Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Feb 26, 2018 13:31:50 GMT -6
I normally record everything at 192K and I've been trying to figure out why I like it better. Although I've taken very good care of my ears, I'm still in geezer territory and age-related loss is an inevitability. If I ever heard 20K, that was a long long time ago. So why do I think it sounds better? Here's what I've got so far. 1) Aliasing (this is to the point Bob make a few posts back). Aliasing is what happens when any frequency above nyquist is converted. Sonically it's a lot like amplitude modulation in the sense that it reflects back into the audible band. And of course that schmutz gets processed through EQ, compressors, reverbs, gain control and the injury is compounded. Good filters on the converter help minimize the amount, but it really helps if the majority of that signal is a couple of octaves above anything you hear. 2) Spacial resolution. Depending on a speaker's ability to deal with transients, it's quite possible that higher sample rates will give you more accurate arrival times for a signal, resulting in a better sense of placement. 3) Filters in the DAW or plugins. Most IIR filters (the vast majority of filters in the audio world) will have slightly different responses depending on sample rate. Action around the passband is pretty much the same in all cases, but the filter's curve will typically extend up to nyquist. If nyquist is higher, the response of the filter is slightly different. Most of this effect is at very high frequencies, but it could help to add a sense of openness.
Of these, I think that the elimination of aliasing components is probably the most helpful, since those components end up all over a mix. Best not to have them. And oddly enough, the recordings I make at 192K sound better downsampled to 44.1 than they would have if they'd originally been recorded at 44. That's probably still an artifact of the aliasing thing, but it may also indicate that the software in a good DAW (usually running in non-realtime) can do a better job than you'd get with the filters in your converter.
|
|
|
Post by Martin John Butler on Feb 26, 2018 14:18:36 GMT -6
Hearing isn't everything. I believe we sense frequencies we can't technically hear. Sound is vibration, and our bodies, inside and outside are sensitive to that. So something might feel better, even if we can't say exactly why yet.
I'll try to complete one session in 96 and see how it goes..
|
|
|
Post by pope on Feb 26, 2018 16:25:47 GMT -6
I believe that the maker of this video is actually a RGO member
|
|
|
Post by Tbone81 on Feb 26, 2018 18:39:28 GMT -6
Hearing isn't everything. I believe we sense frequencies we can't technically hear. Sound is vibration, and our bodies, inside and outside are sensitive to that. So something might feel better, even if we can't say exactly why yet. I'll try to complete one session in 96 and see how it goes.. If I may expand on that thought a little...it's intriguing to me that we don't really hear with our ears, we hear with our brains. Sound is changes in vibration and pressure, our ears turn that into nerve impulses but our brains interpret it. We could just as easily see sound and hear colors if our brains were wired a little differently. So to your point, how you "feel" and "sense" sound is definitely more than what can you physically hear. So much so that your brain will create auditory illusions in many cases.
|
|
|
Post by Martin John Butler on Feb 26, 2018 23:25:58 GMT -6
I used to get so much pushback on audiophile forums in the 90's regarding digital vs. analogue, and digital in general. I keep saying there's distortions we're not completely aware of, and others would say it's all zeros and ones, essentially calling me idiot. That stopped when jitter became a known and accepted factor.
It's always the same, know it alls can prove something I say I can hear or feel is wrong, and then science proves me right later. These days I keep my own advice when it comes to what I'm hearing. I do enjoy learning things from the more knowledgeable cats here though.
Much of listening is training. Once you learn to recognize an effect, it becomes obvious.
|
|
|
Post by svart on Feb 27, 2018 8:51:10 GMT -6
There was a interesting article I read once.. I might try to find it at some point, but the story is about "ghosts" and "hauntings" and what makes people feel uneasy in certain places that people say are haunted.
There was a study of a laboratory in a building's basement that had a history of high employee turnover on 3rd shift(overnight). People would quit and leave on the spot, while others would tolerate it for a while but end up demanding being moved to a day shift, etc.
Eventually, a daytime manager decided to debunk the "haunted lab" by working overnights for some period of time.
The first night, at some point, he started to feel really uneasy and anxious, like someone was watching him from the shadows. He told the story of how the paranoia started to really ramp upwards to the point where he had to leave the lab!
He went back after feeling a bit better, and his anxiety and paranoia came back almost immediately.
From that point on, he decided to do some DIY ghost hunting of his own, and devised a large amount of tests like EMI/RFI sensors, temperature sensors, sound sensors (ultrasonic and infrasonic), thermal cameras, etc.
A long story short, he found that every night around 1am, there would be a huge infrasonic peak around 11-15Hz that would pop up in portions of the lab. Not really loud enough to feel, but loud enough to perceive.
He continued to investigate and found that the factory across the street would start up a certain machine around 1am every night, and it's vibrations would travel through the ground and focus in a certain portion of this basement laboratory.
When the machine wasn't running, nobody felt strange or paranoid, but when it was running, almost everyone did.
So the real reason this happens is because most large predators have fundamental frequencies in their vocalizations (growls and such) that are in the 10-15Hz range. Humans have evolved to still perceive these as a trigger for the "fight or flight" response which we all know increases adrenaline in the blood, which leads to the anxiety and paranoia.
There's also some belief that the fundamental waves in earthquakes are also resident in these frequencies.
Even the military has experimented with infrasonic area denial weapons, and have found them to work, but only in small areas in certain situations and otherwise be impractical.
So yeah, there's something to "perception" that isn't actually "hearing" something..
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Feb 27, 2018 9:13:00 GMT -6
It's important to keep in mind that phenomenae such as infrasonic effects are at the other other of the spectrum from the range that we get from high sample rates. Infrasound's transmission through the body can be accounted for by things like bone conduction. But response to high frequencies require very small resonating bodies, such as the tiny hairs of the inner ear (same reason tweeters are small). And those very high frequencies generally don't go very far before they disappear by air absorption or other effects (like the bones that won't resonate so rapidly).
So I think we have to be very careful in talking about unexplained perception of things as a blanket explanation. It could be correct. It could be just 'woo'. Or it could be confirmation bias.
|
|
|
Post by Bob Olhsson on Feb 27, 2018 12:12:08 GMT -6
Some friends did an experiment using ultra-low distortion filters that took hours to calculate and determined that it is possible to make 44.1 recordings that are indistinguishable from 192k but totally impractical.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Feb 27, 2018 12:51:19 GMT -6
Some friends did an experiment using ultra-low distortion filters that took hours to calculate and determined that it is possible to make 44.1 recordings that are indistinguishable from 192k but totally impractical. As you know Bob, our customers are infinitely patient and will happily wait for that
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Feb 27, 2018 15:29:18 GMT -6
Some friends did an experiment using ultra-low distortion filters that took hours to calculate and determined that it is possible to make 44.1 recordings that are indistinguishable from 192k but totally impractical. Yes, was gonna say, may be impractical now but give it another decade. Gotta love the steady increase in possible fidelity with time. I still use FinalCD for all my downsampling, even though it takes forever, because it sounds better.
|
|
|
Post by Bob Olhsson on Feb 27, 2018 15:49:26 GMT -6
I understand there's no reason we couldn't be using analog computing other than the desire to save money by recycling decades-old binary code that was developed at government expense.
|
|
|
Post by johneppstein on Feb 27, 2018 17:33:34 GMT -6
It's important to keep in mind that phenomenae such as infrasonic effects are at the other other of the spectrum from the range that we get from high sample rates. Infrasound's transmission through the body can be accounted for by things like bone conduction. But response to high frequencies require very small resonating bodies, such as the tiny hairs of the inner ear (same reason tweeters are small). And those very high frequencies generally don't go very far before they disappear by air absorption or other effects (like the bones that won't resonate so rapidly). So I think we have to be very careful in talking about unexplained perception of things as a blanket explanation. It could be correct. It could be just 'woo'. Or it could be confirmation bias. They are likely a number of different causes for unexplained perceptions of differrent types.
|
|
|
Post by joseph on Feb 27, 2018 19:21:56 GMT -6
|
|
|
Post by christopher on Feb 28, 2018 2:15:53 GMT -6
To me this stuff is similar to my brother in law's inablity to lose his accent, even though he's been speaking English in America everyday for last 45 years. All this time, yet he can't seem to hear his own accent and correct his speech to match everyone else's? Why not? Maybe he isn't really listening close enough, or maybe he's not even considering that it's possible his accent could be heard as incorrect and be re-adjusted? Yet actors who work hard can learn to adjust their accents, so it must be possible to both hear an accent and reproduce it. If we conclude that something isn't possible, and further conclude that it's not worth trying, then we certainly never advance any further. When I see people tell others it's not possible for them to hear something, I imagine my brother-in-law telling me that he doesn't have an accent, which would be pretty hilarious.
|
|
|
Post by svart on Mar 1, 2018 13:08:04 GMT -6
I understand there's no reason we couldn't be using analog computing other than the desire to save money by recycling decades-old binary code that was developed at government expense. Care to elaborate on this?
|
|
|
Post by Bob Olhsson on Mar 1, 2018 13:48:06 GMT -6
I first heard about it from a Cray VP at a conference around 2000.
|
|
|
Post by adamjbrass on Mar 8, 2018 5:34:40 GMT -6
sample rates are like jeans...Not every cowboy can fit into the same pair....
|
|
|
Post by Ward on Mar 8, 2018 8:35:39 GMT -6
sample rates are like jeans...Not every cowboy can fit into the same pair....
|
|
|
Post by johneppstein on Mar 8, 2018 13:17:24 GMT -6
I've owned converters that made cymbals sound like indistinct trash can lids at 44K1, and fine at 88K2+. I don't hear that sort of difference with my current setup. I've found several times that the debater shouting loudest about 44K1 being fine for everything was only using virtual instruments and samples for everything they did, and had zero experience working with audio they'd captured themselves. They usually also were 100% ITB, no analog paths. And I'd go farther and speculate that most or all of the loops they're using were originally recorded at 44.1 and probably the VIs had a native sampling rate of 44.1 as well. Note that libraries sampled at higher rates are MUCH larger...
|
|
|
Post by johneppstein on Mar 8, 2018 13:20:26 GMT -6
sample rates are like jeans...Not every cowboy can fit into the same pair.... Yeah, tell me about it.... no, don't bother - I already know.
|
|
|
Post by cowboycoalminer on Mar 8, 2018 16:26:24 GMT -6
I haven't read through the entire thread (I will), but I've been down this rabbit hole before, came out, went back in, and on and on. What I've personally discovered are none of my mixes are good enough to discern a difference in sample rate. I'm back to 44.1 to stay. Higher rates are just way too taxing on a system with very little return that I can hear in a final product.
On a side note, I of coarse don't do this for my sole income. If I did my reasoning might change. But I will say that after listening back, some of my best stuff was done at 44.1. It's all a matter of opinion and whatever floats your boat.
|
|
|
Post by cowboycoalminer on Mar 8, 2018 16:32:54 GMT -6
I do everything I can at 96 because multiple over-sampling within plug-ins, i.e. up/down, up/down, up/down, up/down, up/down, etc. where latency is generally optimized over precision makes no sense. The entire process from the microphone to iTunes needs to be treated as a system. I will say this post is very interesting concerning plugins. And I'll add to my above post concerning this, my best sounding mixes were also done printing everything through hardware at 44.1. I'm quite sure that makes a difference.
|
|
|
Post by Bob Olhsson on Mar 8, 2018 18:19:00 GMT -6
I did a lot of experimenting before moving to 96 because it meant investing in a new computer.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Mar 9, 2018 2:25:17 GMT -6
Bob, what do you use for upsampling client files to 96 if they come in at a lower sample rate? I'd like to give it a go on a few projects (with all oversampling turned off in the chain) to hear if I can notice a difference. I have the iZoptope SRC in RX5 Advanced, care to share your settings if that's the one you use?
|
|
|
Post by Bob Olhsson on Mar 9, 2018 16:57:12 GMT -6
Saracon.
|
|