Originally Posted by dyn
Happy New Year Everyone...
I have a question to start the thread for this year. I hope Otto can answer this question.
I fully agree and understand the fact that there are countless of matters that contribute to how a speaker sound. Even a tiny change will result in different sound. And sometimes the tiny change results in a significant change in sound quality.
Given this fact, how do you design a speaker with the same characteristic over 40 years as shown on the recent review by LowBeats?
To take this further, how do you know that the new design (for example the new esotar3 or the new hexis design or even different internal cable) will result in better sound quality and, at the same time, same characteristic with the previous speakers?
How do you even know if the new design will result in better sound quality? Once again, we all believe that small difference can result in significant difference in sound quality and charcteristic. And we just do not know whether that change will be better or worse until we listen to it.
I really do not understand. You can have better design, more advanced technology, etc which, you think, will result in better sound. But the truth is you do not even know how they would sound. For me, this is like walking with closed eyes and hoping we will get where we want to go.
I am a Dynaudio fanatic and wondering how the sound, which I am so facinated with, is designed.
Happy New Year!
Interesting questions. I will take the liberty of breaking them up in two parts: How do we create a persistent sound characteristic, and how do we validate that a change is an improvement.
I think the answer to the first part is that we do not have a fixed "reference" speaker that we try to match. Our reference is not a product, our reference is the music. Since the beginning in the late 70's, the mission for Dynaudio has always been to try to avoid having a "house sound", and instead present to you what's in the source. Now, let me be the first to say that this is an impossible mission, for a variety of reasons - but nevertheless, we believe that getting as close as possible to this imaginary goal will benefit the listener. The harder it is to identify the "sound" of the speaker, relative to the "sound" of the recording, the better. We do validate against various other of our products, to make sure we are on the right path, but sounding similar to old Dynaudio products is never the goal - that is just a consequence of working towards the same universal goal as we always have.
Perfection doesn't exist, so there will always be some variations, sometimes caused by practical or economical restrictions. But we constantly try to improve, and we believe strongly in incremental updates. Think about compound interest - audio works in a similar fashion. A 5% improvement 20 times is not 100%, it is much more! That takes us to the second part of your questions, how do we validate that a change is an improvement? That is complicated, and a lot of the answer is simply experience, and a lot of patience - yes, we do have lots and lots of trial-and-error. Based on knowledge, experience and imagination, an engineer may come up with a theory on how to make an improvement. First step often is to simulate. If that looks promising, build a prototype of the part. If measurements look promising, build a prototype of a speaker - and start listening. Generally, objective measurements can tell you if something is wrong - but it can't tell you that everything is right. For that, you need ears.
One of the things I have concluded from working with our staff of engineers is that "experience" is more than simply working with something for a long time. It's also about having a good frame of reference. Listening to bad speakers for 20 years won't teach you how good sound sounds. Listening to a good speaker for 10 minutes will teach you a lot more! Simply using high quality speakers in every day life will train you to not accept flaws in poor speakers (especially if you are an audio geek from the outset, but even if you're not, you will still experience this phenomenon).
I definitely don't agree with your analogy of walking with your eyes closed. With experience and knowledge, the quality of the trial-and-error process increases greatly, and you become better at predicting the outcome. You also become better at realising where the flaws and limitations in your current design is, so you can focus on improving those - without introducing new ones.
In general, we believe that eliminating weaknesses is better than tweaking your strengths to the extreme (at least for loudspeakers). If the sound is fantastic in one area, but fails spectacularly in another, over time we believe the annoyance of the flaws will outweigh the enthusiasm for the strengths. A flaw or weakness will draw your attention to the speakers, when it should be on the music. So the answer to how you can improve while retaining the same overall character, is that "change" and "improvement" are two different things - we don't change for the sake of changing, we change if it causes an actual improvement - if you can remove a flaw, without introducing a new, different flaw, then you have improved. If you sacrifice one parameter to improve another one, very often you have just changed, not improved. For instance, for the Contour 20 the engineers were given the specific task of improving bass performance over the previous model, but not sacrifice midrange performance while doing it. Getting more bass is easy - doing it without sacrificing midrange is a lot more challenging.
The really fun part of building loudspeakers, and what makes it an art and not simply science, is that the whole concept of replicating the sound of one room inside another room, is physically impossible! It's all literally an illusion, and our job is to make that illusion as believable as possible. This means that in the end, physics (objective measurements) is not enough, the final validation is always listening. But, simulations and measurements allows us to easier, better and faster predict changes that will improve the actual sound.
I hope this is a satisfactory answer - in the end, there is no simple answer to your question, other than good engineers get better results than bad ones...