Evidence-based Speaker Designs

[Bjorn]

Of course it will sound unatural when the room almost become anechoic, which is the case if you were to treat the floor of non-environment room. An non-environmental room is already a very dry and quite dead room and lack the important psychoacoustic cues of other designs.

 

[andreasmaaan]

What about a measurement:

Fig.7 Acoustic Energy AE3, spatially averaged, 1/3-octave response in JA's Santa Fe listening room.

To see how these quasi-anechoic measurements translate to the in-room balance, I perform 10 1/3-octave spectrum analyses of pink noises for left and right speakers individually in a 72" by 20" "window" around the listening position.
Averaging these spectra minimizes the effects of room standing waves and gives a curve reflecting the mix of the speakers' direct sound and the reverberant soundfield in the listening room.
That for the AE3 is shown in fig.7. The useful bass extends to below 35Hz while the generally smooth response trend is broken only by residual steady-state room effects (the peak at 63Hz and dip between 200Hz and 400Hz) and a slight excess of energy centered on 2kHz.
The gentle slope from 500Hz to 10kHz should give the speaker a musically natural tonal balance, broken only by some brightness due to the 2kHz prominence, while the drop in energy above 10kHz again will render the AE3's sound rather too mellow for some tastes.

https://www.stereophile.com/content/acoustic-energy-ae3-loudspeaker-measurements

What does this measurement show in your opinion?

We know from the research that these steady-state in-room curves don't correlate as well to perceived tonal balance as listening-window frequency response.

 

[tuga]

Interesting comments on the subject:

Floor bounce and some terrible ignorance in the audiophool world
I’ve read a few articles, forum posts, and social media posts mentioning Dr. Floyd Toole’s claim that controlling floor bounce in a home audio environment isn’t necessary, Why? Because, according to Dr. Toole, ground reflections are the single most common reflection on the planet, something humans evolved to deal with, recognize, and/or unconsciously compensate for when we are listening to all sounds. After all, throughout all human history we have been standing or sitting on the ground and ground reflections are always present in everything which we hear. Since we instinctively deal with ground reflections without awareness, and since it is 100% completely natural and always present, there is no reason to attempt to address floor reflections in a room designed specifically for critical audio reproduction.

continues here -> http://theaudioannex.com/forum/thre...ible-ignorance-in-the-audiophool-world.13165/

 

[tuga]

What does this measurement show in your opinion?

We know from the research that these steady-state in-room curves don't correlate as well to perceived tonal balance as listening-window frequency response.

What do you mean with "the" research?

Do you think that a short sweep measurement would not show the cancellation? How short before Haas kicks in?

 

[tuga]

The Elephant In The Control Room
Why Only Subwoofers Can Produce Accurate Low End

This frequency–response graph from a monitor system in an award–winning studio prior to optimisation is typical for a pair of monitors mounted on stands behind a console or on a large console meterbridge, though in this case, a somewhat asymmetrical control room is causing the left and right monitor response at around 100Hz to deviate. Note the ‘Grand Canyon’ of missing low–frequency information between the peaks at 125 and 63 Hz in the circled area. This means that monitors tend to deliver ‘one note’ bass, and engineers tend to boost the frequencies in the ‘canyon’ region, leading to bass–heavy mixes.

https://www.soundonsound.com/techniques/elephant-control-room

 

[andreasmaaan]

Do you think that a short sweep measurement would not show the cancellation?

I meant that the frequency response you see on a steady-state measurement is unlikely to correlate closely to perceived tonal balance. I wasn't suggesting that floor-bounce cancellation wouldn't show up in steady-state measurements.

How short before Haas kicks in?

The precedence effect will always occur on-axis to real sound sources in rooms, regardless of delay, since the delayed arrival will necessarily be lower in level than the direct arrival.

What do you mean with "the" research?

I mean the research in which listeners were asked to draw the perceived frequency responses of loudspeakers in rooms. The perceived response correlated much more closely to the speakers' listening-window response than to the in-room response.

EDIT: this research.

 

[pierre]

precisely. and and what we get in a simple discussion such as treating first reflection points will seemingly dwell down into the religious making completely exaggerated claims such as "you'll turn the room into a padded cell!!". it's just immature and childish, like they are "saving you" from making bad decisions with your room - as if you are not smart enough to make these decisions yourself and must consult the bible and not veer from its guidelines.

the "blind spot" has never been more obvious for me with these folks than for discussions about critically-accurate studio models. i will raise the LEDE/RFZ model, which attenuates all early-arriving indirect energy to form the effectively anechoic ISD gap and the religious users will put their blinders on and claim the control room models result in completely dead spaces "with no reflections" - or that the goal is to "kill all reflections", or liken it tantamount to a "padded cell". completely ignorant and oblivious to the fact that LEDE/RFZ has a requirement for a reflection-rich indirect sound-field, albiet delayed in time and managed appropriately. when you show them ETCs of such rooms and the clear proof of a very dense, semi-diffuse, exponentially decaying sound-field (mimicing what develops naturally in large acoustical spaces), they completly block reality out and act as if the room/model is somehow still anechoic and "padded cell". you provide these facts to refute their claim about "killing all reflections", and the next time the conversation arises they repeat the same "padded cell" claims and ignore what was previously presented to them. they simply can't accept it and continue their bias in the name of "science".

and just yesterday i had a guy on this forum telling me "diffusers" are only for professional studios and not home listening spaces and that "studios not being pleasurable to listen to music in". how are you even supposed to respond to that and take anything they state from that point on seriously?



or those that think RT60 is a valid measurement in home, residential-sized rooms. the main guy that posts here was called out for his erroneous understanding of physical sound-fields, and he then went on to write an article about the "reverberant sound field" in his garage and even claimed critical-distance of 1.3ft. yes, a Dc of 1.3ft from his loudspeaker setup in his garage! and he published and bragged about it as if he proved the acoustics community wrong! so much for science.

What would really help are measurements of a set of speakers in a room, after and before room treatments. If the speaker as been reviewed here, even better.

My experience so far as been: studios mostly LEDE with toons of treatments, non // walls and so on, studios which are Dolby certified and residential places with none or very few treatments.

It is not super clear to me which one is "better". They sound different, ok with that but do we have a preference? I get used to any situation quickly.
Except when mastering, I do not care too much.

New speakers with great directivity integrate very well in normal rooms without treatment. Bass is well compensated with subwoofers and you can easily get excellent crossover now. Taming treble is easy too. You can play a bit in the time domain but it is not clear it is easy to detect (and I really would like to believe I hear a difference between my time aligned speakers with XO and a basic EQ).

Please show some figures and people will be more receptive.

 

[ctrl]

That's only part of it and not the really the key here.

Below is an early measurement with a vintage driver (we use a better driver now) of the midbass horn over a reflective surface at different heights and zoomed in on the area where speakers always get huge problems because of the reflective floor. Take note that no EQ was used here. So this is raw measurement.

Vertical overlay over a reflective surface with old vintage driver, 1/12 oct. smoothing and no EQ of the driver/horn:
View attachment 63924

@Bjorn
Could you please tell us how far away from the horn the microphone was? And at what height was the microphone and the horn?

Only then we can estimate the expected frequency and amplitude of the floor bounce.



Perhaps one should keep in mind that there is also a ceiling bounce, two sidewall bounces and front/back bounces as first order reflections. In addition there are "arbitrarily" many higher order bounces with weaker amplitude.

Only as an example if you consider only floor and side reflections for an 8'' chassis in a floorstanding loudspeaker (red = 1m side distance, blue = 0.5m side distance)

Whether the floor bounce really has a dramatic effect should actually be determined by a correctly performed room measurement at the listening position.

 

[QMuse]

What would really help are measurements of a set of speakers in a room, after and before room treatments. If the speaker as been reviewed here, even better.

Please show some figures and people will be more receptive.

I fully agree.

P.S Sorry for OT, but what are these scores on your site based upon?

 

[QMuse]

Whether the floor bounce really has a dramatic effect should actually be determined by a correctly performed room measurement at the listening position.

Can you plz explain in more detail how to measure that?

 

[pierre]

I fully agree.

P.S Sorry for OT, but what are these scores on your site based upon?

View attachment 63968

it is the pref score normalize (the best speaker from the set get 100, they you go down ...). I could change it to pref score, it may be more clear.

 

[QMuse]

it is the pref score normalize (the best speaker from the set get 100, they you go down ...). I could change it to pref score, it may be more clear.

I guess you used the same method to calculate preference score from vendors spinorama charts as the one used by @MZKM?

 

[ctrl]

Can you plz explain in more detail how to measure that?

Something along the lines of these measurements

 

[QMuse]

Something along the lines of these measurements

Aaaa, same good old MMM from LP.

 

[MZKM]

I guess you used the same method to calculate preference score from vendors spinorama charts as the one used by @MZKM?

His scores are a little different from mine, this is because for part of the score we have to segment the frequency range, and besides the # of frequencies in each segment, the documentation doesn’t state where exactly to start/stop.

On Monday I’ll try and contact Sean Olive again, as the last time it was about a different issue (effect of slope).

 

[tuga]

The precedence effect will always occur on-axis to real sound sources in rooms, regardless of delay, since the delayed arrival will necessarily be lower in level than the direct arrival.

What I asked is whether by using a short sweep the measurement will not also show the floor-bounce cancellation?
In my experience it does but I could be doing something not quite right.

The precedence effect will always occur on-axis to real sound sources in rooms, regardless of delay, since the delayed arrival will necessarily be lower in level than the direct arrival.

Smooth off-axis response is important in wide-directivity speakers because of the precedence effect - the listener will perceive the combined balance direct and reflected sound as one.
Why is floor bounce different? Why is a reflection which creates a dip in the perceived balance at th listening spot not objectionable (or downright offensive)?

I mean the research in which listeners were asked to draw the perceived frequency responses of loudspeakers in rooms. The perceived response correlated much more closely to the speakers' listening-window response than to the in-room response.

EDIT: this research.

The paper is not free, but it seems to be dealing with "speech reinforcement". Can you be more specific on its relevance?

It has been established that a high ratio of the direct to the reverberant field is necessary in the listening area for effective speech reinforcement. However, equalization networks which employ shaping of the frequency characteristic of the reverberant field to increase gain before howlback also distort the direct field. The effect of such distortion on perception is examined by talker identification methods, using tape-synthesized combinations of the fields, with one field containing the original talker and the other field containing a frequency-distorted image of the original talker. Initial data in the ongoing study show a 70% choice of the direct field as the spectrum determinant suggesting the need to avoid distorting this field.

 

[pierre]

His scores are a little different from mine, this is because for part of the score we have to segment the frequency range, and besides the # of frequencies in each segment, the documentation doesn’t state where exactly to start/stop.

On Monday I’ll try and contact Sean Olive again, as the last time it was about a different issue (effect of slope).

If Sean Olive answers you, what would be great is if he could release the raw data that leads to this pref score methodology.

 

[MZKM]

what would be great is if he could release the raw data that leads to this pref score methodology.

You mean the headphone measurements he used as well as the user ratings?
I would want that too, but that is very, very highly unlikely. I’ll ask him after/if he answers the 1/2 octave band question.

 

[QMuse]

Something along the lines of these measurements

And how would you recognise dip is coming from floor bounce?

 

[tuga]

And how would you recognise dip is coming from floor bounce?

Isn't the purpose of ground-plane speaker measurements to avoid floor-bounce cancellation?