Typo, LF information, which is the important one.

Explained or noted would suffice. Explicitly noted is a tad hyperbolic.
But I don't see a rationale here, we use Auto Gating all the time when using Frequency Response Graphs. Nobody else seems to have a problem with that.

Nope, I explicitly stated that the specific research I was referring to was specific to professional mixing engineers.

Clearly I have never suggested that particular controls supplied in REW and FM etc. are useless.
What I have said is that Auto is the norm in common usage. A Lingua Franca.
e.g. The OP presented an auto gated FR, which is a norm. That is what we worked with and solved the puzzle. i.e. The usual two speaker HF conflict thing.

DD

You keep changing your mind on this. First, in post #18 you state that there is no reason to gate a response:

Then, in post #29, you change your tune and see how gating might be useful:

And now, in your most recent post #31 you go back to your original claim:

I hate to be this tedious and it really shouldn't be necessary to have to quote you in posts you made literally hours ago, but you have an uncanny ability to say one thing and then immediately say the opposite (in spite of the fact that your original statement is recorded in the previous post!) .

As before, you are specifically getting things backwards. Floyd Toole explains that professional musicians and mixing engineers are more sensitive to the room sound than the untrained listener (in other words, they are less able to "listen through" a room's sound, since they are so sensitive to reflections):



You didn't use the word "useless" but you did say that it wasn't "useful" to gate a response. Once again:

Why all this tedious back and forth, saying one thing and then the opposite in the next post, and denying you said things that are right before our eyes, Dan? Why not just do the normal thing and say "Oh, I don't understand gating or how/why it is used, please explain more", and have a normal conversation about it?

As I said before, I'm not convinced of this. I don't see how playing two speakers at the same time will cause a broad dip in FR from 3k-15k. The typical case would be to see a comb filtered response in that range. Probably the graph is overly smoothed and we are not seeing what is actually going on. I think the most likely explanation is still attenuation of high frequencies in the room due to thin absorption and/or furnishings (carpet, curtains, etc.). Gating the IR to see the frequency response of the direct sound above approximately 3kHz would allow us to see if this is in fact the case.

I think if you accept evidence that it actually happens then the reasons why may come to you.
https://www.gearslutz.com/board/4217370-post13.html


DD

The delay will decorrelate the L & R signals above frequencies with wavelengths shorter than the (unintentional) delay, resulting in 3 dB summing instead of 6 dB summing that happens below. The transition is of course gradual and you'll see it as a smooth HF rolloff (towards -3 dB) in the smoothed response.

Signals don't get decorrelated because of a delay but rather the phase relationships change. You get nulls and bumps in the frequency response (i.e. comb filtering) due to the difference in path length, not decorrelation and a shelf filter.

As far as average (aka smoothed) frequency response is concerned, they do get decorrelated once you have enough nulls compared to averaging bandwidth (at which point it doesn't matter if the nulls are distributed linearly or randomly per Hz). When the delay is very short, the first peak covers the lows and mids (meaning they are boosted 6 dB). By the first null, the distance between nulls (in Hz) is much less than the width of the averaging, resulting in a shelving filter. Another way to look at it is that the nulls cancel some of the energy once they start happening.

Obviously. The first image is a comparison of single vs two speakers running. The second is the OP with two speakers running.
They are very similar. The OP subsequently tested with single speaker drive. DD

Here are some quotes from Toole:


” In professional audio outside of the movie-sound domain the traditional pink-noise/real-time analyzer (RTA) process of measuring steady-state amplitude response has been superseded. Sound reinforcement system design and setup now routinely includes time-windowed measurements that can capture the direct sound, with selectable amounts of early and late reflected sounds. Generally the goal is well described as a variation on flat and smooth direct sound.”

“The author’s book [1] (chapters 2 and 18) illustrates the past and present situation in consumer and professional-monitor loudspeakers. Flat on-axis frequency response is clearly the engineering objective for most of these systems. Those that deviate significantly earn lower ratings in double-blind subjective evaluations. Although there is more to be considered, a flat direct sound delivered to listeners is the basis for most reproduced sound.”

“From the beginning, loudspeakers intended for sound reproduction have been designed with the goal of a flat onaxis frequency response so that the first sound to arrive is an accurate representation of the spectrum of the recorded sound. Double-blind subjective evaluations of loudspeakers conducted by the author and his colleagues for 35 years have shown consistent preference for those having flat and smooth on-axis frequency response, accompanied by wellbehaved off-axis response—i.e., a smoothly changing or constant directivity index ([1] chapter 18). Changing the room has little effect on this observation ([1] section 11.3.1). Based on this, a generalizable performance standard for sound reproduction systems would require a loudspeaker that radiates a flat, smooth, direct sound, and that has gradually changing or constant directivity.”

” A flat, timbrally neutral, direct sound appears to be the logical objective.”

“The target for the direct sound should most likely be flat: neutral.”

https://www.google.se/url?sa=t&rct=j...wwHEmQ&cad=rja

An Acoustic Basis for the Harman Listener Target Curve | InnerFidelity

DD

A flat, timbrally neutral, direct sound appears to be the logical objective.”

“The target for the direct sound should most likely be flat: neutral.”

the ISO 226:20031. curve

https://fr.wikipedia.org/wiki/Courbe...27isosonie.png


Flat and neutral are an oxymoron. Wtih a calibration around 110 db, why not.

With a Mixing Desk 20M or more from the speakers, how would one capture the 'Direct Sound, eliminating the hall contribution" ? And even if it were possible, it would obviously misrepresent what the audience or engineer actually hear, i.e. a sum of speaker, hall, backline, monitors, and drums. Speaking as a Professional Live Sound Engineer since the mid 70's, I think I can speak for many of us in that our goal is to present the best possible sound to as much of the audience as possible.
We use a combination of system and individual channel Eqs.
A so called 'flat' system response, as heard at an audience or mix position, would make life extremely difficult. In reality the summed or averaged responses we see on measuring gear in situ, is again HF rolled off, B&K like, or with a bit more LF just for fun.

DD

JBL lsr6332

http://www.jblpro.com/ProductAttachm...BL.LSR6332.pdf

Flat or not flat ?

What we hear, on axis or off axis ?

What we hear, direct sound, mix between direct sound and the room sound (Badly chosen expression, sorry) ?

Interesting specs, I would love to hear those speakers. Are JBL part of Harman....?
I have noticed that large PMC's have an actual rising HF response on axis.
The recommended location is for the tweeter to be quite high in elevation, I seem to remember 10-11 feet. Plus very little toe in.
The idea is that the listeners over a broader area are never on axis.
I found the IB1S too bright in my little dead room. I asked PMC about solutions. They said most of their IBIS customers also bought the passive 'film' filter. Or used speaker controlling boxes.


The expression is totally fine dino, in a typical real world listening scenario we hear a mixture or sum of all those sources mentioned.
Logically, to translate our mixes into such scenarios, we need to listen in them or something similar, or something with a similar balance.
Note the BBC recommendations of 400mS to match the average listeners' sitting room.
With 'progress' and perhaps faster and more impulsive music, the trend now is towards shorter room decay. 200mS EBU. Newell and TJ go below this and go spectrum, in one direction. This is way way different from real world listeners' transfer functions.
This is the elephant in the room, with very divergent views. Toole seems at odds with Olive.
As best I can tell the 'flat' cohort don't seem informed by actual experience by themselves or professional practitioners. It seems a theory driven viewpoint, elegant in it's simplicity. i.e. No colouration = perfect linear response.
But that completely ignores the fact that listeners........
Except for those in high quality headphones..... which it turns out have HF rolled off too as the last Harman/Olive link shows.
Plus of course these HeadFans are missing 3dB of Vocal, Snare, Kick, and Bass......

So yes indeed, what who where?
For some of us this is not a pondering matter. We have to deliver product which sounds similar to as many listeners as possible.
Acousticians would do well to heed our experiences and preferences.
I talk to myself quite a lot....;-)

DD

Did you read the Toole paper I linked to?

Numerous studies has shown that most listeners prefer a flat direct sound (as far as the higher frequency range is concerned, above 1 kHz or so). How this effects the response curve in a specific room if looking at the steady state response (ungated impuls response) will depend completely on the room, and often a HF tilt will appere. What I'm trying to communicate is that it's important to understand why that is, and instead focus on the direct sound (that most listeners will want flat) if doing adjustments in the higher frequency range.

Just a note that way back when I did a two year stint as a researcher (Helsinki University of Technology Acoustics lab), the method was to always conduct a formal listening test with multiple people or reference published literature before making claims about such preferences. Given Toole's academic background, I doubt he's making claims purely based on theoretical approach or personal preference.

Indeed that is the academic normal approach. But the Titanic was built by professionals wasn't it, while the Ark......;-)
Just joshing, there are many many academic and other testing sources which confirm the usefulness of HF roll off in a Zone Without Early Reflections, in order to duplicate the Frequency Response experienced in a more normal room. e.g. Bruel&Kjaer, Harman research by both Olive and Toole. http://www.aes.org/tmpFiles/elib/20170509/17839.pdf
I have never come across an actual working engineer who uses a Flat FR.
One might remember that Bob Clearmountain placed toilet paper over the tweeters of his NS10Ms.
Bob Hodas-I will say it again, people who actually make records have experienced preferences, trained ears, and valid reasons for the preference. i.e. the B&K family of curves works in terms of translation, and is actually the norm. Everything else doesn't sound right. The fact that this concurs with the frequency response that the average listener actually hears is hardly just luck.

I hope that those who want to understand why a seemingly tilted HF response might actually be "flat" if gated properly, reads the paper linked to in post 38, and also the sources on how the brain "hears" that mpos provided in post 21.

From the Bob McCarthy quote, "The first perception zone is the tonal fusion zone, which runs for the first 20–30 ms..."

If this means what I think it means (sound within the first 20-30 ms is integrated and provides the subjective perception of frequency response), then I'm not entirely convinced that the subjective and ungated response would actually differ significantly apart from the modal bass regions. My reasoning is that that's enough time to incorporate the first and maybe second reflections, and those with the direct sound are what's providing the majority of the total received energy (apart from the said modal region). So the windowed (direct + er) and total (direct + er + late reverb) response aren't going to differ much (a decibel or two).

Yup. Auto 300mS or so vs short.

DD

As usual, you have posted a link which states the direct opposite of what you are claiming. The article is refers to a 1dB per octave "room gain". I.e. the in-room response when using "high-quality speakers measuring flat in an anechoic chamber". In other words, a flat frequency response of the direct sound (equivalent to a flat response within an anechoic chamber) becomes a 1dB/octave in-room response.

This is exactly what Jens was referring to, and is what Floyd Toole was referring to in the quotes that you are having a hard time understanding.

Again, to quote Floyd Toole:

The difficulties you're having here, including the unwarranted assumption that Toole and Olive are contradicting each other, goes back to your lack of understanding of windowing and what it's used for. As I've said before, I'd recommend looking at the REW manual for you to grasp the basics of the topic. (Start with page 14.)

If you read the whole section of the book, he is very clear that there is a distinction between the way the brain processes the direct sound and the sound that arrives after it (both early reflections and later reflections). For example, the start of the section (pg. 158):

You will see the same distinction made in any book or article on psychoacoustics.

In virtually every case, the direct sound dominates the overall perception of the sound. This is central to the Haas effect as well as to understanding why similar in-room frequency responses can sound very different if the direct sound frequency responses are very different. A discussion of why this is the case is also the larger context of the Philip Newell quote I provided.

It is flat. And this should give us a clue (in addition to all of their very clear statements saying the same thing) that the engineers at Harman are proposing a flat direct frequency response (i.e. flat anechoic frequency response), which translates into a 1dB per octave rise towards the low frequencies in the room frequency response.

i'm ok with that but the gating is rarely quoted when people use the word flat Hence the lack of understanding.

Agreed.

Not easy for a novice to understand that we gate the impulse to study it properly.

Extract from the service of A(rt) Audio Science - Floyd Toole


Measurements that give us a very good prediction of how this loudspeaker will sound in a room.
In a large anechoic chamber, at a distance of 2 m, a total of 70 frequency-response measurements were made at 10° intervals on horizontal and vertical orbits. The data were processed to yield these curves.
The on-axis curve describes the direct sound, the first sound to arrive at the ears of somebody seated in the ‘sweet spot’.
The listening window describes the average direct sound for listeners seated or standing within a ±10º vertical, ±30º horizontal region directly in front of
the loudspeaker – the entire audience in a home theater, for example.
The next curve describes the sound of the average strong early reflections from the room boundaries, and the sound power is a measure of the total sound output of the loudspeaker without regard to direction. The bottom two curves describe the directivity (DI = Directivity Index) of the
loudspeaker: how uniformly it radiates its sound into the room at all frequencies.
The top DI curve is an adaptation of the ‘classic’ one, based on a comparison of listening window measurement and sound power.
The lower DI curve is based on the ‘early reflections’ curve.
In a truly good sounding loudspeaker, the on-axis and listening window curves will be smooth and flat, and the other curves will be smoothly and gradually changing.

Yes. Thus a gated impulse of the speakers will show a flat HF response curve. The stady state response (un-gated or a longer gate setting to include reflections from the room) will look different. Instead of trying to match the total system response curve to a tilted target curve; we (who do this fot a living) gate the impulse and make sure that the direct sound measures flat in the higher frequency range.

And how define the good gating ?

Just gate to the first strong reflection. In a good room; this is not a problem since the first 12 or more ms are clean (-20 dB or lower after direct). In a problematic room, it might be trickier if there are strong early reflections present.

"In a problematic room, it might be trickier if there are strong early reflections present"

I reach only -15 db

thank you.

Most, probably every one, of the Frequency Response graphs presented here by GS members interested in their in room speaker response use Auto Gating. Simple measurements of what the microphone/listener is receiving using software operated as designed.
Anyone here with a measurement of their own room can adjust the gate, as I posted, and see what happens or more importantly what does not.
Most, probably ever one of us has access to Anechoic and various other measurements by the speaker manufacturer.
They strive for Flat or on axis HF rising Anechoic response.
When deployed in a well treated room, the HF will remain similar in the Direct Field.
This is very skewed away from what real listeners with speakers experiencing room gain actually hear.
In my experience, without two exceptions in well tuned rooms, for both myself and/or client, it has been necessary to turn down the HF either at the speaker or externally in order to achieve a successful working environment. When measured this always ends up in the B&K/Harman family.
If we could take any one of those rigs to an Anechoic Chamber, the -HF or DRC filtering still applied, any measured response will obviously include the filter.

DD