Sure, and if you look closely enough, you’re guaranteed to find differences that can’t be heard. That’s probably also true if you’re comparing analog speaker signals that originated from two completely different digital sources that started with the same digital signal (e.g. a Sony CD player and a Philips CD player both playing the same CD).
But there might be a sound whose fundamental is within the range of human hearing, but which has a detectable (with the right equipment) overtone at 50 kHz.
Recording music onto DAT from a digital source, which could be a CD, works great. I have done it loads of times. If the source was at 44.1 kHz, it stays that way, being an exact copy of the audio. What is absolutely true is that you cannot use a DAT deck to record a CD digitally; you’ll want a CD burner. There could be SCMS somewhere, but that was ignored by all equipment I ever used.
I have seen some good (expensive, but for music, not scientific equipment for bats) speakers reproduce up to 23,500 or 24,000 Hz, according to the datasheet.
I guess the idea is that humans can’t (normally, anyway) hear it, and certainly not directly as a pitch. However such tones do occur in natural sounds, and the right equipment can detect/record them.
I do recall, from a couple of decades ago at least, the implication that CD sampling at 44.1kHz required a hard cap on frequency. So the audio has to pass through a filter before further processing. Since the hard cap is close-ish to frequencies we can hear, that filter could distort the audio in those frequencies.
Thus, bumping the sample rate to 96kHz would minimize the impact of this issue, if it was actually a real issue to start with.
What sound format are you getting on your hi-fi music streams? MP3 goes up to 48 KHz, which is already more than 44.1; then again Tidal.com (for example) advertises “up to” 24-bit, 192 kHz. Random Youtube music video I checked was 48 kHz Opus.
This is true and I was going to make this same point. But my criteria for “indistinguishable” is within the parameters of human perception, so any loss or distortion above 20 KHz is immaterial.
My requirement is that the quality can theoretically be detected by humans. But I want to use a technical analysis rather than humans to test it, since humans are so variable.
I know you put that out there because of sampling rate, but MP3 is lossy. I do not know the technical details of how MP3’s are compressed but right out the gate I’d bet an MP3 file can be determined to be lower quality than an audiophile-quality analog recording. (I personally cannot hear the difference between an MP3 and a CD, but I might be able to if I did an A/B test on high-end playback equipment. But I have hearing loss in the upper frequencies so who knows.)
And this notion is the central problem. You’re asking for a square circle.
We do not have a robust understanding of what features of complex sounds humans detect and process. Yes, we have a pretty good idea of total frequency range for average and outlier-good humans. But all the rest of what constitutes nuance, timbre, emotion, etc., are not things that can be measured by a machine since we don’t know how humans create those perceptions from the raw time-varying pressure wave entering their ears.
And those are the things that you’re asking us to machinify even if you don’t quite recognize that yet.
A quick summary on why vinyl is inferior to digital audio:
-
Produces large amounts of subsonic rubbish that needs to be filtered out.
-
Lots of distortion, especially towards the end of a side.
-
Wow, because even if the turntable is perfect, the hole in the record can only be placed so accurately.
-
Flutter, as the arm holding the cartridge moves up and down but does not follow surface irregularities because of the resonance between cartridge compliance and arm+cartridge mass. The stylus therefore moves back and forward in the groove, frequency-modulating the signal.
To my mind this is the worst feature of vinyl tech. -
And the recording deteriorates every time you play it.
I can’t find the direct cite right now, but I recall reading that during the Pink Floyd recording of Shine on You Crazy Diamond, that David Gilmour had his telecaster jacked directly into the mixing/recording board, it was not recorded via an amplifier and microphone, so it was just a pure analog signal from guitar to tape. Unusual for the time, but common now to skip the whole “amp & mic” portion of recording.
All I want to compare are two raw time-varying pressure waves created by two methods. I am not interested in the wetware response to it other than using human perception to determine the physical parameters for comparison. That is, I don’t care if one can reproduce sounds at 50 KHz and the other cannot.
Ha! Hahahahaha.
The winning move is not to play.
I remember when CDs first came out and I was able to hear quiet parts in songs that I had never heard before. Wasn’t that when people first noticed Glenn Gould singing while playing?
Anyway, CD specs through some kind of wow and flutter machine, with a compressor to bring the dynamic range back to 80 db or whatever the best vinyl is, would make it indistinguishable, I guess. Maybe not a compressor, but something that rolled off the quiet and loud parts, so you can’t hear Glenn singing.
If you have a special room with a big microphone array and expensive test equipment, I am sure you will be able to distinguish the characteristics of different audio equipment like different speakers and different amplifiers. As I suggested in my first post, I do not think digital (or not) signal processing is the limiting factor here. It can be done very wrong, but I believe that proper use of digital technology can serve to minimize distortion and maximize the recording quality.
Since this has ran quite a bit in GQ and most of the direct answers seems to have been done, I wanted to add some what of my own experience that may or may not mean anything as this was while tripping on shrooms that I did about a year ago. As I understand it, it is common to experience a sense of time changing, and for me time slowed down. I was listening to music and it appeared to be playing slower and I started to notice digital artifacts using high bitrate streaming music. It got really bad as things intensified and the music sound started to ‘break up’ and no longer enjoyable. Switching to CD music they were a lot less artifacts, I can’t say none as it was a while ago and I don’t really remember as I was not focusing on that, but I recall it was a lot better. I did wonder how a record would be. But to me it does seem like we may be capable of sensing digital artifacts but typically ignore it or don’t use the full ‘sampling rate’ we may be capable of, as it is normally not needed and that may vary per person.
What you are describing are the anti-aliasing and reconstruction filters. These are a requirement for any sampled stream. Nyquist showed that in order to correctly sample a given bandwidth that the sample rate must be at least twice that of the bandwidth. So 20kHz needs 40kHz sample rate. But you must ensure that the bandwidth fed to the sampling mechanism is only what you want. Otherwise frequencies outside the band will heterodyne with the sample clock and fold back into the sampled signal - appearing as aliases of their original frequency.
But using a 40kHz clock on 20kHz is an impossible constraint, as you cannot slice off the band at exactly 20kHz. There needs to be room for a physically realisable filter to remove the unwanted frequencies.
Next, you must provide an equivalent filter on the reproduction side. This is sometimes incorrectly also called an anti-aliasing filter - but is more correctly called the reconstruction filter. It is required to form the mathematically exact reconstruction of the sampled signal.
CDs at 44.1 kHz made this a difficult task. Engineers talk of brick-wall filters. And early CD players had some fairly evil filter sections. But even here the result was usually only some very minor ripple in the frequency response at very high frequencies - where the ear is pretty much insensitive to such small errors, or a very slight drop in frequency response at the very high frequencies. But this was enough for the purists to get excited.
However this is only a problem if the sampling and reconstruction is done at the base sample frequency. One of the neat things about the operation of an ADC or DAC is that they are just mathematical constructs, and there is a lot of wiggle room as to how that are realised. Modern DACs run with very high sample rates and do a lot of the needed work in the digital domain, leaving room for very simple and easy to implement analog bits. This was termed oversampling, and appeared pretty early on. Most CD players by the late 80’s had oversampling DACs.
There is stupid stream in modern audiofoolery building non-oversampling DACs. Often seeking early DAC chips and building DACs that run at the base sample rate. They claim all sorts of magical powers for these designs. However fidelity is unlikely to be one of them.
The same exact physical CD played on the same equipment 6 hours apart will produce a measurably different recording by your technique.
The problem is that while you could create a gap-plot of the time-varying delta between the two recordings, you would have no way of knowing whether that matters to a human.
And you could play the same CD on the same gear in the same studio 5 more times over the next week and now you have 7 samples and 6 different delta plots. Which one represents the “reality” of the difference between what’s really the same thing?
Your obstacle is not technological. It’s epistemological.
I am not interested in distinguishing characteristics of the playback equipment but of the source. A turntable and a [whatever playback device we would use for my imaginary recording technique] through the same amp and speakers.
“Superior” is in the ear of the beholder. Solely in terms of sound fidelity? No, analog will never beat digital at modern sampling rates and bit depth.
However one could argue that the enjoyment of listening is more than sound coming out of a mechanical hole and going into your earhole. Consuming someone else’s art is a social activity between you and the all the artists involved, even if you’re not actually with anyone.
So if you find it adds to the experience to hear the occasional pop or crackle, if you enjoy handling vinyl and manipulating a phonograph, looking at the cover art, sorting and caring for your collection, then these are all “superior” to digital in terms of subjective experience. A Stanford professor did a study about 15 years ago which surprisingly found that students preferred the degraded, compressed sound found in .mp3 format over higher-fidelity digital. But that was in 2009, and .mp3 files aren’t so popular anymore, so tastes may have changed.
Anyway, subjective sound experience is in the ear of the beholder.
People can’t hear ultrasonic pitched sounds- at least not directly. But doesn’t their interaction with other tones via a heterodyning process yield a difference one can hear?