AI Creates Traditional Bagpipe Music, Adds Animation

AI is now composing bagpipe music. 

The development is sparking debate among musicians and traditionalists about the future of one of the world’s oldest instruments.

Scott Haynes is a former piper, having performed in bands including the Boghall & Bathgate Caledonia and the 78th Fraser Highlanders. Now semi-retired from his piping days, Haynes works in AI for Neural Frames, a startup developing an AI-powered animation generator.

Intrigued by the potential intersection of his past and present careers, Haynes sought to see if AI could create bagpipe music. He turned to the text-to-music platform Udio to generate piping music.

To begin the experiment, Haynes recorded himself playing a simple jig on a practice chanter, then uploaded the file to Udio along with the following prompt: "Breton Celtic folk music, Rock, Folk rock, Celtic, European music, Regional music, European folk music, Instrumental."

To complete the project, Haynes used the Neural Frames platform to generate a short animation to accompany the music.

“The result was incredible,” Haynes told AI Business. “Although not every generation turned out perfectly, there's always an element of unpredictability with AI creations.”

However, not everyone shares Haynes' views. The piper reported that the reaction to his work was split down the middle.

Related:Will.i.am Launches Interactive AI-Driven Radio App

“Most people thought it was pretty cool, some people thought it was kind of silly. Sometimes some people just hate AI and they’ll just start attacking that.”

A screenshot of Haynes creating the animation sequence for his piping music | Credit: Scott Haynes

Bagpipes Beyond

Contrary to popular belief, bagpipes are not just the stereotypical Scottish instrument. Cassandre Balosso-Bardin, founding director of the International Bagpipe Organisation (IBO), said there are more than 130 different kinds of bagpipes around the world, including Eastern Europe's Dudy and the German Dudelsack.

The IBO, founded in 2012, organizes a biennial conference bringing together musicians, instrument makers and scholars. Additionally, they promote International Bagpipe Day on March 10 each year, encouraging people worldwide to celebrate and play bagpipe music.

Speaking to AI Business about Hayne's AI piping experiment, Balosso-Bardin highlighted that it raises wider issues of bias and representation in AI-generated content.

"There is a problem with AI-generated stuff in general, just like there is with a lot of stuff that comes out of Silicon Valley, is that everything is biased,” Balosso-Bardin said. “Everything being created by a specific type of people who are creating it with a specific type of audience in mind and it's basically them.”

Related:Video Game Voice Actors Strike Over AI Concerns

Balosso-Bardin said she did her own recent AI experiments, trying to get image generation platforms to create content featuring bagpipes that were not associated with the traditional Scottish stereotypes. The results were not effective, however, returning images that featured tartan patterns and four-armed people with bagpipes that do not exist.

“You've got these inaccuracies because the system hasn't learned, or doesn't know, that there is so much diversity and so much richness in our actual real world that AI isn't able to reproduce and it's completely wrong,” Balosso-Bardin said. “This is what my organization is trying to undo and level the playing field for all other kinds of bagpipes.”

Udio's Legal Challenges

The platform Haynes used to generate the AI piping music has recently come under scrutiny regarding the music used to train its underlying model.

The app is powered by the Udio v1 mode, though it has not revealed the data used to train the models.

Some record labels have said Udio used their tracks without permission. The platform was hit with a lawsuit from the Recording Industry Association of America (RIAA) on behalf of members, alleging it appropriated thousands of songs without permission to train its music generation model.

Haynes said he wished platforms like Udio would be more transparent.

“Being from the traditional music side of things with bagpiping, it's less of a concern because it is traditional music, so almost all of it is public domain. But on Udio I typed in a few DJ names like Deadmaus and it said, ‘Hey, you can’t do that,’ but then replaced it with other keywords and the result was surprisingly close to the actual artist.”

The slow process of enforcing copyrights through the courts means music artists require faster means to protect their content. In Tennessee, for example, a new local law called the Elvis Act offers improved publicity rights to songwriters and performers over unauthorized use of their content but measures like these are minute in the face of the wave of generative content that's beginning to emerge.

“Like most AI, there is an ongoing concern of how Udio and Suno were trained and whose original recorded work was used to train them," said HP Newquist, who founded the National Guitar Museum and now serves as executive director of AI consulting firm The Relayer Group. “This is something that needs to be guarded against and legislation to protect musicians from having their work used for AI training is long overdue.”

The Future of Bagpipe Music

AI will never replace the iconic sounds of the bagpipes or other instruments, as it will never truly be able to replicate the feeling of having sound resonate through your body from an instrument.

But in the wake of his AI experiments, Haynes told AI Business that he could see more bagpipers using AI.

“A lot of the pipe bands I was involved with do professional concerts, which are two hours long and have all these sets and arrangements. How cool would it be if the band was playing something and then all of a sudden, a whole orchestra came in? It would just all be AI, you can do a whole bunch of creative things musically, which ultimately is what I'm interested in, using AI to enhance the music that we already have.”

AI-generated bagpipe music could offer increased accessibility and efficiency, according to Mick Kiely, CEO of music generation platform IAIAI Technologies, enabling musicians to experiment with new musical styles and compositions.

“Given that bagpipe music is not in high demand and accessing the ability to record real bagpipes can be challenging, AI generation could help boost its popularity, albeit likely in a more contemporary style rather than the traditional sound,” Kiely said.

Balosso-Bardin of the ISO said AI will likely be used as a tool by some musicians but was cautious about it being used in traditional music.

“I don't want to be negative, just wary,” she said. “It's great for people to have fun... but I'm just wary of what comes out and how that affects the people who are at the source of this music.”

Haynes' experiment with AI bagpiping was just that, an experiment. While it showcased the potential of AI in music generation, it also highlights the complexities and concerns surrounding this technology in traditional music spheres.

The concept of AI-generated bagpipe music is in itself a microcosm of the wider conversations going on in creative industries about AI: Rights ownership, the creative nuances of the individual artist and the preservation of cultural diversity.

As Balosso-Bardin notes, "I still think that humans have a great role to play in music, especially in traditional music. I would personally not want to go to anything that's AI-generated unless it's kind of backing, something that's already there, that's kind of traditional with the musical elements, that's there for the human element."

While AI may find a place as a tool for musicians, much like it has for marketers and business leaders, the human touch remains irreplaceable in capturing the nuances, cultural significance and emotional depth of traditional music. Traditional music like piping has endured for thousands of years and will most certainly endure beyond AI, for now.