And the inevitable next question is "if we can detect "ultrasonic" frequencies, do we want to?" i.e., do they add to or detract from our perceptual experience.
I can certainly feel a hammer dropping on my head, but I've learned to avoid that if possible.
If by some mechanism we "sense" these frequencies, does that necessarily mean that we should strive to record and reproduce them faithfully?
I can certainly feel a hammer dropping on my head, but I've learned to avoid that if possible.
If by some mechanism we "sense" these frequencies, does that necessarily mean that we should strive to record and reproduce them faithfully?
Hi,
IMHO, yes you do want that.
They add to the illusion of listening to live instruments not a recorded, lifeless sounding reproduction.
Another aspect to consider is that with steep filtering at both ends of the so called audible range you'll inevitably introduce phase anomalies.YEMV.
Cheers,
IMHO, yes you do want that.
They add to the illusion of listening to live instruments not a recorded, lifeless sounding reproduction.
Another aspect to consider is that with steep filtering at both ends of the so called audible range you'll inevitably introduce phase anomalies.YEMV.
Cheers,
5th element said:
Well, earlier in the thread Tiroth questioned whether distortion measurements above 10 kHz were useful since the lowest possible harmonic fell above 20 kHz. That's one way of looking at it, but one must remember that non-linear distortion also produces intermodulation, and those products can fall in the passband. For example, harmonic distortion on 12 kHz and 16 kHz tones would generate a 4 kHz difference tone (most tweeters would reproduce this), the amplitude of which would be related to the level and transfer function of the distortion. This is part of the reason 19-20 kHz IM tests are used to characterize systems, although making sense of the measurements is not obvious.
I'm agnostic about whether humans can perceive (note I didn't say "hear") much above 20 (or even 16 [I'm over 45, gimme a break]) kHz, but I'm quite certain system nonlinearities in the top octave will colour the sound well below that range.
Now, if you want to speculate whether high amplitude ultrasonic frequencies might produce beat tones due to nonlinearities in the ear canal, then that's an interesting subject. Trumpets in particular have fairly high levels of content above 20 kHz, and I haven't yet heard a system, whether digital or analog, which has convinced me on that instrument.
Francois.
Well im 19 and can hear just above 18k my dad can get just past 12k i guess his ears are just not wonderful cause he certainly didnt abuse them I dont think.
Either way I am sceptical as to if stuff above 20k can actually bring any benafits. However one reason I am convinced to as to why 96/192khz gives better quality is not because they contain the ability to produce high frequencies. But in their ability to produce, with greater accuracy, the frequency range we can perceive.
Two samples per time period of a 20khz wave doesnt strike me as very much to go on to produce an accurate representation of infomation in music that is indeed at that frequency. The same goes for 10k and even 5k only four and 8 samples per time period of the wave. Im sure there is mathamatic proofs which say you only need a certain number of points to reproduce faithfully the analogue wave form after the ADC/DAC process. And when I mean faithfully I mean sine wave in sine wave out, not a sawtooth at 20k like my soundcard produces when asked to do a 20k sine.
Either way we know digital produces really good bass thats no surprise because at 40hz and 80hz sine waves will have approx 1000 and 500 samples per time period of a sine wave at those frequencies.
I would be interested to see if it is indeed the extra bandwidth or the increased quality within the audible range which actually provides the benafits of SACD and DVD-A. Its probably a combination of both, but I would imgaine the increased quality would win over bandwidth.
Either way I am sceptical as to if stuff above 20k can actually bring any benafits. However one reason I am convinced to as to why 96/192khz gives better quality is not because they contain the ability to produce high frequencies. But in their ability to produce, with greater accuracy, the frequency range we can perceive.
Two samples per time period of a 20khz wave doesnt strike me as very much to go on to produce an accurate representation of infomation in music that is indeed at that frequency. The same goes for 10k and even 5k only four and 8 samples per time period of the wave. Im sure there is mathamatic proofs which say you only need a certain number of points to reproduce faithfully the analogue wave form after the ADC/DAC process. And when I mean faithfully I mean sine wave in sine wave out, not a sawtooth at 20k like my soundcard produces when asked to do a 20k sine.
Either way we know digital produces really good bass thats no surprise because at 40hz and 80hz sine waves will have approx 1000 and 500 samples per time period of a sine wave at those frequencies.
I would be interested to see if it is indeed the extra bandwidth or the increased quality within the audible range which actually provides the benafits of SACD and DVD-A. Its probably a combination of both, but I would imgaine the increased quality would win over bandwidth.
Someone asked whether we can hear input above 20kHz...
First off, your upper hearing limit depends on a number of factors, such as the integrity of your hearing (e.g. age, and hearing damage), SPL, your gender, your nationality, your native tongue, and whether you have any diseases related to your respiratory system (e.g. asthma).
Assuming no hearing damage, for men, high frequency hearing loss occurs steadily over their entire lifespan, while for women, most of the loss occurs relatively late in life.
People from countries like Japan appear to have higher upper limits, although I'm on thin ice here. Physiological changes do occur at about 2 weeks.
Anyone whose native tongue is something along the lines of !Xhosa will hear higher frequencies better, due to the 108 or so clicking sounds in their 'alphabet'.
People with asthma tend to have a higher upper frequency limit. Myself and another person back in the highschool days were able to detect sinewaves of 22kHz and 23kHz respectively, as an upper limit. Turns out both have asthma, and the connection was verified later on, though I can't remember the publication.
Early Russian, and later other nationalities', experiments show measurable brain activity resulting from frequencies as high as 250kHz. Air-borne sine wave without lower frequency components, but I'm not sure if they're talking about headphones or speakers.
Danish scientists operate with ~14Hz as the limit where our perception transitions from a continous tone to discrete pulses. Research is under way to map out infrasonic hearing thresholds.
Lastly, remember that secondary effects, such as tactile input, all contribute to the total experience. For example, at low frequencies, headphones tend to seem unsatisfactory, due to the absence of tactile input. Stax earspeakers are better at this than most, since your head picks up the tactile bits.
I'd take statements to the effect that there is a world of difference between the presence or absence of supersonic signals with a pinch of salt, but similarly, discounting the effect of supersonics is rather silly. Especially in light of 100kHz ultrasonic carriers for spot-speakers giving tinnitus (granted at extreme spl's)..
However, the need for flat response is another matter. Air absorbs high frequencies, dependent on distance and humidity. ISO recommends reducing high frequency content above 10kHz with a gentle slope (6dB/oct, IIRC) for realistic output. Now, if people would just trade away the flat response in the top for a wider bandwidth, we'd be getting somewhere
PS: As to SACD, the format is not entirely flawed, but the design of a new medium should really be left to dCS or somesuch. 1MHz 4-bit would suit me just fine, and be easy to do computations on.. One of the cutest advantages of DSD is their potential applications in directly driving electrostatic headphones or speakers; let the air do the lowpassing.
First off, your upper hearing limit depends on a number of factors, such as the integrity of your hearing (e.g. age, and hearing damage), SPL, your gender, your nationality, your native tongue, and whether you have any diseases related to your respiratory system (e.g. asthma).
Assuming no hearing damage, for men, high frequency hearing loss occurs steadily over their entire lifespan, while for women, most of the loss occurs relatively late in life.
People from countries like Japan appear to have higher upper limits, although I'm on thin ice here. Physiological changes do occur at about 2 weeks.
Anyone whose native tongue is something along the lines of !Xhosa will hear higher frequencies better, due to the 108 or so clicking sounds in their 'alphabet'.
People with asthma tend to have a higher upper frequency limit. Myself and another person back in the highschool days were able to detect sinewaves of 22kHz and 23kHz respectively, as an upper limit. Turns out both have asthma, and the connection was verified later on, though I can't remember the publication.
Early Russian, and later other nationalities', experiments show measurable brain activity resulting from frequencies as high as 250kHz. Air-borne sine wave without lower frequency components, but I'm not sure if they're talking about headphones or speakers.
Danish scientists operate with ~14Hz as the limit where our perception transitions from a continous tone to discrete pulses. Research is under way to map out infrasonic hearing thresholds.
Lastly, remember that secondary effects, such as tactile input, all contribute to the total experience. For example, at low frequencies, headphones tend to seem unsatisfactory, due to the absence of tactile input. Stax earspeakers are better at this than most, since your head picks up the tactile bits.
I'd take statements to the effect that there is a world of difference between the presence or absence of supersonic signals with a pinch of salt, but similarly, discounting the effect of supersonics is rather silly. Especially in light of 100kHz ultrasonic carriers for spot-speakers giving tinnitus (granted at extreme spl's)..
However, the need for flat response is another matter. Air absorbs high frequencies, dependent on distance and humidity. ISO recommends reducing high frequency content above 10kHz with a gentle slope (6dB/oct, IIRC) for realistic output. Now, if people would just trade away the flat response in the top for a wider bandwidth, we'd be getting somewhere
PS: As to SACD, the format is not entirely flawed, but the design of a new medium should really be left to dCS or somesuch. 1MHz 4-bit would suit me just fine, and be easy to do computations on.. One of the cutest advantages of DSD is their potential applications in directly driving electrostatic headphones or speakers; let the air do the lowpassing.
My personal preference leans heavily to ribbon tweeters. All dome tweeters I've heard, without an exception, sound smeared and 'heavy' compared to ribbons. Domes are not unmusical, they just lack detail in comparison. Newform ribbons offer good value for the money, and sound very good indeed (I have owned a pair). Ravens are exquisite.
Another very interesting driver that extends to high frequencies is the Manger. OMG S E A M L E S S ! I personally have watched a year go by deciding (failing to decide) whether to purchase Raven 3 or Manger, and soon will be like the donkey who died wondering which bale of hay to eat.
Another very interesting driver that extends to high frequencies is the Manger. OMG S E A M L E S S ! I personally have watched a year go by deciding (failing to decide) whether to purchase Raven 3 or Manger, and soon will be like the donkey who died wondering which bale of hay to eat.
5th element said:
The math is really quite simple. Try calculating the phase shift at, say, 5kHz, given an analog brickwall filter at 20kHz. Same reason why you really want valve output transformers to have decent response below 30Hz..
There is mathematic proof that given unlimited resolution, no samples above the Nyqvist frequency, and perfect antialiasing on the output, two samples per cycle are enough to reproduce steady-state signals. Sounds to me like a set of assumtions that have little to do with what I see when I analyze music waveforms.
Actually, this probably has more to do with lower distortion. Low frequencies are rather steady-stateish, so the number of samples there isn't as relevant as for higher frequencies.
In general, we need good frequency localization but poor time localization for low frequencies, and poor frequency localization but good time localization for high frequencies. Incidentally, wavelet based upsampling, like that used by dCS, preserves this quality, and even improves upon it, since all the information can be processed while it still behaves according to theory, and then be put out into the 'real world' with higher resolution, less jitter, etc.
Probably.
I worked for a pro - audio company, and the bandwidth of their mixing desks was 300kHz, although the microphone inputs at the highest gain was limited to 100kHz. This was deemed necessary to preserve the fidelity, and fellow staff assured me that the difference is real, but I never heard a demonstration myself.
It is not easy to preserve that bandwidth throughout all stages of a mixing desk, so they must have had good reason to do it. Even the transformer isolated stages had this bandwidth - I used to measure them.
It is not easy to preserve that bandwidth throughout all stages of a mixing desk, so they must have had good reason to do it. Even the transformer isolated stages had this bandwidth - I used to measure them.
angel said:
Hi,
Depends what is meant by "upper hearing limit" as normal hearing tests but beg the question. I've been tested for my high frequency hearing limit which, on the test, was about 17KHz. The "test" was but a tone test using earphones. I suspect real-world hearing concerns more than tones and that so-called beyond-limit high-frequency soundwaves impart a feel to sounds in the so-called audible range.
Tom
planet10 said:
Oh, yeah. That kind of reminds me..
The 250kHz test was, as far as I can recall, for a monophonic source. Time domain information for monophonic sounds is less detailed- recall that the brain is more concerned with the relative timing between ears than the exact time localization.
It also bears mentioning that we have about as much physiological grounds for detecting supersonics as we do for frequencies < 60Hz, which is the lowest tuning frequency of any detector in the ear. Could account for some of the abysmal sensitivity.. Though at low frequencies, I guess tactile takes over.
Hmm... Excuse me while I go build a 1st order 60Hz crossover between my earspeakers and my subwoofer
There's life after 20KHz
Have a look at this study by James Boyk of Caltech. It's all in the harmonics I guess .
http://www.cco.caltech.edu/~boyk/spectra/spectra.htm
Have a look at this study by James Boyk of Caltech. It's all in the harmonics I guess .
http://www.cco.caltech.edu/~boyk/spectra/spectra.htm
Anotherthing that you guys are forgetting about in terms of ultrasonics is sideband frequencies- if you mix a 1k sine with any other higher frequency sine wave that is not an overtone you will generate significant amount of audible frequencies above and below!!! 1k. This can easily be heard/measured on a hammond organ- Hammonds have an equel tempered overtone series (as there are only 91 tone generators there are not enough to create all just tempered overtones) The sound of a hammond playing a middle C is much lower than a piano playing one even if the hammond is adjusted to have more upper order overtones than a piano does- This is because the overtones are essentally "out of tune" past the first one a just 5th is a little more sharp than a even tempered one)-
Some argue that this does not matter as the lower audible sidebands will still be recorded in the initial process- but there are would be sidebands that will be lost due to left to right panning
Some argue that this does not matter as the lower audible sidebands will still be recorded in the initial process- but there are would be sidebands that will be lost due to left to right panning
Oopps accedentally hit enter too many times there!!!
Anyway- I also think that DVD-a is much better than CD pcm as it is 24 bit!!! This is easily worth any negative effects of ultrasonics (if there really is one)
I can tell you all that after years of making recordings in the digital domain that 24 bit is a very noticeable upgrade!! It is rediculous to use any thing less as a production format but if you like music with uncompressed dynamics then 24 bit is the only way to go- A signal that is -15dB on a 16 bit system will only have the resoloution of 14bits or so- if you like classical music to be untouched (with a compressor or limiter at least) you are dealing with a massive dynamic range could be 60dB - 60dB on a 16bit system is like internet audio.
Anyway that's my 2c.
BTW: I love this forum!!!
Anyway- I also think that DVD-a is much better than CD pcm as it is 24 bit!!! This is easily worth any negative effects of ultrasonics (if there really is one)
I can tell you all that after years of making recordings in the digital domain that 24 bit is a very noticeable upgrade!! It is rediculous to use any thing less as a production format but if you like music with uncompressed dynamics then 24 bit is the only way to go- A signal that is -15dB on a 16 bit system will only have the resoloution of 14bits or so- if you like classical music to be untouched (with a compressor or limiter at least) you are dealing with a massive dynamic range could be 60dB - 60dB on a 16bit system is like internet audio.
Anyway that's my 2c.
BTW: I love this forum!!!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.