Norges musikkhøgskole Norwegian Academy of Music
No En
Search

Surprising journey in the service of improvisation

The project group behind Goodbye Intuition is constructing a machine that can improvise musically, but the objective is human to the highest degree: developing as an improvising and listening performer of music.

This goal has united guitarist Ivar Grydeland, pianists Andrea Neumann and Morten Qvenild, and singer Sidsel Endresen. The project saw the light of day in 2017. Until 2020 they will feed a machine – or a computer program, actually – with their own sounds. Afterwards they will listen, reflect, play with and against the machine, and become wiser musicians through the project Goodbye Intuition.

“The computer is still in the process of being developed. As of now it can decide itself, within the bounds of reason, when it wants to record what we’re doing, and how much it wants to record, and then play it back as a mirror reflection of what we’ve done. It decides itself when the mirror is turned in our direction to play our own music back to us. In simplified terms, this is what it’s doing now,” Ivar Grydeland explains.

It is Notam, a centre for technology in music and the arts, that is behind the technological development of the machine used by the project group, while the musicians themselves provide all the musical input. So far it is still at the starting gate. Slowly but surely the machine is building up an extensive sound repertoire. The hope is that this will enable the possibilities to keep expanding.

“The idea is that we should be able to give it qualities that will gradually produce a more advanced and sophisticated reflection. Right now the machine is rather basic,” says project coordinator Grydeland.

Human machinery

Each of the musicians feeds the machine individually, but the sounds it emits might have come from one of the other members of the project’s quartet.

Is it the desire to challenge yourselves with something unpredictable that is the driving force for you here?

“Yes,” Sidsel Endresen confirms, while Grydeland hurries to clarify: “But not in the sense that we regard ourselves as so sophisticated that we need something even more difficult than people to challenge ourselves with!”

“Absolutely not. Our only goal is self-improvement.”

“What we’re actually doing is exploring how we, as improvising performers, listen and make choices. In daily life we can assess this with other musicians, and we have a lot of experience in that area. This is why it’s interesting to see how listening and making choices become a challenge when we communicate with a computer that thinks and acts in an entirely different way,” Grydeland says.

Endresen nods affirmatively. “It’s an artistic project.”

“Yes, and that’s one of the keys. In the phase we’re in now, we’re talking a lot about the machine we intend to develop, but it’s really more about ourselves. We’re challenging our own abilities to listen and work through our collaboration with technology,” Grydeland confirms.

Do you have any idea as to where this will end?

“The way I am experiencing the project now, that is actually the question we want to answer. The process of developing the machine, what kinds of algorithms are put into it, and our wishes as performing, improvising people will all define what Notam gets the machine to do,” explains Endresen.

Machine aesthetics as a challenge

What do you mean by the title “Goodbye Intuition”?

“My first impression was that the machine would stimulate intuitive work on the part of the improvising musician, because machines are governed by different parameters than people are,” says Sidsel Endresen. She continues, “After all, intuition is really recognition and memory, and that covers a lot of what applies to an improvised interaction between people. The machine represents something else. It makes its own choices, and has its own music-making behaviour. At the beginning it was about the machine not having its own taste, its own agenda, its own virtuosity to display.”

Grydeland elaborates on Endresen’s train of thought.

“Many of the choices made by people who play music with other people are based on attributes that the machine doesn’t have. It doesn’t have feelings. It’s all about getting away from – or challenging – our way of playing. And letting the habits – both good and bad – connected with our own playing be challenged by a machine rather than by a fellow musician; it becomes an opponent,” says Grydeland.

The two musicians reflect on their answers while they are responding. That is typical of the way they work together in the project group. They acknowledge that the entire project is a process of reflection. Nobody really knows the goal, or the road leading to it.

A complex machine

“Are we devising a mechanical replica of a human being? Does it recognise you – how you play? One of the questions is how we should program it in relation to what we want it to do – how the machine will be to play together with.”

The singer has already discovered some weaknesses, and is interested to see whether they will persist.

“It’s impossible to predict what it will give back to us, but the way it reacts in the course of a short period of time, around five to ten minutes, has already become somewhat predictable. That was unexpected.”

“Yes, it’s surprising how quickly it creates patterns. The values the machine selects as of now are so complex that it’s hard to believe. But it seems predictable. We grasp the patterns quickly,” says Grydeland.

Is that disappointing?

“No, it’s more boring,” Endresen and Grydeland say nearly simultaneously, and Grydeland continues enthusiastically: “We think it’s not complex or sophisticated enough, and that we need to add more properties and possibilities in order to make it less predictable. But then we would be designing…”

Endresen finishes his sentence: “…a machine that is like us.”

“That’s right, but that’s exactly what we don’t want. That is the real challenge. It’s easy to want an idealised human fellow musician, and that’s often what we miss when we’re working with the machine: that it can do what a fellow musician of flesh and blood would be able to do,” says Grydeland.

But why build a machine, then?

“Yes, that’s what we have to use to navigate by. The goal is to create a machine that is so complex and self-willed that by the end of the project it won’t remind us of a human being at all. With a clearly mechanical aesthetic. A machine that makes choices that perhaps we wouldn’t make,” Grydeland explains.

What kinds of discussions do you have in the project group?

“Everything from why we play with a computer – can’t we just play with people? Why are we doing this? Is this a good idea? Luckily we have very different kinds of experiences with the purely technical subtleties, ranging from the techniques that nobody understands to those everybody understands. We discuss quite a bit whether playing music with machines holds any value at all, and I think we take a healthily sceptical view of that question. I think that’s a reasonable starting point for the kind of investigation this project entails. But we also discuss how the machine can develop itself, how it can create its own playing style, and how our interaction with the machine changes how we ourselves play and listen,” says Grydeland.

Satisfyingly unpredictable

Endresen wants to challenge her own logic as to how she works and organises things.

Is breaking away from your mindset a goal?

“Yes, and from expectations and viewpoints about what is a satisfactory form. I think that a lot of things can be moved around.”

“For my part, I want the machine to help me create music that is different to what I create without it,” Grydeland adds.

Does getting better imply less predictability?

“Nooo….”, Grydeland thinks about it. “But maybe predictability is more satisfying to me. It’s a question of the level. In a place like this, the Academy, we know a lot about music even before it’s performed. There are predictable, but certainly also unpredictable, performances even of music that’s hundreds of years old. It must have some sort of freshness, whether we are talking about performing old music or music that arises in the moment,” the guitarist emphasises.

“I don’t believe that unpredictability is automatically better. I think very egotistically. I believe that if I find something interesting, the public will, too. In any case when we work to such a high degree with music that is created in the moment, it can be a problem if it is too predictable.”

On a walk

The atmosphere is very intense just an hour before the first laboratory concert is due to begin in the Levin Concert Hall at the Norwegian Academy of Music. Even the most experienced technician must have lost track of the number of cables twisting all over the floor. Not everything is working as it should, and the planned lunch break between rigging, the sound check and the concert has been reduced to a case of multitasking, with each musician trying to have a bite of something while sitting at his or her instrument.

What does the musical dialogue consist of for you when you’re improvising on stage?

“If we’re talking about music with a high degree of improvisation, then the surprise factor is at one level. If it’s music where a lot is determined in advance, then it’s something entirely different. Maybe a microscopic level where the elements you’re very familiar with fuse together in yet another new way. That even if it was a tiny bit different yesterday, it is different enough today to be able to surprise you, to have a freshness about it. This is perhaps the nearest I can get to a common goal if I view all of the music I’m involved with as a whole,” says Grydeland.

For Sidsel Endresen, it has to do with a kind of freedom.

“The ideal for me is someone who doesn’t stop me, who I can go for a walk with, who understands me. The walk might take us in an unforeseen direction, and I can be surprised in a good way. It’s all about choosing to play music with people who open you up rather than trying to block you,” the singer says.

What do people need to do in order to manage that among themselves?

“Quantity training can help. At the same time, one-off encounters can be surprisingly productive. But it all boils down to a large amount of mutual trust and listening,” says Endresen, and Grydeland nods in agreement.

The Bonus

But the machine does not represent a musical partner who encourages trust and interaction. On the contrary, it can be rebellious. And in the Levin Concert Hall it decides to be temperamental precisely when Morten Qvenild is going to demonstrate it to the audience. Restart! Ivar Grydeland tries to make the seconds and minutes pass.

“That’s what we call chit-chat,” he announces, before the machine starts acting normally again.

“We can’t hide behind the research label. We want it to work, of course. But that’s part of what is exciting: that we can allow ourselves to do things slightly differently than we would in a concert that was completely public. I hope and believe that this can generate music that is a little different.”

One thing is a laboratory concert, but can you imagine going out onto the stage in four or five years with an unpredictable machine that you’ll be improvising with?

“It wouldn’t be unthinkable for me,” Endresen responds quickly.

“It seems impossible with the creature we have at the moment,” says Grydeland, laughing. “But it would be very satisfying if this resulted in a work that we could actually use. All the same, I don’t want that to be our goal, because then that will govern the choices we make,” he says.

Sidsel Endresen admits that these somewhat unclear objectives might make the project difficult to understand for outsiders.

“But it’s a voyage of discovery. We want to displace our own musical strategies in our work,” Endresen ascertains.

“We do have a clearly defined goal of learning something about our own way of listening,” Grydeland adds, being more explicit. “That’s our main objective. And it would be a fantastic by-product if the machine could still be used after the project is finished,” he concludes.

Articles relevant