What is the ideal High Freq extension?


I think for a world class system to sound like live music, a good smooth grain free silky (if music calls for) high freq extension is mendatory. But in my mind where the 'flatness' should roll off (at what db rate?)is a big unknown and not quite clear. Is it 10k HZ, 12.5k HZ, 15K HZ or the ideal 20k hz.

This assumes you already have excellent bass extension and mid range.

One would think at ideal 20k hz there will most likely excessive ambience and thinner sound.

What is your opinion?

Thanks.
nilthepill
I think spectral analyzer would be more accurate way to measure system system response at your listening position. I think the Ratshake sound meter may not be accurate the freq extremes even with so called corrections applied.

I measured my system response at about 7 ft away (my seating position) and it is RULAR flat ( with corrections applied) from 25 hz to 16 khz. with 2 db boost at 20 hz and 7 db down at 20 kHZ.
Nil, I borrowed a hand held spectral analyser with a microphone built in its "nose" cum pink noise generator in a wall wart type of thing. Can't remember the name of it, but it was US-made, if I recall right, had a lot of LEDs in its display. A nifty little thing.
I have heard speakers with purposed response to 50Khz.While you might not hear that high up,the over-tone structure may be more completely revealed.A lot has to do with the crossover and phase linearity.
You must remember when digital "brick-walled" everything above 20Khz.Thank goodness we got past that!
It depends on how old you are and how good your hearing is. When I was in my late teens, I could hear up to 18Khz. Now, 30 some years later, I could hear no more than maybe 12 or 13Khz and music sounded very different to me than when I was much younger.
Detlof,

Sure I would certainly be interested to know the your acaoell tweeter experiment. How did you measure your Sound lab extension to 20K hz? Via Sound meter? with necessary correction applied? ( Here in US we have Radioshack sound meter that I typically use).

CKoffend, thanks for reminding me that live music extends well past 20 Khz (How can i forget? with recent threads comparing analog vs digital medium) as much as like 70k hz and beyond for certain instruments.

So in a perfect system you would want 'flat' response to and beyond 20khz- No roll off. Hmmm Now It makes sense why some audiophiles(Detlof and some other I know)prefer super tweeters hooked up to their main speakers.

Now that I think and look back I do remember auditioning a super tweeters in an all audio note system and discerning difference between these tweeters being on and off!

Any recommendation which super tweeters to try with my Dunlavy Vs?
I agree completely with Mrtennis here and in experimenting through the years, especially with Quads, but also with other ELS, the only really satisfactory (to my ears of course) top end was found with the tweeter panels of the long defunct Servo Static speakers, which I tweaked, so I could attenuate them. Later with the Quad63s after much experimentation I settled with the aCapella plasmas which I placed on top of the Quads and drove them separately so that I could adjust their SPLs. They go up to 35khz. For the bats there of course, but the addition of these speakers had a tremendous influence (positive in the sense of more natural) on the entire sound stage and the rendition of transients, bloom,overtones and ambience .
Another thing is to be considered: Experiments have shown (unfortunatly I don't remember where I read about this, so I cannot point you to this, it was about 20 years ago), that listeners, unable to hear tones above 10khz, felt uncomfortable when music was played to them with its upper freqencies cut off above 18khz. So even if we cannot hear it, we obviously feel something is missing, when it is not there. So I personally am not a believer in fixed roll off points in the highs, but prefer, as MrTennis suggests, if I read him right, to set the highs myself, so that I have all the liveliness and ambience I wish for, without the sound per se getting thin, even if it practically means to set up a second chain to do so. (By careful placement, phase anomalies can be easily avoided)
At the moment, from what I've measured, the Sound Labs seem to start attenuating close to 20khz in my room and to me they sound just right. I still have the aCapella tweeters and will in due course experiment with them, driving them with the Zanden9600s to see what happens andI shall report, Nil, if you are interested.
Real/live music extends well past 20K HZ. While some claim this is not audible I suspect it has an impact on the overall presentation. If you look at some of the finest components on the market, many exceed well past what has become the standard 20K range. While this alone may not put them in the absolute upper echelons of performers, it certainly is likely to contribute.
if live unamplified music is a reference, and the concert hall in particular, the attenuation in treble will vary with distance from the stage, the acoustics of the hall and the number of bodies in front and to the side of you.

in addition, the voicing you prefer should also be a factor. it is helpful to have access to an attenuation circuit, if you perceive too much energy in the treble region.
"I think the music should dictate the frequencies" Absolutely agree. I also agree that room and its contents would give you treble limitation. But say if you are 'designing' a dedicated room around your system what would you target the roll-off?

Per Shadorne it is the certain tweeter (silk dome, metal dome?)design that seem to be the problem/limitations and that electrostats do better in reproducing a true high freq response that does not sound hockey at live levels. How about ribbons? horn? Berrylium and the thingamjig?

Why can't the high end manufacturers finally solve this common basic problem then?
My opinion is that the behaviour up to 12 Khz is much more important than between 12 and 17 Khz and frankly above 17Khz it is mostly for the "bats & dogs".

All too often I hear tweeter grain and harshness from compression at realistic levels. Another issue is resonance from light weight materials that ring like a "bell" - this can be very intrusive as it is completely unrelated to the music playing.

Tweeter compression and ringing is all too common and it is this which causes a loss of "smooth grain free silkiness" at realistic "live" levels, IMHO. It is one of the principle reasons that many high end systmes are "highly resolving but fatiguing" or have "etched" sound - and just don't sound natural in the way electrostats do.

In short, I do not see it as a high frequency extension issue at all. This is unfortunate for most people who tend to buy on "specs" and manufacturers that put too much money into high cost impressive extension tweeters instead of improving other areas of their designs. I am old school though - so feel free to diagree.
I think the music should dictate the frequencies - not your own judgement. In this case, the extension from the speakers should always be higher than that of the recordings. For good transient control, you need high system bandwidth - the higher the better.

But the thing you are forgetting is the room. It ultimately will dictate the extension regardless of what you think it should be or what your speakers are rated at. Typical in-room measurements using an averaging technique will show that most rooms start to roll off the treble around 9kHz (like my room). The amount of furniture and the texture of the materials will change it up or down - but generally down is much more likely.

Arthur