from design to audio design

This year I had something to bring to the european Triode festival (eTf): a lecture about the nature of design.

The idea for this lecture came to me a couple of months ago. I am a user interaction architect and I also teach interaction design at a university of applied sciences. One of the pillars of this course is a design methodology that is universal. I always tell my students: ‘If the next thing you have to design is a water tap or a film poster, you can use the same steps.’ I am not going to explain the methodology here, that would take to long. But I did have this thought: if design processes are so universally applicable, what could I say about audio design?

So I thought about it for a while, and here we are. What follows below is the complete lecture, where I take my two decades of experience in interaction design and translate it to the world of audio design, based on a selection of axioms that I use all the time in my job. It is not my goal to convince you, it is to share my thoughts and let you reach your own conclusions.

a tale of two worlds

To get started, we need some definitions of what I am talking about.

What is this interaction design that I do? In short: I make digital stuff work for users. Whether websites, mobile, touch (tablets) or desktop applications. I can make:

• an info terminal in a museum usable from the first second;
• software for experts incredibly powerful;
• the flow through software very smooth;
• users feel very safe while interacting with software;

but not necessarily all of that at the same time.

As far as I am concerned, design is solving the problem. And when you have identified the real problems and tackle them, you will find yourself automatically innovating. It is the nature of design to move forward in jumps, not by itty-bitty incremental improvements.

not skin-deep

What is this audio design that I want to talk about? Let me be clear that this article is not about industrial design of audio gear, like this…

…or this slightly more pedestrian effort:

Instead, I want to talk about the design of how it works:

The same for loudspeakers, it is not about whether to paint this speaker—I don’t know—blue or pink, it is about the plan of how it works:

And the same deal for front-ends:

you vs. them

Some more definitions. In this article I will talk quite a bit about designing audio products. Most of us are still engaged in buying audio products, so what can be learned about the design of them is of interest to us. Some of you are also involved in creating, manufacturing and selling audio products. But in the world of eTf and ultra-fi, a sizeable group of people are at the opposite end of the spectrum: we build and modify audio gear, solely for our own use.

I had to find a name for this world, because it will also feature quite a bit in this article. Then I remembered Tim de Paravicini, who was an eTf guest six years ago. In his lecture he really wanted to make sure we got the difference between hi-fi (i.e. him designing audio products) and my-fi (us building for ourselves). From a designer point of view I can now understand why he wanted to press that point: designing products is really a different world. But I also happily adopt the term my-fi for our ultra-fi world of DIY.

tea time

The final definition I need is that of user interaction in the world of audio. Obviously listening is part of this. But there is more. Herb Reichert wrote in Sound Practice issue 13 about the tea ceremony of listening. You select and put on a record, and then… what do you experience? It is not just what you are hearing.

The room and its lighting; the location of your audio gear; its arrangement on a pile/rack/altar; the way the gear looks: it simply matters, even if you do not want to admit it. Every time you touch your audio gear, it matters how that feels. A flimsy volume knob made of plastic, or one made of solid brass which turns with smooth but firm clicks: a very different experience.

Oh dear. Just a few paragraphs ago I defined that industrial design was not the point and now I conclude that it matters every time audio gear is used. OK, learned a lesson there. Still, for the rest of this article I will concentrate on the design of how it works on the inside.

travelling in three dimensions

With these definitions in place, let’s cut to the chase. This is central to my work as an interaction architect: design is bringing together the dimensions of product, users and technology. OK, let’s translate that:

‘audio design is bringing together the dimensions of product, listeners and technology’

These three dimensions are orthogonal to each other. In design projects, the audio designer ties these together; the audio design itself is the one and only embodiment that connects these dimensions; finally, the people who live in these three dimensions have a hard time talking to each other and the audio designer acts as a translator between them.

Let us explore these three dimensions one by one.

1. technology

All engineering lives in this dimension. One trait of western society is that it speaks for itself that technology is needed to get a piece of audio gear realised. Both listeners and the press speak in terms of ‘engineers developed this amp/speaker/player.’ But:

‘technology is only ¹⁄₃ of the design equation’

You see, it is only one of three dimensions. Now I see some of you thinking ‘bullshit!’—I see quite a few engineers think that in my job—‘All this product and listeners stuff is just fluffy mumbo jumbo. If I just do a fine engineering job and turn out a fine set of measurements, then the result is fine audio gear.’

Well, that does not make the other two dimensions go away. By ignoring them you are making a lottery out of your audio gear development process. Furthermore:

‘the job of technology is to disappear’

Yes, the use of technology speaks for itself, but when it becomes the argument—‘this dac is 25bit/200kHz’—then you are in trouble. Instead the argument should focus on the other two dimensions.

2. product

Product is not some silly marketing game. Product is the essence of what companies sell and exactly what consumers are looking for when buying something. Let me explain this by showing you what makes a good product definition. Whenever I start a design project, I sit down with the leading persons of the project and ask them:

‘what is it we’re creating, who is it for and where is the value?’

The answer to this defines the essence of the project, the big goal that the design is going to realise. Let me demonstrate this with a fictive example:

What is it we’re creating?

‘a phono preamp’

Maybe we can be a bit more more specific?

‘a phono preamp to be used with a line amp’

Ah, this gives us a much better defined goal. We can say something about the output levels required from this design and categorically refuse to design in a volume control—with its compromises. If the answer would have been

‘a stand-alone phono preamp’

—you know, one that is hooked up directly to a power amp—then the output levels and drive need to be higher and a volume control is needed. The latter could compromise our plans, e.g. if we planned to direct-couple the whole preamp input–to–output. If the answer was

‘a full-function preamp’

then you know you have to provide line-level inputs and include a line amp in the design, either by compromising the phono design to include the line amp as its last circuit block, or run the line amp in parallel to the phono part, both with extra complication in switching and making the volume control work for both.

But in itself the ‘what’ question is usually the easiest to answer. It gets already harder with:

Who is it for?

‘everybody’

This is the classic answer. But I can tell you straightaway that we can royally forget about this. There are things that are for ‘everybody’, but they are really boring, they are infrastructure: drinking water, electricity, roads and streetlights are examples. Although hideously complex to provide, no consumer really cares about infrastructure—until it goes missing. Audio is not like this and ‘everybody’ will not work:

‘if you make it for everybody, you make it for no one’

Some hard, viable choices will have to be made.

‘record collectors’

That is better, a designer can work with that.

‘audiophiles’

That is mostly a different user group and a phono preamp can be designed for them too.

‘accomplished musicians’

Ah, one thing I encounter at really plugged-in companies is that they uncover and focus on new but viable target users. If they play there cards right, they can really rule that segment.

And now the most difficult question of all: where is the value? Features are not value. Adjustable phono deemphasis curves or a phase switch are examples of just features for a phono preamp. Value is something much deeper and the essence of why a company is allowed to ask money for their products.

I had a hard time coming up with a statement of value for a phono preamp, but here is one, complete with the ‘what’ and ‘who’:

‘a phono preamp, to be used with a line amp, that will reconnect jaded record collectors with their whole collection’

Now we are talking, this is something a designer can work with. The ‘reconnect […] with their whole record collection’ part is believable as of great value to record collectors, provides visionary direction and is still vague enough to leave room for its design and realisation.

The statement I end up with in my design projects is slightly longer than the one above, two or three paragraphs, but not that much longer because it needs to be used as a tool during the whole project: ‘let us focus again on what we are trying to achieve.’ I call a statement like this a product vision and it really helps with tough design decisions. I actually do not work on a design project without one.

That is a lot of product, what about my-fi? It does not take much to modify the statement above for this context:

‘a phono preamp, to be used with a line amp, that will reconnect me with my whole record collection’

I can advice anyone engaged in my-fi to work with a personal vision like this. It is so much easier to achieve your goals when you’ve defined them. But I also know it is nearly impossible to define a vision on your own. You will need a moderator to get you there, a friend that asks you ‘what is it you are going to build, and what value should it bring you?’ and doesn’t take clichés for an answer.

3. listeners

Into the third dimension: listeners. From the product vision we now know which target group we are dealing with (collectors/audiophiles/musicians) and before we start to design we can research exactly this group. There is however a problem:

‘listeners are always right with what they need, but always wrong with what they ask’

Just asking your group of listeners what they want (i.e. market research) will get you nowhere. What is needed is ethnographical research (this is what usability researchers do in the world of software). A researcher that goes to six or more participants recruited from the target listener group (say, record collectors), asks them questions—‘why do you play records?’—and observes how they talk about it, how they experience the tea ceremony.

culture club

Meanwhile, it really does not hurt to develop a deep understanding of why music exists. Many millennia before audio reproduction was invented, music was already part of our culture. Here are some observations:

• dancing is completely intertwined with music; with dances as a community event and a legitimate occasion for boy-meets-girl, to find a partner for life (very different from the high-end show, isn’t it?);
• think of how much music was developed to be played at official functions, either religious, community, political, military or governmental;
• trace the minstrel tradition, storytelling set to music, from ancient and medieval times, to blues singers and folk troubadours;
• music as a means for groups to assert their identity; from the oppressed throughout the centuries, to youth culture after WWII;
• music as a cultural transport vehicle: used to travel to other parts of the world in times that mass tourism did not yet exist;
• music as a time machine: travel decades or centuries back in time.

The list goes on and on. Having a good grasp of the culture of your target listener group is required of designers. Even if all this plays on a subliminal level, the human factors involved can be very strong. Of course one will have to check how much all of this really applies. I can imagine that after thorough research one has to conclude that not much of this music culture matters to audiophiles.

tock

Overall, a designer has to analyse how the target listener group ticks and what their needs are. And in the case of designing products this can be done with an objective distance.

In the case of my-fi this becomes much more difficult. It is very hard to analyse oneself; it is all about you. But do not despair, in a moment I will present a shortcut.

the third degree

I expect from audio product designers that they can talk with equal depth, length and respect about the three dimensions: product + listeners + technology. That they are able to connect these dimensions, to harmonise the value that has to be delivered with the needs of target listeners, using technology. For instance, deliver that ‘reconnection’ to record collectors.

For my-fi, the dimensions change to vision + you + technology. To be able to call yourself a designer of your own gear, I expect you to be able to balance your vision with your needs and the technology available to you. Note how between these two worlds technology is the common factor. That is exactly the point: technology is agnostic of goals and needs, just a means towards an end.

‘if you make it for yourself, you make it for no one else’

That’s right, when designing a product you will have to take yourself out of the equation. When you evaluate your design—e.g. listen to it—you cannot rely on what you like. Instead you will have to picture thousands of target listeners and how they tick, and evaluate from their perspective.

For my-fi, I invite you to read that same axiom the other way around: if you are doing it for yourself, why would you try to please anybody else? Why conform? Take RCA plugs, we know they are a non-standardised joke of a connector. Why use these things if you make all the rules? For a couple of €/£/$ you can buy a precision-made connector for scientific sensors. If you free your mind, that could be a much saner choice.

to listen, or not to listen

What about peer pressure? Some audio mates visit you and listen to your my-fi system. Should you simply ignore their critique? No, you are not doing yourself a favour with that. A designer’s way to handle this is to weigh the feedback by your personal vision and act accordingly. Are they trying to push you in another direction? Ignore them. Are they pointing out that you missed a goal you set for yourself? Start redesigning to do better.

Here is the shortcut I promised you for my-fi. You can short-circuit the part of finding out how the target listener group (i.e. you) ticks with the evaluation part. Put yourself in the middle of the equation and evaluate purely by feeling. By all means, work with a personal vision to direct where you are going. But you can skip some of the intellectual work that product designers have to perform.

more on (audio) design

‘design is done on paper; stay on paper for as long as you can’

This is an aspect of design I cannot stress enough. What we want to practice is the art of throwing away; to ensure that you are not ‘married’ to a certain solution too early in the process; to be able to whip up a radically different design in a matter of minutes:

(please do not check out these sketches too closely, there are probably some errors in them)

Paper is the place where that innovation, moving forward in jumps, happens. Drawing the same thing on a computer, simulating or prototyping it is working in small, incremental steps, compared to the raw power of working one paper. Related to that is:

‘when I have finished the design on paper, I already know that it is going to work, and how’

This one can really freak out my collaborators in the world of software. How can I be so sure before anything has been simulated or built? It is not simple to explain, but it has a lot to do with knowing what makes your field of design tick. And with having years of experience of how your design choices relate to the final result.

In this context I was reminded of Tim de Paravicini again, who said in his lecture that after he was done developing an amplifier—which meant in his case designing it and bench testing of a prototype—he never listened to the result; he already knew it was going to be great. That statement freaked out a lot of us, including me for the last six years.

That is, until a couple weeks ago, when I connected it to the axiom above. Now I can understand how Tim can work like that.

what about SPICE?

Speaking of simulation, I asked myself if I could say anything about the usefulness of SPICE in audio design. When I evaluate the user interface of SPICE I see no support for expressing a vision or listener needs. SPICE lives solely in the technology dimension, which means it can cover at most ¹⁄₃ of the design equation.

I say ‘at most’ because that ¹⁄₃ has to be multiplied with how reliable SPICE is in your own experience in giving the right technological answer. Make up your own mind and that is the usefulness of SPICE in audio design.

‘the mid-level stuff is easiest to grasp for junior designers; master-level design is in the big plan and all the details’

I learned this while mentoring younger colleagues. Let us first check out what this mid-level is in audio design:

Setting up a tube stage, finding an operating point and component values. And I do not mean that only from a technology point of view. Here is a great place to demonstrate how one starts learning how design choices relate to the final result. Increase/decrease the plate load and observe how the sound gets leaner/richer. Increase the time constants of capacitors and hear the amazing bass, and crappy treble; decrease time constants and hear how everything starts breathing organically, with wooly bass.

Applying all this to set up the stage is mid-level stuff. But let us look at a bigger plan:

The whole amplifier, its overall architecture, the gain structure. Additional to the voltage type of view (very mid-level) we can take the current view: analysing the multitude of AC current paths through the amplifier and start managing them.

And when I say the whole amplifier, I mean the whole amplifier; how it all works together as one thing. It is very much a western culture thing to divide the problem of designing an amplifier into sub-problems—input/driver/output stage, power supplies, heaters and filaments, grounding—and solve them one by one.

Finding the ‘best’ and coolest solution for each of these sub-problems is not the same as designing a good, whole amplifier. Seeing the whole picture is master-level stuff and can involve picking ‘inferior’ and uncool part-solutions so that everything fits much better together, for a superior overall result.

Let us look at an even bigger plan. A system—a record player, phono/line/power amp and a speaker:

Again there is an overall architecture. Gain distribution has to be harmonised with the input sensitivity of all components, including that of your ears. Input and output impedances, and drive, harmonised. Again creating a whole system is master-level stuff.

There is one more component between the speaker and your ears:

The room. Here we see that my-fi has an advantage. You can either match speaker and power amps to the room (and your listening needs), or build a room to match your system (and needs).

For a designer of speaker products, things are more complicated. You cannot design for a specific room. A design strategy to handle this looks as follows:

1. decide on the overall range of types and sizes of rooms you want to design for—remember, audio is not infrastructure, so you do not have to handle everything;
2. analyse this range and try to find big groups of similar rooms;
3. six groups is a good number to aim for; if you find some small groups, ask yourself if they are a market you want to be in;
4. for each group of rooms, you can then design one or more matching speaker products.

get down

And now the opposite direction, taking care of details. This starts with stopping to assume that things are perfectly straightforward. For instance grid current: way before you cross zero volts, you are riding a roller-coaster of changing grid impedance. While I gather Miller capacitance and heater–to–kathode ratings to be mid-level stuff, taking care of details means also dealing with heater–to–kathode leakage—and for dual tubes, kathode–to–kathode leakage.

The list goes on and on. The supply current draw of class A and/or push-pull circuits being not so constant as advertised—it follows the envelope of the music. The thermals of semiconductors, their parameters wandering all over the place, depending on instantaneous power dissipation—again envelope-following the music. I am sure that there are many more master-level details that I cannot even name.

going underground

For a designer, taking care of details is about ensuring that they do not spoil the party. It is about protecting the big plan and the mid-level set-up from being sabotaged.

Thinking about this, I can now understand why some audio designers—and some of my ultra-fi friends—got into custom parts, e.g. transformers, capacitors, speaker drivers, even wire. From specifying custom part orders, via modifying, to manufacturing parts with their own hands, they are taking care of details.

what about listening tests?

To finish off this article, let’s look at something tempestuous. In my job of interaction architect I deal quite a bit with testing: usability testing. One thing I always have to keep in mind is: there is a clear difference, and a hard trade-off, between ease of learning and ease of use. Let’s translate that:

‘there is a clear difference, and a hard trade-off, between the experience of learning and the experience of use’

Experience of learning is what happens when you audition a piece of audio gear, or a system (sometimes hard to differentiate between the two, no?). This happens in the shop, when you visit a friend, at the high-end show and the eTf, and right after you bought, borrowed or built something. Experience of use is what happens further down the road, when the honeymoon is over and it speaks for itself that a piece of audio gear is in your life.

Designers will have to make a hard choice for the trade-off between these two. Go for 100% experience of learning and there will be zero experience of use—and vice versa. Some linear combination, say 30% learning–70% use, is where the answer is.

choices, choices

If you design audio products that are sold in the shop, then you really have to plan for the experience of learning, that wow effect that grabs peoples’ attention. If you, like some of my friends, design audio products that get sold by word of mouth, then tuning towards experience of use looks like the a winning strategy. You will have long-term ambassadors that swear by your product.

For my-fi practitioners there is also a choice, it really depends on what your ultimate goal is. If you are doing it for the satisfaction of building stuff, then you can prioritise the experience of learning and knock yourself out every 2–3 months with a new amp or speaker. You will also need to keep building because that thrill will evaporate in weeks. If you are into my-fi to create your personal system, then I advice to optimise for your experience of use, even if that means somewhat unspectacular results, at first audition.

so what about the !@#$%^& listening tests?

OK, what is the case with usability testing is that you can only work with participants for an hour, maximum hour and a half. Fatigue sets in really fast. I am sure it is the same with listening tests. Being relatively short, these test measure the experience of learning. For some of the projects I do I could really use a testing method that works over a month or more so that we could measure the experience of use, but no such method exists today, AFAIK.

What to do when experience of use has priority in your design goals? Listening tests are not going to help you with measuring that. As an audio product designer you will use other methods. First of all there is that knack of making it already work on paper that I discussed before. Second, as a designer you evaluate when you listen to your designs, which is faster, more detached and analytical—‘not through your own ears’—than a listening test.

A my-fi practitioner can avoid listening test, and its pitfalls, by implementing a simple long-term test method: live with it, for a month or more. For extra bonus points: do not choose yourself the records to play in the first days, to avoid test-record bingo to multiply with the learning experience. Let your partner pick the records, or listen to some good radio.

postscript

To recap, design is about dealing with the whole thing; understanding all the dimensions and connecting them. It is about getting the goals clearly defined and getting a good basis to start from; find out what works and what not at this moment. From there, design is solving the problem and innovating in jumps. And with that, design is about creating the future.

That’s it for the lecture itself, but stay tuned for the interesting q+a that followed and some bonus material—a whole section that did not make it in the final cut of my lecture.

But for now I leave you with a quote from my friend Aaron Day, who is a musician, recording engineer and producer, sound designer for software and devices, general user experience shaman and an electronics tinkerer:

‘Remember, the most important part of the signal path is a good song.’