Frequency Range

 

New member
Username: Sjordan872

Post Number: 7
Registered: 01-2004
How important is frequency range when matching a set of speakers with a good receiver that has a wide, dynamic range? My HK630 has a very wide range and in my quest for speakers, I've been steered towards some great ones that I've been auditioning. My only concern is the freq. range. Example; are the Athena S2 bookshelves, which have a range of only 50Hz - 20kHz, more dynamic in what they can produce compared to (for example)the Canton LE102 bookshelves with a range of 42Hz - 26Hz? Some speakers I've auditioned have even gone higher almost to 30Hz. Low range shouldn't matter too much if used with a good sub, but what about the higher ranges? Especially when used with a dynamic sounding receiver such as an HK? I know the human ear can only hear up to a certian level, but I'd like to get as much out of my receiver as I can. Is there a flaw in my reasoning?
 

Silver Member
Username: Geekboy

Post Number: 173
Registered: 12-2003
Blazer: I had an eloquent response but my computer ate it!

First a minor correction. The Canton LE102 are 42Hz to 26,000Hz (42Hz - 26kHz). That's a big difference! Most common speakers today have a response in the 80Hz - 20kHz range. Most humans can't hear anything measurable above 13kHz (13,000 Hz) so frequencies above this... are for the dogs! :-)

The Athena and Canton's have relatively the same frequency response. The base extension down to 50Hz and 42Hz, respectively, is pretty good. I would say, though, that most bookshelves and even loudspeakers (floorstanding) don't do that well in the lower frequencies. I have Paradigm Reference Eclipse/BPs that are rated to 35Hz with base extension (due to cabinet design) to 18Hz.

Albeit these speakers are rated pretty low, these speakers generally don't do well in the bass frequencies below 80Hz anyhow. The bass will seem tighter but not boomy. For real base extension, you would add a subwoofer, as you write, which can extend down even lower. Depending on where you like to have your cutoff frequency, you can adjust how the base response is for your music. (I have mine set to 60Hz and my Paradigms seem to be ok with that.)

So, the frequencies above 20kHz... I don't know. Reproducing frequencies past the upper limits of human hearing... I don't have any opinion. I guess it's like a 1080i progressive HDTV broadcast on an NTSC 525i TV. Sure, the broadcast is better, but the TV can only do 262 lines a second. :-) Probably not a good analogy. I just don't know what the super high frequencies do for the overall sound. It may be like the "butterfly affect" -- meaning that the inaudible frequencies in the above 20kHz range do have "some" affect on your hearing below 20kHz.
 

New member
Username: Goose

Post Number: 6
Registered: 01-2004
The nominal range of human hearing is 20Hz-20kHz. Depending on your age and gender (older males faring worst), the upper limit will very likely be lower than that (12-15kHz by middle age, from what little I have read). As an aside, the frequency response of the ear-brain system is not linear, and is also volume-dependent. That is, the perceived loudness of a sound depends on its frequency, and how loud it actually is to start with! So make sure you audition those speakers at the volume you intend to listen to them at ...

Anyway, getting back to the topic, there are those who argue that there are audible sub-harmonics of ultrasonic sounds that do make a difference to what you hear. Two questions spring to mind:

1) are those ultrasonic sounds actually captured on the recording medium?

2) are the sub-harmonics somehow not captured and have to be produced at playback?

Standard CDs can only record up to 22kHz at best, so say goodbye to information at higher frequencies. And if those sub-harmonics are so important, wouldn't they be present in the original source material, and thus get recorded to begin with?
 

New member
Username: Sjordan872

Post Number: 8
Registered: 01-2004
That makes a lot of sense. Thanks Geekboy and Goose!
 

New member
Username: Gecko

Post Number: 10
Registered: 01-2004
http://arts.ucsc.edu/EMS/Music/tech_background/TE-02/AcNumbers/AcNumbers.html

that link gives a pretty good overview of the important factors in how the human ear perceives sound.

Hope that answers a few questions for you.

Pete
 

Silver Member
Username: John_a

Post Number: 155
Registered: 12-2003
All correct, I think. Excepting that the audible frequencies from ultra-sonic tones are difference tones, not "sub-harmonics", Goose. Difference tones have frequencies that are the difference between the two higher frequencies that produce them. The microphone captures difference tones just as well as we do.

The 20 kHz upper limit became fairly standard with the introduction of CD. Analogue sources (high-end tape and LP) could usually go much higher. The reason for the CD ceiling is that you need to take up far more digital information on the disc for those very short-wave, high-frequency waveforms than for longer, lower ones, so the law of diminishing returns sets in early for use of digital storage capacity. Also, the CD sampling frequency is only 44 kHz, so its ability to re-create accurately a 20 kHz waveform is very limited. DVD-A gets back to the high-frequency response typical of analogue systems.

Whether anyone can hear the difference is a another question. If you can hear the 14 kHz carrier frequency "whistle" on terrestrial FM broadcasts (few can) then you may get some benefit from having accurate reproduction of waves up to and beyond 20 kHz. Not that you can hear the waves themselves, but you may hear a more accurate reconstruction of the original analogue waveforms of slightly lower frequency. But I am doubtful if even that difference is really audible, and, anyway, you need to get out of CD and FM territory into analogue, DVD-A, or satellite broadcasting as your source.

KEF claim for their "hypertweeter" "...flat response to 50kHz and useful energy up to 70kHz". KEF's argument is by analogy, always suspect. "....this is not as futile as it may at first sound - whereas conventional CD's carry no audio information beyond 20kHz, extended bandwidth formats such as SACD and DVD Audio contains signals up to 40kHz and more. Consider the analogy that even though no-one drives at 200 mph, cars that are engineered to do so are inherently better performers at lower speeds." (KEF Audio).

So, just to be really sure, why not travel by LearJet when next you nip out for a pizza, ....

DVD-A will deliver sound even to 96 kHz in two-channel mode, I think.

There is absolutely no point in playing CDs through speakers that go higher than 20 kHz: there is no sound there to reproduce.

That is high frequency. Low frequency is "a-whole-nother" issue. While 26 kHz is unlikely to be perceptibly different from 20 kHz, 42 Hz is better than 50 Hz, Blazer - other things being equal, which they are almost certainly not. Also take care they quote "at -3dB" or similar for the roll-off at the stated frequencies. If they don't, they are hiding something!
 

New member
Username: Goose

Post Number: 9
Registered: 01-2004
I take issue here:

... the CD sampling frequency is only 44 kHz, so its ability to re-create accurately a 20 kHz waveform is very limited.

The Sampling Theorem says that as long as you sample at *at least* double the maximum frequency present, you are OK. No information is lost this way, and none is gained by sampling at a higher rate.

(Actually, the Sampling Theorem deals in bandwidths, and you only need to sample at double the bandwidth, but for 20Hz-20kHz it makes little difference.)
 

Silver Member
Username: John_a

Post Number: 157
Registered: 12-2003
Goose,

Thanks. I always have to go back to basic principles. You are saying: make n waves, each out of 2n 16-bit samples, and do you get a good approximation to the original? Yes, I guess so. But I always imagine you could see the "corners" in the waveform if you get up close. Try it with 2-bit samples and I doubt you are OK. The sample size must come in somewhere.

Hey, tell me about this sampling theorem. It is news to me!

Please allow that the amount of info you must encode becomes huge at higher frequencies, and the question comes whether approximations are allowed, or do they produce discernable deterioration of the sound.

Such considerations were part of the analogue counterblast to CD in years gone by. I have been an agnostic on digital all this time.

Best.
 

Silver Member
Username: John_a

Post Number: 158
Registered: 12-2003
After some more thought, Goose, the "Sampling Theorem" must say that sound frequencies above 22 kHz are not OK at 44 kHz sampling frequency?

I guess this would predict that the CD era brought in a 22 kHz ceiling....?
 

Silver Member
Username: John_a

Post Number: 159
Registered: 12-2003
...and Sampling Theorem also predicts why sound at up to 96 kHz is OK with the DVD-A sampling frequency of 192 kHz. I like it. Please tell me more!
 

New member
Username: Goose

Post Number: 10
Registered: 01-2004
I just grabbed this out of Google, but for starters try here for the Nyquist-Shannon Sampling Theorem:

http://en.wikipedia.org/wiki/Nyquist-Shannon_Sampling_Theorem

and here for the accompanying Nyquist-Shannon Interpolation Formula that tells you how to put it all back together:

http://en.wikipedia.org/wiki/Nyquist-Shannon_Interpolation_Formula

Thee pages deal in Fourier Transforms, which are also described on that site (follow the links).

Basically, what it amounts to is that, yes, if you are Nyquist sampling a sinusoidal waveform you only get two samples per wavelength. If you try to reconstruct the signal in the time domain, then you will see a square-wave result. But a square wave is a Fourier sum of harmonics of the fundamental, and these higher harmonics are not "real" in the sense that the original signal did not have them (we are assuming that it was band limited to this fundamental frequency and lower). Limiting the band of the output to be the same as the band of the input gives you back the sinusoid you were looking for by throwing out those higher harmonics. All CD players have anti-aliasing filters to achieve this end.

You only run into troubles if you are undersampling your signal. Then you do not have enough information to reconstruct the input, and the higher frequencies of the input will be aliased back into the band you can reconstruct, thereby corrupting it.
 

Silver Member
Username: John_a

Post Number: 160
Registered: 12-2003
Goose,

That is really helpful. Thank you. I never knew that.

I expect the anti-aliasing filter is at 20 kHz and this is what sets the upper frequency limit for CDs. So we never get "stepped" waves. This is an "analogue myth" I have long believed. I think I read it once when Tiefelbaum of Linn was attacking the whole idea of CD as an inferior medium to LP. I could be wrong and would not wish to defend that in court!

So my "more accurate reconstruction of waves" is tosh. I take it back. My argument does not rest on that, of course.

What a useful site this is.

I still maintain that sounds above 20 kHz are inaudible to humans. The industry must be in quite a panic if even reputable makers like KEF are telling us their speakers can reproduce sounds nobody can hear. I suppose they have to go along with the hype for DVD-A and SACD. In my view the real progress in those formats is multichannel capability. 24-bit sampling may make a difference, too. Reproducing sound up to 96 kHz is absurd, and a waste of money. The same industry told us, twenty years ago, that CD, with its 20 kHz upper limit, gave "perfect sound".
 

Bronze Member
Username: Gecko

Post Number: 13
Registered: 01-2004
John A. I have to clarify a few things with sample rates and bit depth.
First off, the fact that DVD-A can reproduce sound up to 96kHz is misguided.
The DVD-A format samples the sound at 96kHz, that means (in quick overview form) that when recording the sound, the voltage level and picture of the sound wave is taken 96000 times per second.
This is just a statement of the rate at which pictures are taken, where as CD's take pictures of the sound wave about 44100 times per second.
The second number is 24 bit,this is the length of the binary number that is used to keep track of the voltage every time a picture is taken. CD's are 16 bit, meaning there are approximante 65500 different levels that the sound wave can be at (2^16) which looks like alot. Take this value and divide it by two and you have the theoretical dynamic range that is possible in 16 bit recordings.
24 bit audio ups the number of steps possible to about 16,777,216 possibities. So the possible dynamic range and resolution is increased by roughly 256 times. That is the most noticable difference between CD and DVD-A/SACD.

The 96kHz compared to 44.1kHz adds a little more "air",or space, and higher frequency accuracy that makes the audio almost exact to the origional source. The 24 bit depth adds dynamic range and detail to the entire sound.

Now if we could just recreate the head bump, tape saturation, and tape distortion accurately - there would be no more need for analog recordings. :o)

-Pete
 

Bronze Member
Username: Goose

Post Number: 12
Registered: 01-2004
There are subtleties to all of this. The reason for the plethora of over-sampling techniques common to CD players is that it is actually hard to construct a filter that passes 20kHz unmolested, but stops dead anything higher than 24kHz (where the aliased signal shows up if your input signal is limited to 20kHz). Over-sampling helps by moving this limit higher in frequency, so it is easier to make a filter that gives you what you want down at 20kHz.

The problem with digital audio is not that it is inherently worse than analog, just that, when badly done, it is awful. Yes, there is quantization noise (but show me a noise-free LP or tape). However, I believe that it is the quality of the digital-to-analog conversion that is crucial. For example, I have an old Rotel CD player. It is not the greatest thing around, but I like it, and, despite being 10 years old, it makes a brand-new Sony DVD player sound like garbage when playing CD audio. The Sony has, in comparison, very thin bass and very harsh treble. The digits coming off the disc are the same, but how they are turned into analog signals is different.
 

Bronze Member
Username: Goose

Post Number: 13
Registered: 01-2004
Incidentally, all of this applies to digital photography as well. Any decent digital camera (and I mean decent) uses an anti-aliasing filter to band-limit the image before it hits the sensor, so that the sensor does not undersample the information and produce an artefact ridden image. A notable exception is the Sigma line of SLR cameras, which omit this filter. I suggest you go to Google groups and enter "Sigma SD10" for the roaring debate on this subject. Make sure you have lots of time, and just keep chanting "Nyquist-Shannon" to help retain your sanity!
 

Silver Member
Username: John_a

Post Number: 164
Registered: 12-2003
Pete,

Thanks. I agree. But DVD-A has a sampling frequency of up to 96 kHz for 5.1, and 192 kHz for two-channel stereo. The Sampling Theorem (thanks, Goose) and manufacturers' claims are therefore in agreement - DVD-A is capable of giving sounds of up to a frequency of 96 kHz.

BTW I have some DVD-A discs and can play then only in the DD and DTS DVD-V formats. But DTS is 24 bit, 96 kHz, too, and the DTS sound is wonderfully detailed and clear, even through speakers that roll off at 20 kHz (as does the receiver). So I am happy with 'a little more "air",or space, and higher frequency accuracy' but I still don't understand where it comes from. According to Sampling Theorem (see Goose's post) you can get a complete waveform over the audible spectrum by synthesis of samples taken at 44 kHz.

This gets back to my point about the benefit of DVD-A extending down into the audible spectrum. But getting back to Blazer's question, I think you do not need speaker that go above 20 kHz to hear that difference.

Goose,

Thanks again. Your post stopped mine, and I had to re-type all that!

The problem with poor digital sound is not what it introduces, but what it leaves out. Play an MP3 file through your system and see. Anaemic, lifeless, washed-out. Give me a good LP anytime. I can stand the little bit of noise if it comes with the sound and "smell" of a flesh-and-blood performance. I hear that, no question. Wish I could account for it!
 

Bronze Member
Username: Goose

Post Number: 15
Registered: 01-2004
CD audio is good, but it is not the greatest. Theorems aside, there are practical limitations, some of which I touched on above. Quantization noise is one (see also Pete's post). The fact that you are reconstructing an analog waveform from a digital representation obviously depends heavily on how well that reconstruction is done. No filter is perfect, so at the upper end of the audible spectrum there are issues. I think most will agree that CD audio has gotten better over the years as some of these practicalities have been improved upon.

DVD-A has every chance to be better than CD audio, because it is sampled at a higher rate to start with, has a greater bit-depth (less quantization noise), and any practicalities at frequencies approaching the sampling rate are so far out of the audible spectrum that they can have no significant effect on what you hear. This "cleaner" high-frequency end of the audio spectrum may explain the extra "air" that you hear.
 

Silver Member
Username: John_a

Post Number: 167
Registered: 12-2003
Goose,

You have given things back, and I thank you for the insight and education.

As regards DVD-A, it is better than CD, no question. Let me flag the thread where I learned some things on that subject, with thanks to all there for the input, some of it heated.

Home Audio: Receivers: Archive through January 07, 2004: What does "DVD-audio" mean here?
« Previous Thread Next Thread »



Main Forums

Today's Posts

Forum Help

Follow Us