do not disclose how this data is gathered and used, raising suspicions about copyright infringements.
“We know that these AI music generators have used songs to train their algorithms without paying for copyrights. So they are profiting while taking advantage of other artists,” he says.
Feelings towards AI
Some people fear that AI can replace their jobs. For now, Braz does not see any potential risks to his own career: “I’ve had conversations with some colleagues and nobody is afraid of a possible layoff. Maybe in 10 years, who knows.”