I'm hardly grumpy, I'm realistic, and I'm reasonable.
You're comparing me to the newer, obsessed-with-perfection audiofools.
Of which I am not.
I do get some amusement though, of the ones who literally nit-pick things apart endlessly on websites.
Complaining about this, that, and everything in between.
Worrying over a single "byte" of information, questioning idiotic things like a tiny clump of stuck-on debri on their phono stylus, and then hoardes of others generating long threads of just how it is to be removed.
Worrying over a fraction of a millimeter of vertical tracking error.... etc etc.
I certainly don't want to be part of that bunch of goons.
So yeah, I'll keep my reasonable way of thinking which fits my lifestyle quite nicely.
@edbarx , To further clarify, the stairstep waveform is called a zero-order hold. The samples themselves are only points and do not have a staircase property.
Often the output of a dac before final filtering is in the form of a zero-order hold. What's left to do at that point is pass the dac output through the analog reconstruction filter. That's the only way to do the final smoothing: in the analog domain.
If we wanted to digitally connect the dots more smoothly then in one sense it doesn't matter if we do a spine-fit or if we filter. Either way we have to upsample to add more points between the original sample points, so we can make a smoother approximation to the final desired analog output (more points = finer, smoother stairsteps). Even if we do that we will still end up with a zero order hold stairsteps coming out of the dac. We will only have made the stairsteps smaller in size, so less final filtering to do in the analog domain. Less HF droop due to the zero order hold FR, as well.
There is no way to completely avoid doing the final filtering in the analog domain, unless we just decide to skip it and live with the HF/RF garbage that shouldn't be there.
https://www.tutorialspoint.com/sign...ts-transfer-function-practical-reconstruction
Often the output of a dac before final filtering is in the form of a zero-order hold. What's left to do at that point is pass the dac output through the analog reconstruction filter. That's the only way to do the final smoothing: in the analog domain.
If we wanted to digitally connect the dots more smoothly then in one sense it doesn't matter if we do a spine-fit or if we filter. Either way we have to upsample to add more points between the original sample points, so we can make a smoother approximation to the final desired analog output (more points = finer, smoother stairsteps). Even if we do that we will still end up with a zero order hold stairsteps coming out of the dac. We will only have made the stairsteps smaller in size, so less final filtering to do in the analog domain. Less HF droop due to the zero order hold FR, as well.
There is no way to completely avoid doing the final filtering in the analog domain, unless we just decide to skip it and live with the HF/RF garbage that shouldn't be there.
https://www.tutorialspoint.com/sign...ts-transfer-function-practical-reconstruction
Last edited:
The popularity of vinyl records could be due to a lot of things: the smooth ultrasonic roll-off and short impulse response compared to 44.1 kHz sample rate digital audio, differences in mastering, the large artwork, the whole procedure of putting a record on a record player, fashion, ...
Last time a visited a mastering studio mainly dedicated to vinyl ( they had a nice Neumann Vms) i spoted some ad/da ( Weiss audio) in the chain.
Asked why they are for : " ...man, the whole things run true the Sadie before running trough the Neumann. Makes life easier for album process ( no need for two times the analog gear and manual setup of gear during other track is printed) and sound better overall as i can perform digital tricks."
Time have changed. For the best imo.
Asked why they are for : " ...man, the whole things run true the Sadie before running trough the Neumann. Makes life easier for album process ( no need for two times the analog gear and manual setup of gear during other track is printed) and sound better overall as i can perform digital tricks."
Time have changed. For the best imo.
Have optical cartridge vinyl playback here with custom preamp. That through dual mono blocks into large panel ESL speakers. Played back Lincoln Mayorga #3 live-to-disk-lathe vinyl the other day. There is no way digital sounds more real than that, at least not any digital I have heard so far. OTOH, playing Hungarian Dances, Wiener Philharmoniker, Claudio Abbado vinyl, its quite clear from the sound it was originally recorded to digital. I heard it not too long ago on another optical vinyl system and could tell it was cut from digital before I was sure what album it was. Had a bit of that grain to the sound.
OTOH, DSD256 can be pretty darn good.
Still think there is something we are missing by overreliance on steady-state time-averaged FFT measurements as though that's all there is to reproduction accuracy.
OTOH, DSD256 can be pretty darn good.
Still think there is something we are missing by overreliance on steady-state time-averaged FFT measurements as though that's all there is to reproduction accuracy.
Last edited:
Maybe it is on topic. I thought we were going off on signals types and digital vs analog tropes.
I have lately been experimenting with Jecklin and other barrier type mics. Mostly with my big fat head as a barrier. The results are interesting, but I'm not yet sure what to make of them.
I've also looked into some line array microphones. I've only heard them, never used them. They don't come cheap. I wonder if a "reverse CBT" (Don Keele) array of mikes would give useful results?
At Record Industry in Haarlem you can either deliver the audio in a high-resolution digital format, or on an analogue tape, or hire their studio full of analogue gear, mostly with valves.
Yes, it is possible to deliver in analog to the studio i talked about too. You can even specify you want the work to be done full analog too but for that you'll have to pay more.
It's more a question of workflow and optimisation of time and gear management as they opened a second 'twin' room ( gear was on trolley and movable).
In my view the analog/digital are just two kind of esthetics. For some kind of music tape brings something for sure. But better use it at takes ( multitrack), mix analogue and fix it on 1/2". Even if you digitalize at one point for editing...
In a way i'm the inverse of Mark4: i can spot when tape was used most of the time, digital i'm not so sure.
It's more a question of workflow and optimisation of time and gear management as they opened a second 'twin' room ( gear was on trolley and movable).
In my view the analog/digital are just two kind of esthetics. For some kind of music tape brings something for sure. But better use it at takes ( multitrack), mix analogue and fix it on 1/2". Even if you digitalize at one point for editing...
In a way i'm the inverse of Mark4: i can spot when tape was used most of the time, digital i'm not so sure.
To my mind it's the next direction for real advancement. Spaced stereo pair miking is the most linear, simple and first run solution imaginable. It completely ignores too many aspects of how we localize.
Kimber's Jecklin disc yields impressive results for a microphone moved with a truck. The link below illustrates some of the directions investigated by these enthusiasts with more practical equipment.
https://www.trackseventeen.com/mic_rigs.html
Actually, my first DDDAC 1543 output trace looked exactly like that. Laughably so. Both I and the engineer whose scope it appeared on got a good laugh out of it. Yet we had both built them and we both liked it a lot. Very engaging sound.
The trace didn't move my appreciation of its musicality at all but It did and still does make me wonder how anything that looked that "bad" could still sound so good.
Though Doede has moved on with the 1794 chip his early design is still a good dac IME.
Probably because of the bandwidth limitations of your ears, you are your own reconstruction filter.
Assuming you can hear up to well above 10 kHz, the attachment will give you an impression of what 44.1 kHz sample rate audio reproduced like that would sound like to a cat. It contains two short music fragments, one bandwidth limited to 5 kHz (Track15exp1OS.wav) and one bandwidth limited to 5 kHz, decimated to 11.025 kHz sample rate and then interpolated to 44.1 kHz again by just repeating each sample four times (Track15exp1NOS.wav).
Assuming you can hear up to well above 10 kHz, the attachment will give you an impression of what 44.1 kHz sample rate audio reproduced like that would sound like to a cat. It contains two short music fragments, one bandwidth limited to 5 kHz (Track15exp1OS.wav) and one bandwidth limited to 5 kHz, decimated to 11.025 kHz sample rate and then interpolated to 44.1 kHz again by just repeating each sample four times (Track15exp1NOS.wav).
Of course, a scope also connects the dots. Monte´s Youtube I linked makes it clear why.
Jan
I suppose perceived recording quality may be down to a whole chain of processes that have very little to do with the underlying reproduction technology.
An artist getting started and just looking to get their tunes out so that they can be played at high volumes in a dance club will probably select the cheapest path. That’s probably a setup with very little customization from mic to mastering. High amounts of compression are used to get the famous “wall of sound”, removing the dynamics of the instrumentation. It’s cheap, and so readily accessible many people choose it.
Then people on this board complain that “all” modern recordings are trash and that it must be down to an all-digital signal chain. Actually, it’s more that very little care is paid to the sound because that’s not what the target audience / consumer wants.
On the other side we have people who pay attention to these factors. They could be recording exactly the same kind of music but make it sound better,. But this might be counter-productive because actually having “good” sound doesn’t fit with the expectations and aesthetics of the musical genre, not fit with the expected environment where it will be heard.
I see people arguing here that digital is inferior because it can’t reproduce high frequencies. But tape has a limited bandwidth as well; more so than digital. With digital we just need to throw more compute power to sample higher and higher frequencies. With tape, there are hard physical limitations to it’s frequency capture. Your recording head can only get so many magnetic particles to move in any given location on the tape — it’s a hard physical process.
Same with vinyl — the cutting stylus was not capable of encoding 20kHz sounds, and the playback stylus is not capable of reading them. So the frequency limitations are actually higher in analog recording than in digital. Most estimates have analog playback topping out between 10 and 15kHz. The difference is that the cutoff is quite a bit smoother, so that it has a natural smoothness, rather than a hard wall. This is a significant factor, to be sure, but it has nothing to do with the frequency capture ability.
Objectively, digital is better in almost every way. It can reproduce frequencies more faithfully, and the sampling rate is far higher than anything the analog signal capture world can produce. Where digital is a bit weaker is in the hard signal roll off, but the high sample rates and advanced signal processing make this negligible for most cases.
Subjectively, you may not like it, and you may not like the music that makes heavy use of it, and that’s ok. But you can’t honestly say that this preference is somehow magnified to be an objective statement about a single component of the overall process. When there are blanket statements like “older recordings are always better because they’re analog” or “analog recording is more accurate” that is not really true. I think it’s more down to the artistic qualities of the recording, the choices made in the process, and what constitutes “good enough” for the target audience. If you think a recording sounds bad to you, it’s probably because you’re not the person the artist wants to reach. what it doesn’t mean, however, is that the technology itself is inferior.
An artist getting started and just looking to get their tunes out so that they can be played at high volumes in a dance club will probably select the cheapest path. That’s probably a setup with very little customization from mic to mastering. High amounts of compression are used to get the famous “wall of sound”, removing the dynamics of the instrumentation. It’s cheap, and so readily accessible many people choose it.
Then people on this board complain that “all” modern recordings are trash and that it must be down to an all-digital signal chain. Actually, it’s more that very little care is paid to the sound because that’s not what the target audience / consumer wants.
On the other side we have people who pay attention to these factors. They could be recording exactly the same kind of music but make it sound better,. But this might be counter-productive because actually having “good” sound doesn’t fit with the expectations and aesthetics of the musical genre, not fit with the expected environment where it will be heard.
I see people arguing here that digital is inferior because it can’t reproduce high frequencies. But tape has a limited bandwidth as well; more so than digital. With digital we just need to throw more compute power to sample higher and higher frequencies. With tape, there are hard physical limitations to it’s frequency capture. Your recording head can only get so many magnetic particles to move in any given location on the tape — it’s a hard physical process.
Same with vinyl — the cutting stylus was not capable of encoding 20kHz sounds, and the playback stylus is not capable of reading them. So the frequency limitations are actually higher in analog recording than in digital. Most estimates have analog playback topping out between 10 and 15kHz. The difference is that the cutoff is quite a bit smoother, so that it has a natural smoothness, rather than a hard wall. This is a significant factor, to be sure, but it has nothing to do with the frequency capture ability.
Objectively, digital is better in almost every way. It can reproduce frequencies more faithfully, and the sampling rate is far higher than anything the analog signal capture world can produce. Where digital is a bit weaker is in the hard signal roll off, but the high sample rates and advanced signal processing make this negligible for most cases.
Subjectively, you may not like it, and you may not like the music that makes heavy use of it, and that’s ok. But you can’t honestly say that this preference is somehow magnified to be an objective statement about a single component of the overall process. When there are blanket statements like “older recordings are always better because they’re analog” or “analog recording is more accurate” that is not really true. I think it’s more down to the artistic qualities of the recording, the choices made in the process, and what constitutes “good enough” for the target audience. If you think a recording sounds bad to you, it’s probably because you’re not the person the artist wants to reach. what it doesn’t mean, however, is that the technology itself is inferior.
When I was still in my twenties, ages upon ages ago, I used tape recording and it was excellent. These days I am incapable of recording anything. This is not progress.
The issue is that that many listeners perceive something in typical in-home digital playback that just doesn't sound right. So, they too freely speculate about what might be the cause. Because their speculations are un-tethered from digital theory, that doesn't mean that they aren't actually hearing some real artifact in typically (imperfectly) implemented digital playback.
Habit create preferences/taste. It is almost impossible to convincing someone who like analog process only and hate digital process to use digital process. He will be refuse to use his brain and only use his feeling. So he will response subjectively.
Distortion can be describe using filter application analogy in camera. Many people will like using filter application although they know the reality look like. But in audio it is more difficult, because many people did not know how it should sound correctly.
Distortion can be describe using filter application analogy in camera. Many people will like using filter application although they know the reality look like. But in audio it is more difficult, because many people did not know how it should sound correctly.
Digital sampling necessarily discards entire curve parts which fall between samples. What is done to make sure minima, maxima and inflexions are accounted for? These are difficult to account for if they happen in a very short curve segment which is not sampled.
Last edited:
That's only true for non-bandlimited signal. A proper anti-alias filter used before sampling limits the curvature which is possible. If done properly then there is only one possible curvature. That's the theory, and its mathematically true. Now in reality, there are no perfect anti-alias filters, no perfect reconstruction filters, etc. However, it can done to higher quality some people have ever heard. A really good dac can be essential for picky listeners. They are not cheap, probably in the range of $5k - $10k for two channels, unless maybe you want one in this class: https://www.mola-mola.nl/tambaqui.php ...Same thing for a really good vinyl playback setup, its not cheap to do well.
This points to a clear misunderstanding of digital sampling. Nothing is lost.
After reconstruction, you get the exact same waveform as the original, no matter what the shape is.
The only requirement is that the highest frequency in your signal is sampled at least twice in a period.
Check the Nyquist Theorem. Or, indulge me, and watch this video, your understanding will make a quantum jump for a few minutes of time investment.
The first 8 minutes suffice.
I know, most people in this forum are not interested in learning anything, but please prove me wrong.
Please.
Jan
Last edited: