AI in Music Production: Dahlbäck’s Inspiring Instrument

John Dahlbäck describes using AI in music production primarily as a source of inspiration and treats AI vocal tools as an instrument similar to Splice or a synthesiser. He says he can feed a beat to AI to generate ideas on how to develop it, and that he approaches AI the way he has approached sampling, by re-creating elements rather than simply including them.

He states that he is a firm believer in voice-changing AI tools, saying that he can sing into a microphone with a poor singing voice and change that into a beautiful singing voice. He expresses excitement about AI when it is used creatively rather than lazily, and he opposes casual use that only produces a pop song for upload to Spotify.

He also mentions plans to train AI on recordings of a Swedish singer from age five so that she can duet with herself.

John Dahlbäck says he turns to AI in music production mainly for inspiration and treats AI vocal tools as an instrument comparable to Splice or a synthesiser. He describes using AI in a way similar to tweaking a synth until it leads to a result, and he reports that he can feed a beat into AI so the system gives him ideas for where to develop that beat. He indicates that this workflow is part of his creative process and that he uses AI outputs as prompts for further development rather than as finished material.

He reports a history of sampling in which he would re-create samples rather than simply include them, and he applies the same re-creative approach when using AI. He states he uses AI to generate stems and backing elements that can be integrated into productions rather than copied wholesale. He says he is a firm believer in voice-changing AI tools and that he uses them to transform rough vocal takes into improved vocal performances. He opposes casual use that produces a pop song which is then exported and uploaded to Spotify.

John Dahlbäck says he is a firm believer in voice-changing tools with AI and states, “I sing something in the microphone with my horrible singing voice, and I change that into a beautiful singing voice.” He paraphrases this use as transforming rough vocal takes into improved performances, describing the tool as capable of taking a poor singing input and producing a polished vocal result.

He presents this capability as part of his practical workflow, using his own recorded vocal takes as material for the tool to alter. He does not offer technical details about the process in the quoted material.

He says, “I’m very excited by AI if the person using it is also creative and not just lazy,” and he criticises casual use that simply generates a pop song for upload.

He recounts an example in which an acapella is loaded into AI to obtain stems and backing vocals, and he describes the result as making the vocals more full. He calls that example “a brilliant use of it” and distinguishes that creative application from what he describes as sloppy use. He does not provide additional specifics about the AI systems or settings used in these examples.

John Dahlbäck says his brother is an old-school music producer who records bands in a traditional way. He says he was scared of AI in music in the past but learned from his brother and now sees value in AI for backing vocals using AI-generated stems. He reports an example where “he just loads in her acapella, and he gets great stems and just backing vocals by using AI,” and he says the result made the vocals more full. He calls that use “a brilliant use of it,” and he presents the generation of stems and backing elements as a way to enhance vocals rather than replace them.

He recently spoke about a Swedish singer whose tapes from age five could be used to train AI to allow her to duet with herself. He says, “Plans are now in motion to train AI on her singing voice at that age, ‘so that she can duet herself’.” He describes that planned training as a practical application of voice-focused AI and expresses approval of the concept. He also says, “It’s kind of incredible… I’m all for that,” and he contrasts creative applications with what he calls sloppy uses that simply generate a pop song for upload to Spotify.

John Dahlbäck says, “I’m very excited by AI if the person using it is also creative and not just lazy.”

He recounts an example in which “he just loads in her acapella, and he gets great stems and just backing vocals by using AI; it made the vocals more full,” and he calls that “a brilliant use of it.”

He also states, “It’s kind of incredible… I’m all for that, and like I said, I don’t like when it’s just anyone sloppy using AI to just [generate] a pop song, then export and then upload Spotify. That doesn’t make any sense.”

John Dahlbäck expresses a cautiously optimistic view on AI in music production, saying “I tend to turn to AI for inspiration mainly. I use it more of an instrument – just as I would Splice or a synthesiser.” He says he can feed a beat to AI so it gives him ideas of where to develop it, stating, “I can feed [it] that beat that I’m doing, and it gives me ideas of where to develop it.”

He says he is a firm believer in voice-changing AI tools and reports that he can sing into a microphone with a poor singing voice and change that into a beautiful singing voice, saying, “I sing something in the microphone with my horrible singing voice, and I change that into a beautiful singing voice.”

He says he is excited by AI when it is used creatively and opposes casual use that simply produces a pop song for upload, stating, “I’m very excited by AI if the person using it is also creative and not just lazy,” and, “I don’t like when it’s just anyone sloppy using AI to just [generate] a pop song, then export and then upload Spotify. That doesn’t make any sense.”

He also praises an example where AI-generated stems and backing vocals made the vocals more full and mentions plans to train AI on tapes of a Swedish singer from age five so she can duet with herself, saying, “So, he just loads in her acapella, and he gets great stems and just backing vocals by using AI. It made the vocals more full. I just thought that was a brilliant use of it,” and, “Plans are now in motion to train AI on her singing voice at that age, ‘so that she can duet herself’.”