• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Vanatoo Transparent Zero Speaker Review

HooStat

Addicted to Fun and Learning
Joined
May 11, 2020
Messages
856
Likes
933
Location
Calabasas, CA
I do show that radar chart

I think that is probably the most important measure -- more relevant than the score. I didn't realize that is what the radar plot reflected.

I read through Olive's patent application. I can't believe anybody got a patent for what is a garden variety statistical analysis, but good for him and Harmon. I think the real value is the experiments themselves and the approach to summarizing complex functions over frequency. The method is just basic statistics. But it was really informative. Assuming that is the documentation for the analyses on this site, it explains a lot.

The one thing I noted was that there was 1 measured preference score that was above 8 and 1 calculated preference score above 7. So calculated preference scores above about 7 are going to have a lot of variance (more than scores in the middle of the range). Plus, we run into ceiling effects (i.e., no measured score can be above 10) and I assume things get non-linear near the tops of the range. By non-linear, I mean we hit the point of diminishing returns for small improvements in any one of the model's inputs. All of this is to say that any "high" calculated score is very uncertain. The things that matter at/above 8 might be very different than what is measured for the model (or maybe not -- we have no data to know for sure since there is no data). Most likely, there are some additional variables that might help discriminate the very high scores. I wouldn't expect to need a completely different model.

Just to be clear, I am not criticizing any of Olive's work or your hard work in putting this all together. I am just trying to lay out some of the limitations of using this kind of model to make inferences about subjective preferences so people don't get too carried away with the different preference scores with EQ.
 

Sancus

Major Contributor
Forum Donor
Joined
Nov 30, 2018
Messages
2,923
Likes
7,616
Location
Canada
The one thing I noted was that there was 1 measured preference score that was above 8 and 1 calculated preference score above 7. So calculated preference scores above about 7 are going to have a lot of variance (more than scores in the middle of the range).

This is a good point. Perhaps scores above 7 or 8 are really just not meaningfully distinguishable from eachother.

It's based on this study. Of the ~70-75(there are 75 points on the graph in question on pg11, but the study references 70) there are 5 measured above 7, and only one predicted to be above 7. No data table is provided, but I scanned the graph and put the data in this post some time ago.
 

HooStat

Addicted to Fun and Learning
Joined
May 11, 2020
Messages
856
Likes
933
Location
Calabasas, CA
Perhaps scores above 7 or 8 are really just not meaningfully distinguishable from eachother.

One of the key assumptions of the method (implicit in the use of linear regression) is that each unit change is exactly the same (i.e., 4-5 is the same as 9-10). That is generally not true. There is a whole literature on how to create scores to reflect things that are subjective and bounded (for example, quality of life scores on a 0-100 scale) and the Olive score does none of them. Again, given how hard it is to do what they did, they did a great job. But the model's ability to project to new speakers and DSP-enabled speakers is limited. Particularly when the calculated scores start to get high.

One other thing I noted was that one of the preference score components is based on the deviation from a standard slope which was calculated from the 90th percentile of measured preference score. That measure HAS TO BE significant in the model because that component of the model was defined based on the the results for the top speakers (i.e., it has a circular logic to it). And, since those top speakers were at the extreme (far right in his graph), they have a lot of influence on the fit of the the regression model.

OK. I am getting off my soapbox. Apologies for going too far off topic here.
 

wadec22

Senior Member
Joined
Oct 9, 2018
Messages
319
Likes
242
couple of things that stood out to me;

  • testing via the 3.5mm will not give you best performance. the internal DAC cannot be bypassed. so you are sending it through an ADC, then the DAC. odd design choice, but I assume necessary with all their tuning adjustments
  • comparisons to the JBL speakers is fine, but keep in mind these stand at something like 7" at their tallest point. heck you can slide them forward and the height continues to go down. They are a great solution where one does not have space for a larger monitor, frankly, that is why I chose them. Great flexibility in placement.
  • I'm not sure if these have seen revisions, or if the 3.5mm is affecting things or if the tweeter is testing that off as the manufacturer stated but it seems very weird. mine came so veiled and soft in the treble, I actually called the company, as I was considering returning them. they walked me through the dip switches and got them to a nice clean output with me over the phone. I had been to lazy to read the manual in regards to the switches and had I relied solely on the remote without understanding the switches as well, they would have gone back.
  • someone mentioned a $100 mic for testing and EQ. fun for us tinkerers, but again, these are loaded with DSP options to adjust the sound. combine that with the handy remote - they are desktop/computer speakers and very easy to adjust by ear to your personal preference. I think the mic and and some software are a great investment for those who turn and change out gear or are setting up a serious listening environment. I think that really goes beyond the use-case for these, but maybe I am wrong.

just some random feedback from someone who has owned them for about a year and half. I am treble sensitive and I find them nice and clear and lacking in bass. Considering size and placement, I kind of expected that without adding a sub.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,370
Likes
234,430
Location
Seattle Area
testing via the 3.5mm will not give you best performance. the internal DAC cannot be bypassed. so you are sending it through an ADC, then the DAC. odd design choice, but I assume necessary with all their tuning adjustments
In the context of measuring frequency response of a speaker, such factors are totally imaterial. An ADC will never have this type of frequency response.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,370
Likes
234,430
Location
Seattle Area
I'm not sure if these have seen revisions, or if the 3.5mm is affecting things or if the tweeter is testing that off as the manufacturer stated but it seems very weird.
I ran my measurements by the company and it matched theirs that were made prior to sending the sample to me. There are variations in tweeters causing the differences that different people see.
 

wadec22

Senior Member
Joined
Oct 9, 2018
Messages
319
Likes
242
In the context of measuring frequency response of a speaker, such factors are totally imaterial. An ADC will never have this type of frequency response.

I did not mention FR. I simply shared a tip for best performance. Simply recommend users not pass through ADC, then DAC if they have an option.

It does seem the oddest choice for any scientific testing, but I understand what you are saying.
 

wadec22

Senior Member
Joined
Oct 9, 2018
Messages
319
Likes
242
I ran my measurements by the company and it matched theirs that were made prior to sending the sample to me. There are variations in tweeters causing the differences that different people see.

I had read that. Just shared I found it weird since mine were so dark when I received them. Such a broad range in variance...
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,370
Likes
234,430
Location
Seattle Area
It does seem the oddest choice for any scientific testing, but I understand what you are saying.
We are not performing scientific testing/research. We are making engineering measurements. And there, we worry about what things impact the measurement at hand. The ADC is not one of them.
 

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,785
So what is the latest EQ/convolution you want me to try with this speaker? And 305P MK II? Can't keep track of the back and forth. :)

I would be interested to hear your opinion on this filter. As the previous one you liked it is also not optimised for best score but to my perception of what kind of tonal balance you prefer. :)

EDIT: this filter requires 1dB of attenuation to avoid clipping
 

Attachments

  • filter.zip
    2.5 KB · Views: 103
Last edited:

ttimer

Active Member
Joined
May 28, 2020
Messages
116
Likes
161
As others have already hinted at: This speaker does have in-built DSP and EQ but doesn't use them to correct the major flaws in FR. Then what is the point? The big thing about active speakers with DSP was supposed to be their ability to be EQed flat at the factory.
But apart from Genelec (and Neumann?), hardly anyone seems to use this option. It seems that currently, most active+dsp speakers are no better than any old passive with a D-amp strapped to the backside.
 

Tks

Major Contributor
Joined
Apr 1, 2019
Messages
3,221
Likes
5,494
In the context of measuring frequency response of a speaker, such factors are totally imaterial. An ADC will never have this type of frequency response.

I'd have to push back on this. Sure the FR perhaps, though the THD metrics could be widely varrying, I'm sure you would agree seeing the poor state of even dedicated ADC's, let alone ones in mixed devices or powered speakers.
 

edechamps

Addicted to Fun and Learning
Forum Donor
Joined
Nov 21, 2018
Messages
910
Likes
3,620
Location
London, United Kingdom
I see no reason why preference score wouldn't be valid after EQ as it is based on the same curves that EQ affected.

I've said this before, but I'd be really careful with statement like these. As far as we know, Olive's test sample did not include any speaker that are EQ'd, or at least, not EQ'd using this particular approach. The model was trained on "standard" speakers with no additional EQ. So we're back to the whole "correlation vs causation" question and the dangers of extrapolating from the model.

Can someone point me to the source for this model? I would like to see the model and the coefficients.

In Loudspeaker Explorer (unfold the various sections under "Olive Preference Rating") I have detailed the model, including links to the original papers, a clearer description of the formulas, and many links to various debates that were had on ASR regarding the finer points of the model and controversies around the definitions. I would suggest reading through that to quickly get up to speed on discussions.

Just out of curiosity, do you ever provide the model inputs? Or just the model outputs only.

The input to score calculations is the frequency response data that @amirm provides in a zipfile attached to every speaker review. If that's what you're asking.

One other thing I noted was that one of the preference score components is based on the deviation from a standard slope which was calculated from the 90th percentile of measured preference score.

No, it's not. The target slope is only used to compute SL. SL did not make it into the final Olive model. Therefore the target slope is not used in the final model at all and is for illustrative purposes only. This is a common source of confusion. (In practice the final model scores are quite consistent with the target slopes, but that's somewhat coincidental.)

That measure HAS TO BE significant in the model because that component of the model was defined based on the the results for the top speakers (i.e., it has a circular logic to it).

I don't think it's circular logic because, in this context, I believe "top speakers" means the speakers that obtained the best actual ratings, not the best predicted ratings.
 

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,785
I've said this before, but I'd be really careful with statement like these. As far as we know, Olive's test sample did not include any speaker that are EQ'd, or at least, not EQ'd using this particular approach. The model was trained on "standard" speakers with no additional EQ. So we're back to the whole "correlation vs causation" question and the dangers of extrapolating from the model.

From what I understood Olive's model hasn't been properly verified at all as for that he would need to test it against different group of speakers than the ones he used to develop the model.

Btw, I am not optimising my EQ for the maximum score but to what I think would sound the best.
 

wadec22

Senior Member
Joined
Oct 9, 2018
Messages
319
Likes
242
We are not performing scientific testing/research. We are making engineering measurements. And there, we worry about what things impact the measurement at hand. The ADC is not one of them.

I fail to see how engineering measurements are not scientific. the title of the website would certainly lead any new web surfer reading their first review to believe that was your intention.

I am not saying the ADC did affect your measurements. I am saying it is the input with the most variables. Without fully testing each input, it leaves the most potential for unknown influence on the measurement. Given the other, easily accessible inputs, I stand by calling it an odd decision.
 

wadec22

Senior Member
Joined
Oct 9, 2018
Messages
319
Likes
242
As others have already hinted at: This speaker does have in-built DSP and EQ but doesn't use them to correct the major flaws in FR. Then what is the point?.

How do you know it can't use them to correct the flaws in this measurement? They were not tested here.
 

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,785
How do you know it can't use them to correct the flaws in this measurement? They were not tested here.

Owner's manual states PEQ filters configuration is not supported.
 

wadec22

Senior Member
Joined
Oct 9, 2018
Messages
319
Likes
242
Owner's manual states PEQ filters configuration is not supported.

Yes, but manipulating the DIP switches and the remote settings can give you vastly different measurements from what is seen here, so one of those settings may not have those same flaws.
 

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,785
Yes, but manipulating the DIP switches and the remote settings can give you vastly different measurements from what is seen here, so one of those settings may not have those same flaws.

You don't ever measure all possible combinations of shelf filters etc, what you measure is when they are on flat/neutral settings.
 

ttimer

Active Member
Joined
May 28, 2020
Messages
116
Likes
161
Yes, but manipulating the DIP switches and the remote settings can give you vastly different measurements from what is seen here, so one of those settings may not have those same flaws.

Even if that is the case, whats the point? If i have to do my own measurements and experiment with seetings which "might" result in a less flawed frequency response, that is not much different from applying my own EQ corrections to any random passive.
In order to realize the theoretical benefits of active + dsp speakers, the FR needs to be flattened by default DSP settings applied at the factory and verified by the manucturer with his measurements and knowledge of drivers and crossovers.
 
Top Bottom