Flat Response

[Zilch]

Re: Flat Response

I do not believe (and it has been proven by many commercial designs) That flat anechoic response does not equal good sound, or perceived sound.

Summarily established by Toole 20 years ago, and supported by 1000s of repeats in the intervening years, flat anechoic axial response is the single most significant correlate of listener preference.

With respect to power response, "flat" in the sense of horizontal is known to be overly bright, so "smooth" or "linear," may be the preferred descriptive....

[John L. Murphy]

Re: Flat Responce

I would like your opinions on actual vs. perceived flat frequency response. Some say a flat response sounds to forward or even harsh when auditioned in typical listening rooms,
yet it seems a flat response is an important design goal for many designers.

<snip>

Let's reframe the question a bit and see what we get...try this:

"What is the optimum (in-room, spatially averaged) frequency response for a home music playback system where the program material will consist of commercially available recorded music from the past 50 years?"

I tried to specify how the response is to be measured and the program material that would be auditioned. These are both critical elements of any serious inquiry into the appropriate frequency response for a sound playback system.

If we were strictly doing acoustics research and were trying to accurately reproduce the sound of a live source (a voice for example) then we would use a microphone with a known flat frequency response to detect the sound and then we would reproduce it through a loudspeaker with a flat (in-room, spatially averaged) on-axis frequency response. This setup would likely provide a good reproduction of the live sound source.

I submit that when listening to the playback of commercially produced recordings the above "laboratory measurement" scenario is not the correct view of the situation. Rather, the real goal of a home music reproduction system is to "reproduce" what the artist intended for you to hear. That is, to reproduce the sound of the playback system on which the music was crafted. This is an entirely different goal than the laboratory reproduction goal.

Let's imagine an extreme case just to make the point. Imagine that the studio monitor system on which the music was created had a 20 dB bass boost and a 20 dB treble cut. Individual tracks of the recording were first auditioned through these monitors and microphones were selected based on how they sounded through these (extremely colored) monitors. Now the recording engineer has a big collection of microphones that have widely varying frequency response curves. At first he picks a fairly flat mic to see how it sounds. But (no surprise here) it sounds muddy, has excess bass and not enough treble. So he selects a different mic that he knows has less bass (-10 dB) and more treble (+10 dB) than the flat mic. This mic sounds much better on the colored monitors but still needs a little EQ adjustment. The engineer dials up a 10 dB bass cut and a 10 dB treble boost and now the mic sounds "right". He then proceeds to record all the parts of the record through this carefully set up recording channel using the highly colored microphone recorded through a highly colored (EQ'd) mixer channel. The final mix sounds great, just like you were in the room with the musicians. The record is released and is a big hit.

Now, here comes the problem. You buy the record and play it back on your "flat" home music system. Your instant reaction upon hearing the first sound is to grab the volume control and turn it down because it is so awful sounding. There is no bass and the treble is horribly emphasized. What is wrong? Well, the bass is (predictably) deficient by 20 dB and the treble is (again, predictably) 20 dB in excess. What is wrong is this, your playback system has a different frequency response than the monitoring system on which the music was recorded, mixed and mastered. In order for you to hear what was intended (the true goal of music reproduction) you need to "voice" your loudspeakers to match the response of the monitors that were used to create the music. That means you need the same 20 dB bass boost and 20 dB treble cut of the studio monitors used to make this product. The response of the playback system clearly needs to match the reference response of the recording monitors in order for the recording to sound like it was intended. There is no need to declare any single reference as the one true frequency response as if it were a moral issue of right and wrong. The correct frequency response is simply the one that matches the reference monitor response.

Identifying the optimum playback frequency response remains THE central problem with standardizing the reproduction of recorded music. Traditionally there has been NO ESTABLISHED STANDARD for the frequency response of studio monitors. However, most loudspeaker designers with whom I've discussed this topic seem convinced that the "correct" response for a loudspeaker is "flat on-axis anechoic response".

The cinema industry is much more organized than the record industry in this respect. The cinema industry actually HAS A STANDARD for loudspeaker frequency response that is shared by both the mixing rooms where the audio sound tracks are created and the local cinema theaters where everyone hears the sound track playback. The actual frequency response of the loudspeakers is not nearly as important as the "match" between the mixing room sound system and the theater sound system. The goal (correctly) is for the theater audience to hear what the sound mixers heard. Imagine that.

In order to have the "best fit" frequency response for auditioning music from the last 50 years or so we would need to know the average frequency response of all the studio monitors used to prepare all that music. If that grand average were flat then I would be the first to agree that we should listen to that material through flat playback systems. But does anyone really believe that on average over the past 50 years studio monitors have exhibited a flat playback response? I know better from firsthand experience. In the past I routinely voiced professional recording studio main monitors with what was considered the unofficial standard of measuring using pink noise with an RTA and then voicing the system to be flat with a 3 dB per octave roll off above 2 kHz. That unofficial standard was eventually adopted by the cinema industry as the official "X-Curve" in use today.

If you accept that the "optimum" playback frequency response is the reference response of the music mixer's studio monitor system then you must realize too that there is a large variation in that reference response for the library of recorded music and also that it is changing in time over the years. There seem to be more and more loudspeakers moving toward flat (typically anechoic) on-axis response in an attempt at being "accurate". Thankfully for listeners these systems do not typically deliver a flat in-room spatially averaged response. That response, fortunately, is somewhat closer to the x-curve with its rolled off high end. So modern speaker designers seem to be torn between paying homage to the "flat response" required for the laboratory reproduction scenario and providing "enjoyable sounding" speakers for auditioning commercially prepared music from the last 50 years.

If you look at the body of recorded music from the last 50 years there is broad variation in the frequency response of the recordings. Is it reasonable to think that any ONE frequency response can do an acceptable job of recreating the reference response necessary to accurately render these recordings? That seems to be what we try to do when we setup playback systems with one fixed voicing. On average it works but occasional recordings will lack a little treble or bass while others have a little too much.

In the future perhaps the music industry will put us out of our non-standard misery and officially adopt either the x-curve, in-room flat or some similar response as a standard but until then we seem to be pretending to make speakers with a flat response while actually sneaking in the treble cut necessary for the optimal enjoyment of the last half century's music.

The need for matching the response of the home user's playback system to the reference response of the studio monitor system seems like a no-brainer. Perhaps the day will come when recording engineers switch to flat microphones and flat studio monitors but until then the best solution for home playback is to have a playback system with flexible voicing. The choices for voicing would start with:

1) in-room, on-axis, spatially averaged, flat

2) in-room, on-axis, spatially averaged, x-curve

3) in-room, on-axis, spatially averaged, small room x-curve

In addition to these fixed voicings, bass and treble controls are appropriate for correcting the response by ear for older recordings where the reference frequency response can vary widely and most obviously.

Some modern recordings are moving closer to the theoretically ideal method of using flat microphones and mixing on flat monitors. These newer recordings have the best chance of sounding good when auditioned on flat speakers. But there is no official way of knowing how a record was recorded in order to select the appropriate playback response. You just have to use your ears and decide if you need an adjustment.

I hope this sheds some light on how the optimum playback frequency response of a music playback system is not a trivial issue with the simple pat answer that “flat is correct”. Rather, I hope I have convinced you that the correct playback response is the reference response of the monitors with which the recording was created. Review my exaggerated example (above) if you do not agree.

<<flame suit on>>

Regards,

John

[ligs]

Re: Flat Responce

Do you have the response curves of the more popular studio monitor?

[Zilch]

Re: Flat Response

Most competent recording engineers are well aware of this stuff, but their job is (and has always been) to produce product that sells, which means, statistically, anticipating the listening preferences of their target audience and the sonic character of their typical playback systems. In this respect, loudspeaker manufacturers have largely accepted the challenge of breaking the cycle of confusion by themselves settling upon a standardized target response -- flat. As Toole has demonstrated, even desktop mini-systems may be seen as striving toward this goal.

The fact that the spectral balance of source material has varied substantially over the course of the history of its production does not argue in favor of standardization upon any particular alternative voicing, but there is clearly advantage in having a baseline from which listeners may exercise their EQ acumen as desired. The central tendency of loudspeaker design is flat response, and the recording industry knows what's up.... ;)

[Guest]

Re: Flat Responce

Rather, the real goal of a home music reproduction system is to "reproduce" what the artist intended for you to hear. That is, to reproduce the sound of the playback system on which the music was crafted.

Hi John,
So under this scenario, I would (per occasion) want my loudspeakers to sound like a Diatone or a tissue enhanced NS-10?
No thanks ;).
I would rather it be a reasonable facsimile of the sound on instruments in my memory.
Actually, the last thing I want to "hear" is loudspeakers (for my type of non-amplified/acoustic derived music). I would rather not hear any loudspeaker signature at all (including my own), so I respectfully disagree with your position.

I will finish with this.
Generally, when a person states that they would rather have a "non-flat" speaker (as if it radiated sound only on one axis..and one perceived sound in room only from this one imagined axis) and would rather distort the "on axis" to "correct" it, what they are really saying (and doing), is that their speakers have problems off axis. By "correcting" the on axis, they have unknowingly actually "corrected" sound power problems due to poor off axis.
The vast majority will have typical (tragic) cone and dome, flat on the face of a box 2 ways, which may have a reasonably linear on axis response...and a disaster in every other direction. If correcting the sound power disaster requires also "correcting" the on axis by distorting it, to produce more palatable in room sound, so be it. But identify it for what it is...a sound power problem, not an on axis one.

cheers,

AJ

[robert]

Re: Flat Responce

Off of standards and such - my most "flat" and lowest distortion measured speaker would be my pair of Zaph Audio ZRT's. Experiences with them pretty much back up other posts in this thread.

1.) On well-recorded music, they are amazingly SWEET and sound "right."

2.) On poorly recorded stuff, they'll shred it.

My 8" 2 ways are easier to listen to playing popular music, and yes I did down tilt the treble a bit, and actually attenuated a touch more at Ft. Wayne from listener input there.


Cheers / Robert

[Pete Schumacher]

Re: Flat Responce

Do you have the response curves of the more popular studio monitor?

This would be typical.

http://techtalk.parts-express.com/showpost.php?p=1655701&postcount=428

[ligs]

Re: Flat Responce

The FR looks like a very good home speaker Even and flat response over most of the range, the kink in very top end is not audible anyway!

[DDF]

Re: Flat Responce

I was reading something in another forum, and the poster dug into the detailed outcomes of Toole's famous preference tests. What he found was that a couple of the most highly prefered speakers had poor off axis response and a couple with extremely good off axis response were poorly rated for preference.

The poster indicated these were less outliers than thought, and that thecorrelation of good sound = good off axis is less strong than the current common wisdom, based on Toole's own results with drove this common wisdom in teh first place. This puts some solid caveats on the ubiquitous goal of smooth power response.

I haven't had the time to go back yet and review the papers for these results but they are like big gongs going off, demanding attention to see if they're true.


Dave

Went back through the Toole papers. Bottom line: its not true what the poster was surmising. While there are a few speakers with very well controlled off axis response that scored poorly, their on axis was very erratic.

However, there were a couple that scored extremely well while having some largish dips in the off axis. A couple scored well with off axis peaks at 5 kHz, but for the most part, large peaks off axis correlated with a lower score. Supports the typical principle that dips are more benign than peaks, seems to apply off axis as well.

[Pete Schumacher]

Re: Flat Responce

The FR looks like a very good home speaker Even and flat response over most of the range, the kink in very top end is not audible anyway!

The key with these monitors is that as you move off axis, the response stays smooth, maintaining the character of the on-axis response.

[auracle]

Re: Flat Response

Summarily established by Toole 20 years ago, and supported by 1000s of repeats in the intervening years, flat anechoic axial response is the single most significant correlant with listener preference.

With respect to power response, "flat" in the sense of horizontal is known to be overly bright, so "smooth" or "linear," may be the preferred descriptive....


The central tendency of loudspeaker design is flat response, and the recording industry knows what's up....

"Flat anechoic response" is a common slogan that has very little actual practical meaning for establishing a design standard. Flat response in an anechoic chamber at what distance - 1 meter, 3 meters, 10 meters? Since the energy available for inevitable propagation loss is directly proportional to wavelength, the frequency response balance is altered significantly with distance - anechoic chamber or not. If that were not the case, there would be no need for products like this in large studios, auditoriums, and concert stages where the compression drivers you use for high frequency playback were actually intended to be used:

http://www.meyersound.com/support/papers/REM/index.htm

When was the last time Toole or anyone else for that matter sat a bunch of speaker evaluators down in an anechoic chamber to test speakers? Anechoic chambers are used as a means to take the environment out of the measuring process so that design change differences can be recorded and tracked - not to inform the design engineer when he's done if his speaker measures flat in an anechoic chamber.

This drives home the point Linkwitz was making - creating a believable "audible scene" does not necessarily require flat on axis response. And it doesn't necessarily require uniform off axis response either. His papers emphasize the role the listening room or environment plays in overall results and that response must be tailored to account for the interaction with that environment.


As for "The central tendency of loudspeaker design is flat response, and the recording industry knows what's up.... " - you've got to be kidding.

Anyone who's heard a popular artist's song like Gwen Stefani's "Luxurious" would laugh at the notion of the "recording industry knowing what's up" from a mixing or mastering perspective. Recording and mastering practices, particularly in the digital age, are as all over the map as the artists themselves.

[Deward Hastings]

Re: Flat Responce

The key with these monitors is that as you move off axis, the response stays smooth, maintaining the character of the on-axis response.

No, actually the key is that as you move off axis the high frequency response falls, giving a (substantial) rolloff in power response (which combined with "flat on axis" still gives a drooping high end for the listener). "Flat" on-axis combined with "flat" off-axis (constant directivity) produces a speaker that is too bright and "forward" . . . one has to roll off the high end to produce an acceptably "natural" sound . . .

[Pete Schumacher]

Re: Flat Responce

No, actually the key is that as you move off axis the high frequency response falls, giving a (substantial) rolloff in power response (which combined with "flat on axis" still gives a drooping high end for the listener). "Flat" on-axis combined with "flat" off-axis (constant directivity) produces a speaker that is too bright and "forward" . . . one has to roll off the high end to produce an acceptably "natural" sound . . .

The off axis response of the midrange unit drops off with frequency too, and the WG loaded tweeter follows that drop off, maintaining a smooth response off axis, the tweeter dropping with the midrange. Minimal peaks and dips as you move off axis, maintaining the same character as the on axis response.

A non WG loaded tweeter in a typical 2000Hz LR4 crossover with a 7" woofer has a peak in response above the cutoff as you move off axis. That problem is minimized with a WG.

[auracle]

Re: Flat Responce

The off axis response of the midrange unit drops off with frequency too, and the WG loaded tweeter follows that drop off, maintaining a smooth response off axis, the tweeter dropping with the midrange. Minimal peaks and dips as you move off axis, maintaining the same character as the on axis response.

A non WG loaded tweeter in a typical 2000Hz LR4 crossover with a 7" woofer has a peak in response above the cutoff as you move off axis. That problem is minimized with a WG.

Seems like we're veering off the path of righteousness and salvation a little. These generalizations above might be true if all tweeters were made the same way or if all waveguides, woofers, baffles, and enclosures were made the same. There is so much variability in all of these characteristics - making general statements beyond the most general (like, high frequency response usually drops as you move off axis) is kind of pointless.

[Pete Schumacher]

Re: Flat Responce

Seems like we're veering off the path of righteousness and salvation a little. These generalizations above might be true if all tweeters were made the same way or if all waveguides, woofers, baffles, and enclosures were made the same. There is so much variability in all of these characteristics - making general statements beyond the most general (like, high frequency response usually drops as you move off axis) is kind of pointless.

We were speaking specifically of the Behringer 2-way with the WG loaded tweeter. At lease, I was.