Kalvos & Damian’s New Music Bazaar, Show #116, 9 August 1997. Kalvos & Damian on the road in Cologne at the composer’s studio. Listen to the interview from the original broadcast [0:36:03–1:40:10].
Henning Berg was trombonist from 1982–96 in the internationally-renowned WDR Big Band. Additionally he play in a duo with Hendrik Soll and in various Jazz Ensembles and improvised music formations. He is the developer of the interactive improvisation software Tango (1992, Steinberg) and TangoII (in development). He is currently Professor of Jazz trombone at the Hochschule für Musik Köln.
[Kalvos] Our guest today is Henning Berg. Tell us about yourself, tell us the kind of music that you do. I’m here, sitting on the floor in this studio, surrounded by speakers and, there’s a drum kit, a piano, computer, printer, two, what look like two Ataris and a Macintosh, a string of Yamaha equipment, a Roland device… what else have we got up there, Ensoniq…
[Henning Berg] Well, I’m a jazz musician, actually not from Cologne, but I’ve been living here for 20 years, and I play the trombone on a good day… and on the bad days, as well. At this point, I play in the West German Radio Big Band, called the WDR Big Band.
[K] The WDR Big Band, I never thought I would hear “WDR” and “Big Band” in one phase.
[Damian] We haven’t, actually, one has been superimposed over the other.
[HB] Well, it’s a very nice institution. We are doing jazz, and jazz-related stuff exclusively, which is a nice thing for a big band, not to have to play any dancers, stuff like that. We are playing nice festivals with guest conductors, guest arrangers, guest soloists, we’ve been playing the Montreux Festival several years with Quincy Jones, and people like that. This year, again, we are even going to go on tour with Phil Collins in July sometime this year. Yeah, so, it’s a very good band, and I’ll have to say that I’ll leave the band next summer, after 14 years of my playing there. I do that because I want to do smaller projects, on my own, with small groups. Also, my electronic stuff became very important in the last couple of years.
[K] Yeah, let’s hear about that, because that’s what we first heard about, was the electronic stuff.
[HB] Well, years ago, I think it was in ’88, there happened to be an International Computer Music Conference here in Cologne: the ICMC, which is one year in the States, one year in Europe, basically that’s the deal. I used to program music before that, in sequencers, with keyboards and stuff, I made my own tapes, and so on.
[K] How far back does that go, when did you start doing that sort of thing?
[HB] 10 years ago, something like that. So, having heard about this ICMC and not knowing anybody there, I just thought, well, if you’re serious about that kind of stuff, working with electronics, and so on, you should at least go there and check out what they are doing. One of the first people I met on that particular Saturday morning was a man named Clarence Barlow, who was the musical director of that ICMC in ’88. As it happened, in the afternoon, he showed to interested people his work, a program named AUTOBUSK. He played with it, it was the first time in my life that I saw somebody working with a computer in a way that was not quite predictable, where something came out of the machine that wasn’t actually put in the same way.
[K] Now, AUTOBUSK, for our listeners, is best described, I think, as a parametric music generator. You can either analyze, or derive for yourself, a kind of a condition in which you want the music to play, and that can be harmonicity or rhythmicity, or leafiness, of the notes… there are any number of parameters that you can specify so that you can produce music that may be similar to plain chant, that might be similar to Mozart, or Stravinsky, or might be totally original because of the parameters you’ve chosen. I think that description might be inadequate, but it at least describes some of it. Can you expand on it?
[HB] Well, what was most striking to me was the fact that the parameters of the program can actually be remote-controlled by MIDI input. I don’t know if I have to talk about MIDI now? No. So, somebody playing the piano, or a MIDI keyboard or something like that, and the data that would come in from playing a Beethoven piece or something actually controlled parameters of the program that generates music on its own.
So, this program makes completely different use of parameters, say, volume or pitch or something like that. Actually, the ear would do some of the notes, a low note, a loud note, a soft note. So that was the interesting part for me, that parameters were actually bent into something completely different. I felt that this is actually what I’m trying to do, is getting something out of the machine that has to do with what was put in, but not in this predictable way that you would do when you’re sequencing, you know, you just program your tracks, edit them, copy stuff, and it’s like a word processor for music.
And this is a completely different thing.
We listen to Earspace by Henning Berg and John Taylor [0:44:05–0:47:35].
[HB] So, I talked to Clarence later on, and said I would love to have something like that, I’m a trombone player, and I want to be able to really improvise, and isn’t there anything available on the market that you can buy? He said no, that if you want to have something like that, you have to learn how to program yourself. I had no idea how to program a computer, I didn’t know what a programming language whatsoever, so he said, “Well, go to a shop and buy yourself a C-compiler, and just get started. Get yourself a good book, and just get started, problems will come on their own, you can always ask people if you get stuck.” So actually, that was kind of the first initial thing, and the second thing was at the same ICMC, the opening concert was a concert by George Lewis, the American fantastic trombone player and computer programmer.
[K] Right, and we have just played a piece by Neil Rolnick that George Lewis performs, a spectacular piece called Wonderous Love, on a new CD that’s only been out about a year.
[HB] Well, he played up here in Cologne in the TV tower, well? Yeah, the TV tower, big thing there, and he played, all on its own with his trombone, a pitch-to-MIDI device into a Macintosh, where there was a program running. Now this program is called Voyager, I don’t know what it was called then. And, this program was playing MIDI synthesizers, like an improvising duo partner. So those were the two things, talking to Clarence and seeing this performance of George Lewis’s. From then on, Clarence and I hadn’t talked for 3 years, and I just sat down and learned how to program, and to learn to come up with something that I can actually duo with, play duo music with. This program will never be finished, because I am working on it every day since then. But, it has reached a state, since 1990, where I can actually perform with it.
[K] Is this a program that you sell? Or is it a program only for you?
[HB] Well, I got a big software company, German software company, interested in this program, and together we worked at making it available and more interesting for more people than myself. So, now it has a professional user interface and that kind of stuff as well, that it didn’t have in the first place, and it sold all over the place, it’s called Tango.
[K] Tango is yours, okay.
[MB] And, actually, I wish it was being sold all over the place, but it’s… available, let’s put it that way.
[D] You’ve said you wanted to work more with small ensembles, but have you written anything for the [WDR] band, big band?
[HB] No, I haven’t. Not at all, no. I’m not really a person that sits down and writes for ensembles and stuff. I’m more like a player, which also shows in my work with the computer. Concerts with Tango work like, well, I sit here and this computer wreck, is 2 or 3 meters away, I play and the program plays, I listen and the program listens, and we hopefully come up with something decent.
We listen to Confluence by Henning Berg and John Taylor [0:52:42–0:55:20].
[K] Has this changed the way you improvise? I don’t mean simply mechanically, but the way you think about music?
[HB] Yes, yes, a lot. Especially in the last couple of years, I work a lot more with pre-improvised music than I did before, because working with Tango basically is free improvised music, most of the time. You can also enter changes, chord changes, which you improvise on. That’s possible on meters, tempi, or half-tempo, and stuff like that. But, many times, just free improvisation, where you play and listen and at the same time, to get this loop straight between the two guys. Yeah, it has changed my way of thinking about these things. I’d say I’m more conscious about what I expect from an improvising machine, or an improvising musician, or, an improvising myself, in the particular moment. I realize that free improvisation never is completely free. Very often, or most of the time, you have to be very conscious of what you actually doing right now. The situation can be that one of the two partners is kind of dominant, and the other provides background material for him. Say I suddenly start playing very soft in a duet improvisation, that can mean different things. It can mean, “Okay, let’s start a big crescendo together. And, it can also mean, “I want to put you in the foreground now and play back on it. It can also mean, “Let’s just enjoy playing very soft together.” So, how to find out what the other one wants now, by playing very soft?
And, I also realize that these kinds of things are also very important in writing Tango, because many times, when I’m writing the program, I don’t want it to be predictable, on one hand, but on the other hand I want the musical output to be as good as possible, which sometimes is a contradiction in itself. Because, just take this simple example of me playing soft. What do I want the program to do? Do I want it always to take it like, “Okay, I’m on now, it’s my solo,” or do I always want a big crescendo to come out of the situation, “Okay, we’re both soft now and we get louder.” Tango actually has got a little mechanism in there, that evaluates the situation that is happening right now.
[K] In a live performance with live musicians, you sort of have body language and other cues, how do you exchange cues with your program?
[HB] It works on different levels. On one level, Tango memorizes what I’m playing, it analyzes my playing in terms of statistics, average pitches, highest pitches, lowest, longest notes, shortest notes, loudest notes, intervals…
[K] Does it learn you, in a sense, at that time, or does it have a history of you playing?
[HB] No, from that time. This analysis actually controls the parameter that I call TSI, which stands for Time and Surprising Input. This parameter, it’s just a bar on the monitor that goes from left to right. On the very left, it means that in the recent time, there has been very little surprising input from me. Surprising input is input that is different from what happened before, in terms of these values. And, basically it means that very little surprising stuff in a certain amount of time means, “I’m bored,” the program is bored. On the other hand, lots of surprising stuff in a given amount of time means the program is attentive, rather. So, if the program is bored, it will have a tendency to play more spontaneous material, to help me with new ideas, rather than follow me in what I’m doing, which the program is doing when it is very attentive. Recent versions also contain an opinion that Tango doesn’t have, not only about my playing, but also about his own playing. Basically it works with the same parameters, but takes “What am I, Tango, playing now? Is it time for a change, or isn’t it?” And these decisions, when is it time for a change, in the given situation, of course, these decisions have eventually been made by me, Henning, writing the program. So, that’s the hard part: making it come to reasonable decisions without pre-defining too much, because what is interesting to me about the whole thing is the surprising bit.
[K] I guess that’s why I was asking you if it kept any kind of history, because it would seem that what you would consider surprising and not, in some regards, would end up in your programming, so that if someone else were to work with it, it would kind of have your bias, rather than that performer’s bias.
[HB] Certainly, certainly. Even though there are many parameters to shape the program to your own needs. I mean, you can really quite flexibly pre-define what you think is a boring situation, and which parameters are more valid in terms of making it interesting. For instance, a trombone player will put much stress on… hmm, a bad word. He will look at pitches as a very important factor, because the trombone has, in the normal, jazzy kind of way of playing, a pitch-range of say one and a half to about two octaves. That’s basically where you spend about 95% of your time. Most of it in even narrower a pitch range. So, if I’m really at the very top of this range, and move to the very bottom of this range, then that’s something that really makes a hell of a difference, from a trombone player. But, if I’m playing the piano or a MIDI keyboard, or something like that, two octaves is like what I’m doing all the time. Just using the left or right hand, could easily cover two octaves every second, so I will look at pitches in a completely different way. Same with intervals, stuff like that. Playing the two-octave thing, or octave intervals on the trombone, will make it very obvious that this guy now wants to do something with big intervals, but octaves and stuff like that on the piano will not be such a big thing.
[D] Is it called Tango because you are, in a sense, dancing with the other instrument?
[HB] No, it’s not an abbreviation for something. I was looking for a word that has to do with music, and that sounds a bit old-fashioned, can be pronounced in many languages (is Latin-based), and…
[K] And hadn’t already been used by someone else… [Laughter]
[HB] …and has something erotic to it.
[K] Oh, tango, yes…
[D] Can you envision a time where you will have two Tangos, where one improvises off of another, so you actually sit in the audience, and watch these two machines improvise off of one another?
[HB] I’ve never tried that, and I’ve got my doubts if it would really be a very rewarding situation, I don’t know. Because, I get the feeling that it’s good to have at least one human partner, in that loop of playing and listening.
We listen to Lineage by Henning Berg and John Taylor [1:07:07–1:09:37].
[HB] Can I just add something to what you said about the two Tangos? That’s a situation where we can only get one Tango, because I mentioned this TSI parameter before, and when this is very low to being bored, and there is nothing coming in from the human partner, say, then Tango has something of a feedback loop, where it puts its own output back into its input, and so plays a solo, providing its own input by its output, you know, which is basically the same situation that you were talking about, with two Tangos playing with each other.
[K] So, it’s equivalent to a feedback loop if you have two, is that what you’re saying?
[HB] Basically, it’s very close to that.
[K] But have you tried it? Have you actually run two of them against each other?
[HB] No. But Tango is using its feedback thing all the time, but of course, not…
[K] I wonder if, and I’ll just run with this a little bit, if you started it with human input, and then let the other one take over, so you ended up with the initial duet being you and the machine, and then let the other machine take over your part, since you’ve given it a germ to work with.
[HB] Well, you have to do that anyway. There has to be the seed initially, for it to start. When you set Tango up it will just wait for your first note, so somebody has to do something, then, in order to get music out of it. But, as I said, I never tried it with two machines actually sitting opposite to each other.
[K] There’s something… Cageian, about that, I think. Anyway. Tell us, you have a CD coming out.
[HB] Um, I hope it’s going to come out soon. That was the first time I actually tried to expand this Tango situation, which was originally a duo situation, from some human and a machine, into a trio situation. An old friend of mine, a piano player, John Taylor, and I had done a radio broadcast in December ’94, here in the WDR, live for an hour, with Tango. He played a MIDI piano, and of course I played the trombone with a pitch-to-MIDI thing and Tango plays its keyboards. These days, Tango is also able to sample what we are doing in real-time. So, that was that, and we thought we should do that again in an ideal situation, in a studio situation, so we spent three days here in the radio, a nice studio, and we recorded music for a CD that will hopefully come out soon.
We listen to The Heart of the Matter by Henning Berg and John Taylor [1:13:52–1:17:45].
[K] Who is your audience? That’s always an interesting question, particularly when we’re talking to you here, in Cologne, the audiences, the performers, the atmosphere with regard to contemporary forms of expression… all of those things seem to be be very excited. And, sometimes in America, maybe more often than not, there is not an audience, or even a body of performers, interested in ground-breaking ideas. How is it different here (or, since you’re here all the time, maybe you can’t tell the differences), but what is your experience with audiences and performers here?
[HB] Basically you described the American situation, as it could have been a description of the situation as I look at it here. So, I don’t think it’s that ideal that the audiences come to concerts, and now I think it’s as slow as you described it from over there.
[K] But you can go to WDR here, and work with them. You know, we can’t see many performers and contemporary composers going to our National Public Radio, or even many of our local stations in the big cities, and having much success.
[HB] Okay, maybe that is a difference, that stations like public radio stations here in Germany still keep departments that don’t necessarily serve the majority of people who would like to have pop music and regular amusement shows, and stuff like that. So, there is on each of these public radio stations a department for contemporary new music, a jazz department, and people who take care of minorities who are interested in something different from the average taste. That obviously works only because public radio here is paid by a large amount of people, who pay a monthly or quarterly-something fee, and is not only paid for by advertising.
So, yeah, maybe I was too pessimistic there, that’s a big difference. But, before you were mentioning the concert situation. As you soon as you leave this subsidized sector of public radio, with also its facilities of concerts, with a prestige involved in having a concert on the radio, also pulls people into the radio halls, and there are people listening to that. But, as soon as you go to little communities where they have societies for new music, or something like that, you will find yourself sitting there, playing in front of 5 people, or 10, or 15 if you’re lucky. I think that’s not too different from the situation you just described. So, a lot of the stuff here happens either in the radio situation, which is probably the same in your world, or in a university situation, where the big music academies have electronic music studios, nicely equipped, which is available in the United States as well, isn’t it?
[K] Yeah, to the university community. Not to the outside world, to independent composers.
[HB] Well, you have a teaching situation there that also involves creative work, and usually these studios are booked up 24 hours a day. People come there to work at night, and stuff, so yeah, that’s how it works here as well.
[K] Let’s talk about your music again. You said something earlier about having to think more when you’re working with Tango, that it forces you to take more things into consideration. Does that conflict with what many people would call the intuitive nature of improvisation? Does that create some kind of conflict in you, as it were, or does it open new directions?
[HB] I don’t see a conflict with intuitive playing, because I don’t believe in playing that is entirely intuitive. There are always rules, I mean, you never drop every possible frame of what you are just doing. I mean, you are playing with somebody in a room, you are playing with somebody over a certain amount of time in front of a group of people, you’re playing certain instruments and stuff. There’s always limitations, and these limitations have to be taken care of also in an improvising situation with a machine. Say you want to play a concert of about 45–60 minutes, which is about the format when I’m playing alone with Tango that I prefer to do. You can say, “Okay, I have to subdivide it into, say, 8 segments of so-and-so many minutes, or 3 segments or 15, or whatever,” and you have to think about what you can do to make it as little boring as possible, and worthwhile as possible.
So, you will start arranging things. You will think about instrumentation, what kind of instruments is Tango supposed to use, what are you going to do with the music, and textures, and so on. That stuff I’ve mentioned before, foreground and background, when are you on and when is the other one on? Tango and I, we also have our little cues, very definite stuff, note combinations that enable me to switch on or off certain functions within the program. For instance, there’s a very important switch at the outside of the program that is called “freeze,” which does nothing but stop Tango from listening any more. If I press this button “freeze,” which is for me now just a combination of notes that I play on the trombone that will trigger this particular function, then Tango will play exactly as it played before, and not change its playing as long as I “de-freeze” the system again. So, Tango will move into some kind of pattern, a repetitive pattern, that might not be exactly repeated because the other parameters suggest not to repeat something exactly, but to stay in the same statistical range. Say, a few short notes and a long one, something like that.
This function is very simple, but very useful in order to have Tango produce background material very quickly. That’s something I didn’t think of before, when I thought about foreground/background material, what can background material be? It can be soft, very quietly played, but it can also be louder, and then it has to have something like a pattern to it. It has to have something repetitive to it, it has to have something where the listener doesn’t expect something new all the time, to be able to concentrate on something that comes from the other guy at that particular moment.
[D] On the CD, is Tango interacting with John Taylor as well?
[HB] Yes, in several ways. Of course, I play the trombone with the pitch-to-MIDI device, which Tango listens to, and John plays MIDI piano, a normal grand piano with a MIDI interface which Tango can listen to in the same way. But there’s another way for Tango to listen to us. We had kind of a feedback loop from the recording booth into Tango, that just fed the stuff that John and I played acoustically.
Now, in the more recent versions, Tango is able to sample sound, not only listen to MIDI data, but also sample the real sound that is produced right now, and Tango then copies those samples into a big chunk of memory somewhere, and looks at them, analyzes them, in terms of how long are they, how loud are they. It can chop them up in very small segments, put them together in a different order, play them forward, backward, at the same time, and does that a lot. Also, it can send them to pitch-shifting effects and stuff. So, in these two pieces (actually it’s the first and last pieces on the tape), Tango doesn’t listen to MIDI stuff at all, but listens only to the real sound, records that sound and plays it back as something different, as it was being played in real-time.
We listen to Garden by Henning Berg and John Taylor [1:30:08–1:31:55].
[K] You personify Tango. Is Tango a combination of a clone of yours, and a slave to you? You give it a real personality.
[HB] Well… It has kind of a personality to it, it changes all the time. At least, it’s an internal construction site that you think of when you wake up in the morning, and many times during the day. In Germany, it would be even more obvious that I think of it kind of like a person, because in German, I always use er, [“he”] for Tango, instead of it [“es” in German], the program.
[K] Is Tango your pioneering act? Or, is the music that you play with Tango your pioneering act?
[HB] That’s a very good question. The reason I have Tango, is because I want to play music with it. Writing Tango, you run into technical problems all the time, that you want to solve, and that are sometimes frustrating and sometimes a lot of fun to solve. Writing Tango is something different from writing, say, a word processor or a statistics program, or something like that, because it’s a long process of growth. Sometimes, you sit there for 2 days and you’ve written 3 lines of code. You’ve spent hours and hours testing it, because as I said, you don’t want it to be predictable in terms of “When I do this, it will do that.” I want it to work properly on a higher level. So, it’s very time-consuming, making this thing grow and get better, and learn, in a way. On the other hand, it’s very time-consuming to get music out with it, to perform, and many times I realize that those technical problems kind of keep me from really going onstage with it, work with it, play with it, think of musical output to be produced with it, which is the main purpose, the only purpose I have this for.
So, actually, this soon-to-come-out CD, will be one of the first attempts to get something straight, that is, only music, and not a fascinating live performance of somebody playing with a machine.
[K] Yes, aha. I understand what you’re getting at, and I understand that having spent, what, 10 years now with that, it should be a hard divorce, should you have to separate yourself from the program, should it ever concluded?
[HB] No, no. I have thousands of ideas that will certainly cover the next 20 years, in terms of things to work at, and I’m very grateful to that, if I ran out of ideas of what to do with it. There’s still so much missing that I can describe very definitely that I want from the program, and just needs time and effort, and growth. Many times I compare it to a building. You have a building somewhere, you live in it, and eventually you find out you need a garage, or sometimes you need to do work at the foundations of the building because you realize there’s something wrong with it. There will hopefully always be stuff that I want from the machine, but what I don’t want to lose is… the reason is, to go onstage and play, or to come up with music on a tape, and play this music to people without saying, “This is what I did with the machine.” The main point is, and I realize that it’s possible now, also after these recordings, that the stuff John and I came up with, for instance, but also a lot of the stuff I came up with before, recordings of live performances, this stuff can only be produced this way. It would be virtually impossible to come up with this kind of music in any other way.
[K] So, there is no “clone” or “slave,” this really is something that begins to have its own personality. It begins to develop its own way of improvising with you. Since it couldn’t be produced any other… you couldn’t have another human there who would work the same way.
[HB] That was never the idea of the whole thing, to just imitate an improvising human musician, because tonight I’m going to be improvising with a fantastic piano player here, from Cologne, and he gives me things that Tango will never be able to give me, and the other way around. It was like that from the very beginning, it’s not that Tango is more patient, or something, it’s a different quality of music.
[K] Henning Berg, thank you so much for joining us today on The New Music Sesquihour. It’s been a great pleasure, and maybe we’ll be back in the fall and get a chance to hear how Tango has developed.
[K] Thank you.
[HB] Thank you for listening.
We listen to Hey Presto! by Henning Berg and John Taylor [1:40:10–1:42:30].