A Little About JBL

[tomp]

Again from an AI search -

Floyd Toole's research on expectation bias reveals that listeners' judgments of audio quality are heavily influenced by factors other than sound, such as brand reputation, price, appearance, and suggestion from a salesperson
. His experiments, often using double-blind listening tests, demonstrated that when these biases are removed, people show a remarkable similarity in what they prefer: a neutral, uncolored, and accurate reproduction of sound.
Key findings on expectation bias

Non-audio cues matter: Toole's work demonstrated that elements external to the sound itself can change a person's perception of audio quality. For example, a listener may rate a speaker more highly because of its price or expensive-looking finish, even when they cannot distinguish it from a less-expensive speaker in a blind test.

The "sight over sound" syndrome: In audio evaluations, visual cues and pre-existing beliefs can override what a listener actually hears. When listeners were given a chance to compare speakers blindly, their "personal preference" disappeared, and most gravitated toward the same objectively accurate speakers.

Double-blind testing reveals true preferences: To combat expectation bias, Toole and his colleagues at Harman pioneered rigorous double-blind listening tests. In these tests, neither the listener nor the experimenter knows which product is being evaluated. This methodology ensures that the listener's ratings are based purely on what they hear, rather than any pre-existing biases.

Correlation between measurement and preference: Using these controlled tests, Toole confirmed a strong correlation between objective measurements and subjective listener preference. The loudspeakers that consistently ranked highest in blind tests were those with the flattest, smoothest on- and off-axis frequency response.

Individual differences are minor: Toole's research found that when the biasing influences of brand, price, and appearance are removed, individual listening preferences are more similar than the audio industry had previously assumed.

The problem with Floyd's double-blind testing is, it gave the listeners only at 15-30 second snippet of music on each speaker. When was the last time you walked into an audio showroom and auditioned a speaker for 15secons?

The rationale behind the 15second listening span is the fact that we only remember short passages of music. But in reality, we "experience" the music which requires lot longer exposure to music.

The human brain works best with A to B comparisons of any type when the time between them is short. When we were comparing photographic prints we tested this by giving photos made with different lenses with a wait time between presenting them to the observer. Then we tried putting the prints side by side and the ability to pick out small differences increased dramatically. Remember with double blind audio tests the goal is not to determine whether A or B is "better", only if there is a statistical difference in the identification of the two. I would agree that longer term listening is needed to determine which sound is more pleasing to you. However, if double blind testing shows no perceivable difference in the sound, how can you make a choice as to which one you like. With me, my choice of how I react to different music or intensity levels varies with my mood at the particular moment.

[Guy]

There is a fix (maybe) online which I have done (I am still waiting on a Buckeye Amp to install in my home theater). I will let you know what happens.

Thanks, Brombo, I'm interested.

[Cogito]

The human brain works best with A to B comparisons of any type when the time between them is short. When we were comparing photographic prints we tested this by giving photos made with different lenses with a wait time between presenting them to the observer. Then we tried putting the prints side by side and the ability to pick out small differences increased dramatically. Remember with double blind audio tests the goal is not to determine whether A or B is "better", only if there is a statistical difference in the identification of the two. I would agree that longer term listening is needed to determine which sound is more pleasing to you. However, if double blind testing shows no perceivable difference in the sound, how can you make a choice as to which one you like. With me, my choice of how I react to different music or intensity levels varies with my mood at the particular moment.

Everything has its role, its place and time.

You are categorical that dspNexus is superior product. Have you done any A-B test, double blind testing with 15 second music segments? I bet you did not.

When you judge DIY speakers, do you, use A-B, double blind testing with 15 second music segements? Again, I bet you did not.

Does it mean you are BS'ing us?
Absolutely not.

You are behaving like us "regular" people, properly audition the equipment to form your opinions and sharing them.

[tomp]

I have said that it is my belief based on over 60 years of audio experience that the dspNexus is the best DSP processor I have owned or auditioned. I can't say it is the best overall because I have not auditioned every one available. I have also done very many double blind tests and have a QSC ABX controller. I even made a special large display so that the box could be mounted near the equipment being tested and far from the participants because I found that in quiet passages, I could distinguish the sound of the different relays for the separate channels during activation. I also made a sound absorbing box to house to further reduce the sound of the relays. I'm enclosing a picture.

Doing any double blind test properly is a very difficult task with the simplest of audio components. Doing it with DSP processors with 8 channels in play is pretty much a non starter. In addition to all the connections that are more than any ABX box I have ever seen, switching digital signals at frequencies well beyond analog frequencies is not a trivial task. In addition to the equipment itself, you have to deal with output filters for the DACs, some of which may not be available in all the DACs. I'm including a list of the filters in the AK4191 section of the 8 channels of the DACs. I am using the low dispersion short delay filter. Doing a double blind test of speakers , although requiring great attention to detail, is in comparison relatively simple and I have done it many times with short segments of the music, often repeated if there is something in the segment of the track that is particularly revealing.

So I believe that comparing ABX testing of speakers to DSP processors is definitely comparing apples to oranges. In theory, the best solution to determine differences in any components is double blind testing. To repeat an old saying I once saw on the wall of an engineering department where I worked, "One of the great tragedies of modern science is that every day beautiful theory is shattered by ugly old reality". I can tell you that I did a double blind test of USB cables, and out of 5 participants, only one statistically identified the different cables. It was one of the most difficult tests I have done because in order for it to be truly double blind, not having any device that could handle the USB frequencies involved, the equipment had to be located out of sight of the participants and I had to manually switch the cables between each portion of the test. The participants wanted to familiarize themselves with the cables requiring several switches for each trial before the unknown part of the trial was conducted. It was very demanding.

To sum up, you are correct in my opinion about the Nexus, but it is an opinion, not the same kind of certainty I had with double blind testing of speakers in the past. So don't mix apples and oranges in an attempt to prove a point.

Attachments

AKM AK4191 filters.jpg (11.14 KiB) Viewed 101 times

[brombo]

To stir the pot a little more:

"If it can't be expressed in figures, it is not science; it is opinion"
―Robert Heinlein

“When you can measure what you are speaking about, and express it in numbers,
you know something about it, when you cannot express it in numbers, your knowledge
is of a meager and unsatisfactory kind; it may be the beginning of knowledge, but
you have scarcely, in your thoughts advanced to the stage of science.”
― Lord Kelvin

[Cogito]

Brombo,

There are lots of things that cannot be measured, like

life
consciousness
soul
taste
smell
how our brains perceive sounds and visuals
...
...

[tomp]

When designing and building a speaker, I always start with measurements because otherwise you are "wandering in the desert". However, I have never built a speaker based solely on measurements that was to my liking. I can only guess that my very precise ACO Pacific measurement mike does not act like my ears and my computer does not act like my brain. At the end of the day, after measurements to get you in the ball park, you need to listen and make adjustments to get the resulting sound to be pleasing to you. Note that I didn't say flat, but pleasing. Manufacturers do not have that luxury because they are aiming for lots of people who may have different tastes so they either go for flat response or some other setup that they thing a large part of the customers will relate to.

[Pelliott321]

I think A/B tests are very flawed. Anything that is processed by our brains is prejudiced by all past experiences. If the same track is played over your system without you knowing whether something is actually changed or not, you will perceive a difference. Humans make terrible witnesses because of this.
Scientific measurements are the only way to really see a change. The problem is measuring instruments are incapable of capturing the emotional part, which we humans want.
We all know that components sound different. The question is....is the difference BETTER. Will one component give a better emotional effect than another?

[SoundMods]

I have said that it is my belief based on over 60 years of audio experience . . .

OK. It is tiring when someone has to verbalize their resume bragging about all the many years experience.

Is it 60-years experience -or- is it 60-years of the same experience?

There are those that know what they don't know -and- those that don't know what they don't know (aka. self-proclaimed "experts")

And my experience has been that I made a lot of money (engineering forensics) off the backs of those those that don't know what they don't know.

[HAL]

Since my name and demo have been mentioned, might as well stir the pot!

The CAF2024 HAL demo of the Magnepan LRS speakers with the 1x12 OB subs as stands was to show what the original speaker series mode passive crossover with DSP EQ can do along with integrating the OB subs with high rate Linkwitz-Riley crossover was to give and idea of what is possible with the Danville Signal dspNexus 2/8 system with AK4493SEQ DAC's running DSP Concepts Audio Weaver development system.

The room had front wall diffusion and side wall absorption to help control first reflections. The floor has a rug and no ceiling changes. I always find it best to start with correcting the room acoustics with material before the final measurements and DSP EQ created.

I measured the in room center listening chair position with a Beyerdynamic MM-1 mic with the 90 degree correction file used in Room EQ Wizard. The EQ filters were calculated from the minimum phase results. Those parameters were put into the Audio Weaver EQ processing before the crossover driving the LRS and OB Subs. I EQ'd them for flat on-axis response at the center chair. I could have as easily EQ'd the system for the Harman response curve for listening area. Both the acoustic levels were matched at 500Hz.

The LRS have with the standard passive crossover a large boost in the midrange that tends to cover the HF end of the frequency spectrum. Also they are dropping off below 50Hz quickly and have limited power handling. One way to help that is to use a high pass filter at 100Hz to greatly reduce the first octave power requirements. That is how I added in the OB sub when used. Otherwise the speakers were run full range. By EQing them flat it sounded much clearer and more even across the frequency spectrum in that room.

The way I look at this is the speaker and room distorts the acoustic signal frequency response to the listener due to things like room modes up to one octave above the Schroeder room frequency. Below this the room owns the system response and needs EQ after the speaker system and chair are setup with some room measurements and modeling that REW, dspNexus with the mic act as the measurement system with a PC.

For all the customers that have the dspNexus 2/8 that I have helped to integrate with many styles of speakers and subs, all have been very pleased with the results. All saying it is the best sound they have ever had in their room. It has been used from simple stock speakers to add subs to full multiamp direct drive speaker crossover and sub intergation, with planars, cone and horn drivers.

Another point of this demo was to show the capability of a new Danville Signal entry level product being developed that is a 2x4 DSP, with AKM ADC and DAC system with digital inputs. This was shown as a prototype at the back of the room where it uses an eARC HDMI input, but the other version would use standard analog and digital inputs to allow easy integration with DSP crossover, time delay for time alignment and PEQ for speaker systems. The dspNexus 2/8 only had 4 channels in use with the same DAC's as the new unit once complete. It will not come with an Audio Weaver license, but be programmable from the front panel. REW can still be used to do the room EQ filters.

Hopefully that sheds light on the CAF2024 room demo and purposes.