What complete twaddle.
What conceited nonsense. A kick drum generates very little above 1kHz in fact - do you know what a kick drum actually is? Soft heavy fluff ball hits drumskin at a few m/s... It might have been a bit more believable if you'd said rimshot on a share, at least that's got a sharp crack.
Kickdrum: no shock waves are generated or anything like. Do you know what a shock wave is? That's what explosives generate, you have to have motion faster than the speed of sound to do that. Cracking a whip is a safer way. In a shock wave the air has a discontinuity in temperature, pressure and velocity. That rapidly tails off to a loud noise with distance (a real shock wave will destroy most sensitive microphones instantly, they split ear-drums for instance).
Air attenuates MHz ultrasound extremely efficiently. > 100 dB/m or loss. So you aint seeing anything like that getting to the microphone, and the microphone diaphragm is a very good low pass filter too, as is the preamp after it, unless its a microphone designed for ultrasound, when you might get response upto 50kHz.
If it wasn't continuous how would your eardrum follow it? Anyway the point is human hearing goes up to around 20kHz in some young individuals, and beyond that there's nothing. Really there isn't, people have played loud ultrasound at people and tested this, there's no response at all (I guess until you start to feel the heat or the air goes non-linear at high SPLs - that's almost certainly not good for you!)
So you could ABX or double-blind detect this? What would you be detecting though, the higher resolution or the higher bandwidth? Several ABX tests you can try there.
A wider bandwidth than 20kHz is needed.
There's life above 20 kilohertz! A survey of musical instrument spectra to 102.4 kHz
There's life above 20 kilohertz! A survey of musical instrument spectra to 102.4 kHz
That's not clearly asserted by the paper. It also comes down to whether we can perceive it or not. It deserves further research though, there are some cited papers I would love to read on the subject but not sure if I could access them (I'll have a dig later in my own time)
Last edited:
Yes instruments put out ultrasound (rattling some keys is a really good way too), but the ear can't detect it at all. It really can't. Get someone to beam loud 30kHz ultrasound at you. All you can hear is the clicks when it turns on and off, when the waveform has a broad range of frequencies.
The ear acts as a low-pass filter, the eardrum and ossicles have mass, they cannot respond as much at higher frequencies, and the hair cell receptors on the organ of Corti are sensitive to a narrow band of frequencies due to the tuned nature of the thing. Hearing sensitivity drops like a stone above about 14kHz (age dependent!), around 13kHz there can be a little more sensitivity than 10kHz due to the third overtone resonance of the cochlea canal itself.
Individuals do vary of course, some lucky few may get to hear above 20kHz, but that will be because of the structure of the ear being different for them.
If you want your pets to enjoy the best hifi sound reproduction, ultrasound response will matter, but not for humans.
In theory, 16 bit 44.1kHz is adequate for music, no doubt about that. But in reality, I think the 24bit-48kHz or 24bit-96kHz is much better. Problems is the music production. Nowadays, the mixing/EQ/mastering in music production is done on DAW, like Merging Pyramix, Cubase with internal processing at 32bit/64bit 352kHz, then the final release is a downsample of the master file (DXD (32bit 352kHz) or PCM 24bit-96kHz).
Here is my practical experiences with Merging Pyramix and Weiss Saracons, downsample from master file DXD 32bit 352kHz to 24bit 96kHz or 24bit 192kHz with TPDF dither yields no difference in hearing test.
But downsample from master file DXD 32bit 352kHz to redbook format 16bit 44.1kHz makes a significant difference in listening test compared to the master file. And with different dithering algorithm (TPDF, POWr1. POWr2, POWr3), the sound is different as well. And to my ear, the 16bit-44.1kHz downsample version, no matter what dithering algorithm was used, sounds worse than the master DXD files.
So for me, if the music is done or remastered recently and available on both redbook and HiRes release, I will choose the HiRes format, at least with classical music.
Here is my practical experiences with Merging Pyramix and Weiss Saracons, downsample from master file DXD 32bit 352kHz to 24bit 96kHz or 24bit 192kHz with TPDF dither yields no difference in hearing test.
But downsample from master file DXD 32bit 352kHz to redbook format 16bit 44.1kHz makes a significant difference in listening test compared to the master file. And with different dithering algorithm (TPDF, POWr1. POWr2, POWr3), the sound is different as well. And to my ear, the 16bit-44.1kHz downsample version, no matter what dithering algorithm was used, sounds worse than the master DXD files.
So for me, if the music is done or remastered recently and available on both redbook and HiRes release, I will choose the HiRes format, at least with classical music.
There is a very big difference between 16bit and 24 bit, and I think most people can hear the difference.
But being able to hear the difference will depend heavily on the system you are listening to.
Look at bit depth in relation to signal resolution here:
Audio bit depth - Wikipedia
It does not matter at all if you have 24bit signal, 24bit capable dac, and whatever else in the signal chain.
You must have a complete signal chain capable of better than 110db S/N to have an easily discernible difference.
If you're using a 100db THD+N amplifier even with everything else of better quality, then you'd most likely have a very hard time telling the difference.
Very good 16bit equipment get around 95db theoretical signal to noise ratio, very good 24bit equipment still struggle to get better than 119db signal to noise ratio.
It's very difficult to maintain signal integrity with such a high standard, designing circuits like that is no walk in the park.
But being able to hear the difference will depend heavily on the system you are listening to.
Look at bit depth in relation to signal resolution here:
Audio bit depth - Wikipedia
It does not matter at all if you have 24bit signal, 24bit capable dac, and whatever else in the signal chain.
You must have a complete signal chain capable of better than 110db S/N to have an easily discernible difference.
If you're using a 100db THD+N amplifier even with everything else of better quality, then you'd most likely have a very hard time telling the difference.
Very good 16bit equipment get around 95db theoretical signal to noise ratio, very good 24bit equipment still struggle to get better than 119db signal to noise ratio.
It's very difficult to maintain signal integrity with such a high standard, designing circuits like that is no walk in the park.
I am very happy with the further postings, some including details about professional music recording.
While the large majority of consumers will continue using the CD format or whatever is better but at least not more expensive, there is a (smaller) group of more audiophile minded people who may support extended formats if they experience a SQ improvement and if they understand why. Maintaining a full audio chain of that quality (and cost) will be a further demand on them but they are typically ready to spend much on the gear.
You hold about the ultimate technical knowledge and it by far exceeds mine such that I cannot judge who is wrong and who is right. But, it seems to me that in particular the present and future producers have not yet come up with the rock-solid arguments for why we should pay-up more expensively to replace our music collections. They seem still to rely on the higher figures (16/44.1 -> 24/192) being a sufficient argument in themselves, and there I believe they may be wrong with that critical (and somewhat difficult) group of potential customers.
Please continue this interesting discussion.
While the large majority of consumers will continue using the CD format or whatever is better but at least not more expensive, there is a (smaller) group of more audiophile minded people who may support extended formats if they experience a SQ improvement and if they understand why. Maintaining a full audio chain of that quality (and cost) will be a further demand on them but they are typically ready to spend much on the gear.
You hold about the ultimate technical knowledge and it by far exceeds mine such that I cannot judge who is wrong and who is right. But, it seems to me that in particular the present and future producers have not yet come up with the rock-solid arguments for why we should pay-up more expensively to replace our music collections. They seem still to rely on the higher figures (16/44.1 -> 24/192) being a sufficient argument in themselves, and there I believe they may be wrong with that critical (and somewhat difficult) group of potential customers.
Please continue this interesting discussion.
Last edited:
Just to get a better perspective on Dynamics and SNR, you can refer to this chart
Assuming a room with noise close to hearing limits (a recording studio is typical in the 20/30 dB range), 96 dB of dynamic is already in the "hearing damage from prolonged exposure" zone; 120 dB is close to the threshold of pain ...
When 99% of music hardly reach 18 dB of dynamic (average power vs peak power) what is the point of 24 bits ? Moreover, the theoretical 96 dB is a sum across all spectrum: using noise dithering/shaping local OSNR can go down to more than 110 dB meaning a sinusoidal signal with 1/4 bit amplitude can be reproduced with enough SNR with 16 bits.
Assuming a room with noise close to hearing limits (a recording studio is typical in the 20/30 dB range), 96 dB of dynamic is already in the "hearing damage from prolonged exposure" zone; 120 dB is close to the threshold of pain ...
When 99% of music hardly reach 18 dB of dynamic (average power vs peak power) what is the point of 24 bits ? Moreover, the theoretical 96 dB is a sum across all spectrum: using noise dithering/shaping local OSNR can go down to more than 110 dB meaning a sinusoidal signal with 1/4 bit amplitude can be reproduced with enough SNR with 16 bits.
There is actually an argument from the "industry" to keep hanging on to 16bit 44.1khz:
It requires less bandwidth.
This is a critical argument towards availability. In this age of streaming, higher bandwidth induces higher costs on servers, storage space, network speeds, number of concurrent users etc.
So part of the audio industry, namely the various streaming services. Heavily benefit from marketing anything higher than 192kbit/s as a high quality service.
Sales at 24bit/48khz or better is a relatively small niche market. Most consumers have no quality preference. Add this to the demands of the signal chain on each and every system to defend the higher quality reproduction of sound, and you end up with the major part of 24bit consumers purchasing high quality files purely on hype.
edit:
enryfox, 16bit really does not compare well to 24bit, dithered or no.
Dithered 16bit may have pretty numbers, but there is a significant lack of resolution on peaks. If the entire recording is lower than 0.4-0.6v (from 2v peak, as in max you get out from a regular CD player) then sure, dithering is wonderful.
Regular 16bit is better at higher level signals.
Proper 24 bit recording offer the best of both 16bit and 16bit dithered and then some.
Dithering is not magical, you still only have 16bits of resolution, something must be sacrificed.
It requires less bandwidth.
This is a critical argument towards availability. In this age of streaming, higher bandwidth induces higher costs on servers, storage space, network speeds, number of concurrent users etc.
So part of the audio industry, namely the various streaming services. Heavily benefit from marketing anything higher than 192kbit/s as a high quality service.
Sales at 24bit/48khz or better is a relatively small niche market. Most consumers have no quality preference. Add this to the demands of the signal chain on each and every system to defend the higher quality reproduction of sound, and you end up with the major part of 24bit consumers purchasing high quality files purely on hype.
edit:
enryfox, 16bit really does not compare well to 24bit, dithered or no.
Dithered 16bit may have pretty numbers, but there is a significant lack of resolution on peaks. If the entire recording is lower than 0.4-0.6v (from 2v peak, as in max you get out from a regular CD player) then sure, dithering is wonderful.
Regular 16bit is better at higher level signals.
Proper 24 bit recording offer the best of both 16bit and 16bit dithered and then some.
Dithering is not magical, you still only have 16bits of resolution, something must be sacrificed.
Last edited:
Dithering is just a way to shape the quantisation noise so that it less concentrated only on frequencies but spread across the entire spectrum. And that allows faint signals, smaller than one bit in amplitude, to be discernible; it's not magic, it's math.
I'm not saying that 24 bit is worse, I'm just saying that more does not always equally with "certainly better". On the production side of music, 24 bits makes a lot of sense; for the simple fruition of music I think it is mostly a placebo effect.
I find this article to be very interesting; it's from 2012 meaning the 24 bits dispute has been going on for quite some time.
There is some evidence here Can you hear a difference between 2 solid state preamps? that a difference can be heard between high bandwidth components even on low bandwidth systems
Enryfox, All your points are valid. Only thing is, I have not seen a single 16bit DAC that claimed better than 98db S/N, and as to THD+N... Well, nothing over 95db or so.
Perhaps you have knowledge of better 16bit hardware.
Even most 24bit hardware is 100-108db S/N, quite good stuff get higher. But the absolute best I've seen gets 119db THD+N, pricey! If you know any better, I'd like to know.
Matt, I guess your point is about Dynamic Range, this is frequently higher than S/N ratio. Not saying it's impossible to hear the difference through a lower S/N chain of gear, but it gets significantly harder to do so the more inherent noise there is.
Perhaps you have knowledge of better 16bit hardware.
Even most 24bit hardware is 100-108db S/N, quite good stuff get higher. But the absolute best I've seen gets 119db THD+N, pricey! If you know any better, I'd like to know.
Matt, I guess your point is about Dynamic Range, this is frequently higher than S/N ratio. Not saying it's impossible to hear the difference through a lower S/N chain of gear, but it gets significantly harder to do so the more inherent noise there is.
Oh, yes.
Now we're talking about sampling frequency. Higher sampling frequencies in general, help reduce audibility and possible disturbance from the post processing filters.
But my somewhat limited experience tells me there is no difference between storing audio in 24bit 48khz vs 24bit 96khz, only difference being: the 48khz file is half the size.
In signal processing however, there can be significant advantages to using 96khz samplerate. But again, the difference between upsampling from 48khz audio vs native 96khz = zero.
I welcome any views or data that says otherwise, it is an interesting topic.
Now we're talking about sampling frequency. Higher sampling frequencies in general, help reduce audibility and possible disturbance from the post processing filters.
But my somewhat limited experience tells me there is no difference between storing audio in 24bit 48khz vs 24bit 96khz, only difference being: the 48khz file is half the size.
In signal processing however, there can be significant advantages to using 96khz samplerate. But again, the difference between upsampling from 48khz audio vs native 96khz = zero.
I welcome any views or data that says otherwise, it is an interesting topic.
Actually well designed 1 bit DAC claims 110 dB of SNR and > 100 dB of dynamic. An example is the old Philips TDA1547
Anyway that is mostly academical, enjoying such huge dynamic is quite complex (very good equipment, super silent environment...) and recordings have to be made specifically to take full advantage of such high resolution.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.