Interview with Dynaudio chief engineer Stephen Entwistle

[Zvu]

https://soundcloud.com/johnhdarko%2Finside-dynaudios-focus-30-w-stephen-entwistle

 

[SpanishPassion]

After hearing this interview, as a Dynaudio customer, I can say, that I am disappointed.

Unfortunately, I didn't take notes while I was listing to it, but what I got from it essentially is the following:

1. Dynaudio doesn't really follow a proper engineering process or methodology when designing the speakers. This is evident by the statement that they spend around 30% of their time doing the measurements and 60% of their time doing subjective listening in the listening room. He didn't describe any methodology that they follow when performing this subjective listening, so this activity probably is just wasting time when it comes to a tangible outcome.
2. They claim that there are characteristics of the speaker that can't be measured and can only be heard. As an example, he gives something like the sound power that speaker produces in the room, I didn't really understand it. Apparently, this is supposed to affect the sound quality a lot.
3. Poor vertical directivity is intentional. Apparently when it's that way, there are less reflections from the floor, so the sound on axis is better. This decision is derived directly from their subjective listening "tests", not from the measurements of any kind.
4. According to him, you can't really hear the "anomalies" in the vertical directivity. Well, I don't know about the "anomalies", but I sure can hear that it is extremely narrow.
5. In general, the Dynaudio's (narrow) directivity gives better image, according to him. I assume that in some rooms this can give you less reflections, but seems to be very room-dependent.

Overall, what I got from this interview, is that we shouldn't expect any significant improvements from the upcoming Dynaudio speakers. They don't want to address the technical issues like poor directivity. Even on-axis, I don't think they want to have a very flat frequency response, because apparently a luminary that performs the subjective tests there doesn't like it that way.

Personally, I don't see any reason to buy those speakers anymore. They are not bad, they are surely reasonably flat on-axis, but given that they have these technical drawbacks which are built-in intentionally, this is just wasting money when compared to competition. And that's a shame, because their speakers are extremely convenient to use and Dynaudio as a company clearly can produce state of art speakers.

I wonder why did they invest in Jupiter (which I think is a very fine tool) if they rely on it 30% of the time when designing a speaker? Seems like a rather poor investment to me.

 

[Zvu]

We do need to be aware of design process to be able to judge it. I'll address only first two.

1. If you listened what Stephen said, he mentioned Jupiter system. You can find pics of it on the web. It is a measurement system that uses over 30 microphones mounted on an arc picking up log sweeps. So they can sample 30 measurements at 30 different angles in one sweep. Of course they spend only 30% of design time doing measurements. Measurements are straightforward job, precise and simple to do. Our ear/brain combo, on the other hand, is much more complicated. Few minutes listening to the familiar music and i can hear tonality with regard to frequency amplitude response. But i need at least a week or more of listening to be able to hear if there are flaws that aren't too obvious.

2. Sound power is not one of those things that can't be measured but can be heard. He says that Sound power is more important to them than directivity curve - and it matches my experience. Directivity curve shows if a loudspeaker will react good to equalization. Sound power shows all of the sound produced by the loudspeaker in all directions which is averaged to a single curve. If there weren't things that can be heard but can't be measured, no company would conduct listening to their products or blind A/B tests. They would just make a crossover in simulator, measured it in a chamber to confirm it and ship the loudspeakers to dealers.

 

[Flak]

After hearing this interview, as a Dynaudio customer, I can say, that I am disappointed.

Unfortunately, I didn't take notes while I was listing to it, but what I got from it essentially is the following:

1. Dynaudio doesn't really follow a proper engineering process or methodology when designing the speakers. This is evident by the statement that they spend around 30% of their time doing the measurements and 60% of their time doing subjective listening in the listening room. He didn't describe any methodology that they follow when performing this subjective listening, so this activity probably is just wasting time when it comes to a tangible outcome.
2. They claim that there are characteristics of the speaker that can't be measured and can only be heard. As an example, he gives something like the sound power that speaker produces in the room, I didn't really understand it. Apparently, this is supposed to affect the sound quality a lot.
3. Poor vertical directivity is intentional. Apparently when it's that way, there are less reflections from the floor, so the sound on axis is better. This decision is derived directly from their subjective listening "tests", not from the measurements of any kind.
4. According to him, you can't really hear the "anomalies" in the vertical directivity. Well, I don't know about the "anomalies", but I sure can hear that it is extremely narrow.
5. In general, the Dynaudio's (narrow) directivity gives better image, according to him. I assume that in some rooms this can give you less reflections, but seems to be very room-dependent.

Overall, what I got from this interview, is that we shouldn't expect any significant improvements from the upcoming Dynaudio speakers. They don't want to address the technical issues like poor directivity. Even on-axis, I don't think they want to have a very flat frequency response, because apparently a luminary that performs the subjective tests there doesn't like it that way.

Personally, I don't see any reason to buy those speakers anymore. They are not bad, they are surely reasonably flat on-axis, but given that they have these technical drawbacks which are built-in intentionally, this is just wasting money when compared to competition. And that's a shame, because their speakers are extremely convenient to use and Dynaudio as a company clearly can produce state of art speakers.

I wonder why did they invest in Jupiter (which I think is a very fine tool) if they rely on it 30% of the time when designing a speaker? Seems like a rather poor investment to me.

I'm not surprised by the Dynaudio statements and they sound quite reasonable to me...
he says that measurements are very important but you then have to take decisions about design choices that produce different measurements.
as Zvu mentioned, measuring with an automated system like Jupiter likely takes much less time than listening tests.
If you listen from min. 31:30 you'll hear him say that directivity matters but total sound power does as well,
one can agree or not but in order to decide, one cannot rely on measurements only... you'll need to run extensive listening tests and I don't think that it's a black vs. white choice.
Also, one may have to choose between a wider radiation pattern that can be more pleasant because of the wider soundstage created by the reflections or a narrower radiation pattern where you'll listen to more of the recording and less of the room.
Again that's a design choice that is likely based on listening, the data are what they are.
And so on... in my humble opinion of course

 

[Kal Rubinson]

This is evident by the statement that they spend around 30% of their time doing the measurements and 60% of their time doing subjective listening in the listening room.

It probably takes relatively little time to do the automated measurements compared to the time for multiple listeners to draw opinions about the sound (including multiple recordings of various music at different levels and in different listening rooms). Essentially, I would not draw any inferences from this statement.

 

[ctrl]

1. Dynaudio doesn't really follow a proper engineering process or methodology when designing the speakers. This is evident by the statement that they spend around 30% of their time doing the measurements and 60% of their time doing subjective listening in the listening room. He didn't describe any methodology that they follow when performing this subjective listening, so this activity probably is just wasting time when it comes to a tangible outcome.

So 30% time spent on measurements I think is a lot (and I'm really into measurements).

If the design process of a loudspeaker is done carefully (which takes a lot of time before the first prototype is build and is not mentioned at all in the interview), then the loudspeaker ideally needs to be measured only twice (the individual drivers are of course additionally measured in advance independently of the later loudspeaker in order to determine HD, IMD,... to make sure that the planned XO is realizable with these drivers and no unwanted resonances are caused by the drivers in the used frequency range).

Once to determine the individual frequency response of each driver in the loudspeaker and to use it to "feed" a development tool like VCAD.

VCAD is then used to develop the active or passive crossover. For this purpose, the loudspeaker no longer needs to be measured, since VCAD simulates the behavior of the loudspeaker very accurately based on the measurements. Step by step, the simulation tool and listening sessions are used to develop the best XO.

Depending on the crossover chosen, it may be necessary to check distortion or for resonances of the speaker by measurements in between.

The second time a full set of measurements is done with the final loudspeaker to make sure that the simulated XO is confirmed by reality and the chosen design does not cause unwanted resonances.

In between there are many listening sessions to optimize the details of the XO.

They claim that there are characteristics of the speaker that can't be measured and can only be heard.

The question in the podcast was phrased differently. Something about whether "audible qualities" cannot be derived directly from the measurements.
That's quite different from saying it can' t be measured.

Predicting from measurements the effect of, for example, a slight change in vertical radiation (e.g. by changing XO frequency) is very difficult, although it may be easy to hear the difference. So it can be measured, but we humans are hardly able to predict the "acoustic effects" of minor changes in the measurement curves.

Conversely, we can interpret the sound changes we hear when comparing two XO variants relatively easily on the basis of the measurement curves.

As an example, he gives something like the sound power that speaker produces in the room, I didn't really understand it. Apparently, this is supposed to affect the sound quality a lot.

He tries to explain that the straightest possible on-axis frequency response (when listening in a room) alone is not a quality criterion.
He said more important is the sound power SP response of the loudspeaker in the room (PIR is very similar and could be used too). He justifies this with the "critical distance" of a loudspeaker in the room. It's the distance from from the speaker in a room were the direct sound (free field) field and diffuse sound (reverberant field) field is equal.

Source: Arta Handbook

This radius around the speaker depends on the room size, the reverberation time and the directivity of the speaker (for the exact calculation see the wikipedia link).

Many underestimate how small the critical distance of a loudspeaker is in a normal room and that the diffuse sound field (a simplified representation is the SP frequency response curve***), especially for wide dispersion loudspeakers, is dominant over wide frequency ranges compared to the direct sound.

For an omnidirectional radiating loudspeaker a critical distance depending on RT60 results:

Source: Arta Handbook

Since the square root of the directivity is used in the calculation of the critical distance, a loudspeaker with a four times higher directivity has only twice the critical distance.

This means that at listening distances of 3-4m in a normal room the diffuse field is dominant for most of the audible frequency band and therefore the on-axis frequency response (or the listening window) does not reflect the "acoustic characteristics" of a loudspeaker.

As you can see, the relationships are complicated and we have not even considered the difference between vertical and horizontal radiation. Or that a listening room naturally does not create a perfect diffuse field,...

What is important, and I would agree with Stephen Entwistle, is that if two loudspeakers have a reasonably straight on-axis frequency response (+-3dB) (this can also be applied to the listening window with some reservations), the loudspeaker with the flatter on-axis FR does not automatically sound better in the listening room.

For the reasons mentioned, it is better to additionally look at the SP or PIR frequency responses, which are the dominant factors for a large frequency range. However, the problem of fixation on on-axis frequency response is not a problem of the Klippel measurement system.

I would disagree with Stephen Entwistle's statement that the directivity index DI of a loudspeaker is of no great importance, put that way - but my post is way too long now....

Update:
*** What I said there is not correct. Direct and diffuse fields are independent of the type of sound source; they are shaped by the acoustic properties of the room surrounding the sound source.
The SP frequency response depends directly on the sound source. Nevertheless, it reflects the behavior in a normal listening room better than the on-axis FR. Since SP is direction-independent, PIR usually reflects the behavior of the loudspeaker in the room better than SP (since PIR has more direction-dependent character with more direct sound content).

 

[SpanishPassion]

1. If you listened what Stephen said, he mentioned Jupiter system. You can find pics of it on the web. It is a measurement system that uses over 30 microphones mounted on an arc picking up log sweeps. So they can sample 30 measurements at 30 different angles in one sweep. Of course they spend only 30% of design time doing measurements. Measurements are straightforward job, precise and simple to do. Our ear/brain combo, on the other hand, is much more complicated. Few minutes listening to the familiar music and i can hear tonality with regard to frequency amplitude response. But i need at least a week or more of listening to be able to hear if there are flaws that aren't too obvious.

I'm well aware of Jupiter system. When they have built it, I was very excited, because I expected them to significantly improve both the frequency response and the directivity of their speakers. Unfortunately, this doesn't seem to be the case.

Then again, why did they bother to have this arc at all if they don't consider directivity to be so important? One would expect that if you have a custom-built measuring tool to produce complete 180 degrees vertical and horizontal measurements, then you would push heavily to improve the directivity of your speakers.

And no, I'm not convinced at all regarding their listening sessions. I agree that there should be listening sessions, but spending weeks listening to music trying to satisfy the subjective feelings of someone is not a valid methodology. Do they at least compare the sound of their speakers to something? Like speakers from their competitors? If not, it looks like a total time waste to me.

He says that Sound power is more important to them than directivity curve - and it matches my experience. Directivity curve shows if a loudspeaker will react good to equalization. Sound power shows all of the sound produced by the loudspeaker in all directions which is averaged to a single curve.

Why is it more important? Why should I care about the sound that doesn't go in my direction?

He said more important is the sound power SP response of the loudspeaker in the room (PIR is very similar and could be used too). He justifies this with the "critical distance" of a loudspeaker in the room. It's the distance from from the speaker in a room were the direct sound (free field) field and diffuse sound (reverberant field) field is equal.

But wouldn't directivity affect this direct sound field much more than the total sound power?

I'm sorry, but I just don't see what is the benefit of this approach. I guess having a narrow directivity can have some benefits, but so far I don't see any data supporting this assertion.

I don't want to be harsh to Dynaudio, quite the contrary, I'm their customer and overall I'm happy with their speakers, as long as you listen in that narrow sound field they produce. But if I had to recommend a new set of speakers to a friend, I wouldn't recommend their speakers, because they have competition that outperforms them. And they don't seem to want to address the issues with their product, which brings the question, why buy them at all?

 

[Zvu]

Why is it more important? Why should I care about the sound that doesn't go in my direction?

If you are listening near field (under 1m distance) or outdoors, your question has a merit. If not, you should know by now that all sound goes in your direction, eventually (be it direct or indirect).

You can have a directivity curve with no large directivity errors and still sound like crap. Directivity doesn't tell you anything about tonality since you can have large amplitude variations in frequency but still obtain good directivity. Sound power curve, on the other hand, tells a whole lot about tonality and partly directivity.

For example, if you have transducers crossed to high, you will get a sound power dip at crossover point. There will also be an aberation on directivity curve.

Sound power curve contains data on amplitude and some data on directivity of a loudspeaker, but a directivity curve has no data about real amplitude of a loudspeaker just a hint if it can be equalized or not to obtain a good sound power reaponse.

That is why i understand their preference to observe sound power curve more than directivity curve.

 

[BoredErica]

He tries to explain that the straightest possible on-axis frequency response (when listening in a room) alone is not a quality criterion.
He said more important is the sound power SP response of the loudspeaker in the room (PIR is very similar and could be used too). He justifies this with the "critical distance" of a loudspeaker in the room. It's the distance from from the speaker in a room were the direct sound (free field) field and diffuse sound (reverberant field) field is equal.

View attachment 225191
Source: Arta Handbook

This radius around the speaker depends on the room size, the reverberation time and the directivity of the speaker (for the exact calculation see the wikipedia link).

Many underestimate how small the critical distance of a loudspeaker is in a normal room and that the diffuse sound field (a simplified representation is the SP frequency response curve***), especially for wide dispersion loudspeakers, is dominant over wide frequency ranges compared to the direct sound.

Is there any settled science on whether listening at critical distance is actually better than closer or farther away than that?

 

[thewas]

Is there any settled science on whether listening at critical distance is actually better than closer or farther away than that?

In stereo reproduction it is always a compromise, closer you get a smaller (less envelopment) but sharper representation.

 

[lowkeyoperations]

I agree that there should be listening sessions, but spending weeks listening to music trying to satisfy the subjective feelings of someone is not a valid methodology.

Are you serious? If you can get a pair of speakers for home demo for a couple of weeks, it is going to tell you a lot more about whether you enjoy them than the published frequency response charts.

I listen to my speakers for weeks, satisfying my subjective feelings , because that is entirely the point of owning them. Listening to and enjoying music. Of course listening to speakers “for weeks” is a valid methodology. It’s quite literally what your customers will do once they buy them.

Why would a few few weeks be too long?
Should it be only 5 minutes?
Is one random song good enough?
Is a few hours a better limit?

There’s no ideal ratio between time spent measuring and time spent listening during development. At least I’ve never read any studies that conclude the less time you spend listening to your speakers during development the better they will sound as a final product!

On top of the listening, they have obviously invested heavily in their measuring tools.

You seem to have used the article as bias reinforcement.

 

[Flak]

I guess having a narrow directivity can have some benefits, but so far I don't see any data supporting this assertion.

On one hand, a narrow directivity means hearing less of the room and more of the recording... on the other hand, the reflections can create a wider soundstage that can be pleasant.
It's a hard life

 

[abdo123]

Are you serious? If you can get a pair of speakers for home demo for a couple of weeks, it is going to tell you a lot more about whether you enjoy them than the published frequency response charts.

I listen to my speakers for weeks, satisfying my subjective feelings , because that is entirely the point of owning them. Listening to and enjoying music. Of course listening to speakers “for weeks” is a valid methodology. It’s quite literally what your customers will do once they buy them.

Why would a few few weeks be too long?
Should it be only 5 minutes?
Is one random song good enough?
Is a few hours a better limit?

There’s no ideal ratio between time spent measuring and time spent listening during development. At least I’ve never read any studies that conclude the less time you spend listening to your speakers during development the better they will sound as a final product!

On top of the listening, they have obviously invested heavily in their measuring tools.

You seem to have used the article as bias reinforcement.

You're confusing comparing two products and using only one of them. you need to alternate between the two products quickly, and seamlessly. Otherwise you would never be able to tell which product is subjectively better. Our memory is just not good at all for anything other than visual input.

 

[lowkeyoperations]

So where is this proven that “you need to alternate quickly, and seamlessly. Otherwise you would NEVER be able to tell which product is subjectively better”?

That’s just like your opinion man.

Give me a set of speakers for 2 weeks. Then give me another set of speakers for the next 2 weeks. I could tell you which pair I thought were subjectively more enjoyable.

I don’t need to eat a mouthful of entre then a mouthful of main course and continue with alternating mouthfuls to know whether I preferred the entree or the main. I can tell you whether I enjoyed a meal more at one restaurant than the restaurant I went to a week earlier. I couldn’t say “the level of salt on the palette in restaurant 1 was slightly more pronounced than restaurant 2”. But I could say subjectively which I preferred.

Their are many here who seem more interested in measuring speakers than enjoying music. You don’t need to listen to 2 pairs of speakers side by side to know whether you enjoy the sound of one of them.

 

[tuga]

After hearing this interview, as a Dynaudio customer, I can say, that I am disappointed.

Unfortunately, I didn't take notes while I was listing to it, but what I got from it essentially is the following:

1. Dynaudio doesn't really follow a proper engineering process or methodology when designing the speakers. This is evident by the statement that they spend around 30% of their time doing the measurements and 60% of their time doing subjective listening in the listening room. He didn't describe any methodology that they follow when performing this subjective listening, so this activity probably is just wasting time when it comes to a tangible outcome.
2. They claim that there are characteristics of the speaker that can't be measured and can only be heard. As an example, he gives something like the sound power that speaker produces in the room, I didn't really understand it. Apparently, this is supposed to affect the sound quality a lot.
3. Poor vertical directivity is intentional. Apparently when it's that way, there are less reflections from the floor, so the sound on axis is better. This decision is derived directly from their subjective listening "tests", not from the measurements of any kind.
4. According to him, you can't really hear the "anomalies" in the vertical directivity. Well, I don't know about the "anomalies", but I sure can hear that it is extremely narrow.
5. In general, the Dynaudio's (narrow) directivity gives better image, according to him. I assume that in some rooms this can give you less reflections, but seems to be very room-dependent.

Overall, what I got from this interview, is that we shouldn't expect any significant improvements from the upcoming Dynaudio speakers. They don't want to address the technical issues like poor directivity. Even on-axis, I don't think they want to have a very flat frequency response, because apparently a luminary that performs the subjective tests there doesn't like it that way.

Personally, I don't see any reason to buy those speakers anymore. They are not bad, they are surely reasonably flat on-axis, but given that they have these technical drawbacks which are built-in intentionally, this is just wasting money when compared to competition. And that's a shame, because their speakers are extremely convenient to use and Dynaudio as a company clearly can produce state of art speakers.

I wonder why did they invest in Jupiter (which I think is a very fine tool) if they rely on it 30% of the time when designing a speaker? Seems like a rather poor investment to me.

It may be the case that the interviewer being an audio YouTuber may have biased the tone of the discourse.

It is not unknown for engineers to adapt the conversation to the target audience and the majority of audiophiles couldn’t care less about measurements.

And after all, audio is not science but business.

 

[tuga]

You're confusing comparing two products and using only one of them. you need to alternate between the two products quickly, and seamlessly. Otherwise you would never be able to tell which product is subjectively better. Our memory is just not good at all for anything other than visual input.

Try listening to a pair of speakers for a week and then replace it with another pair. Will you not be able the identity differences or even state your preference?

I disagree with your comments.

 

[abdo123]

Try listening to a pair of speakers for a week and then replace it with another pair. Will you not be able the identity differences or even state your preference?

I disagree with your comments.

Unless they're very different I don't think i would be able to form an opinion that i would personally trust.

 

[lherrm]

Try listening to a pair of speakers for a week and then replace it with another pair. Will you not be able the identity differences or even state your preference?

I disagree with your comments.

I guess it depends on how close the two products are.
But indeed, I can clearly state my bluetooth speaker is better than my TV, without having to switch seamlessly, because the differences are obvious.

 

[FeddyLost]

1. Poor vertical directivity is intentional. Apparently when it's that way, there are less reflections from the floor, so the sound on axis is better. This decision is derived directly from their subjective listening "tests", not from the measurements of any kind.
2. According to him, you can't really hear the "anomalies" in the vertical directivity. Well, I don't know about the "anomalies", but I sure can hear that it is extremely narrow.
3. In general, the Dynaudio's (narrow) directivity gives better image, according to him. I assume that in some rooms this can give you less reflections, but seems to be very room-dependent.

They don't make (and never tried to, AFAIK) any coaxials, multiple entry horns and even come vertical arrays with directivity control are only for their high-end line.
So what else shall he say?
Like "we know how to do speakers, good according to some other opinions, but we are doing models from 80ies just because we have a company image and our marketing deprtment doesnt's allow us to"?
I think even accepting something like this is not very comfortable for decent engineer.

Typically, audio companies are just trying to make profits. If their products still sound "right" it's nothing more than a luck.

 

[lowkeyoperations]

Unless they're very different I don't think i would be able to form an opinion that i would personally trust.

Have you ever tried? Or is it just speculation on your part?

Give it a go.

Listening back and forth from one to another is beneficial for assessing difference, especially when those differences are minute.

But listening for extended periods of time is beneficial for assessing preference and enjoyment in my experience.