- Joined
- Feb 23, 2016
- Messages
- 20,769
- Likes
- 37,634
One impulse is green and the other one is yellow.
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
High Resolution Audio: Does It Matter?
- Thread starter amirm
- Start date
I havent done the maths but from looking at the plots the ripples die out very quickly, cant get anywhere near 1 second.
Im going to send you a prize for that. Want to test the NC1200 amps in a few weeks?One impulse is green and the other one is yellow.
- Joined
- Feb 23, 2016
- Messages
- 20,769
- Likes
- 37,634
Sure!Im going to send you a prize for that. Want to test the NC1200 amps in a few weeks?
- Joined
- Feb 23, 2016
- Messages
- 20,769
- Likes
- 37,634
Well about that Dirac pulse depends upon how many taps in the filter. I suppose in a perfect world you'd have an infinite tap filter with infinite rippling at some level. With filters of a few hundred taps you get a few hundred samples of ripples.
Of course that is actually an illegal output signal in an ADC/DAC loop which is what Shannon-Nyquist refers to anyway. You don't see ringing in real music samples.
Of course that is actually an illegal output signal in an ADC/DAC loop which is what Shannon-Nyquist refers to anyway. You don't see ringing in real music samples.
Indeed which all makes a mockery of all the hand waving about filter ringing!
Last edited:
Something about filter ringing: https://troll-audio.com/articles/filter-ringing/
I think the cheeky way Amir posed the question implicitly acknowledges that the question can’t be answered yes or no without qualification.
On balance, I think the focus on minute differences in signal transmission technologies are at best a mild salve for the anxiety of trying to create a good music playback experience.
I agree that there are sound arguments for higher sample rate and bit depth recording on the production side, within reason.
In the playback side, whatever psychological benefit this brings comes at a cost.
It is in fact hard to have a good music listening experience if one has high expectations for sound quality and musical expression.
There are a myriad of reasons for this, but whether sample rate is 44.1 or 96khz, or whether this amp has .001 THD vs that other with a mere .01 are not among them!
On the question of hires audio as is being discussed here, it’s like being worried about whether the hasta in the front yard is getting enough water when the house is burning down.
We’re talking about music, which isn’t on the same level as climate change debates, or alternative medicine trends.
But to the extent it matters, a bigger contribution to the preservation of valuable cultural legacies would be the adoption of a trustable metadata system that would give the listener clear information about what version of a particular song they are hearing. I tried to do a quick ab of song on Tidal vs Spotify the other day, and the mixes were totally different. I liked the Spotify version a lot more, so the resolution is irrelevant.
High quality remasters of classic albums would be vastly more “impactful” than allegedly hires versions.
Having attention brought to the ongoing loss of actual musicians in the world would easily justify living with 320 kbit mp3 forever.
Even though I love the resource, my guess is a site like ASR is ironically reinforcing a subjectivist mindset. By focusing on measuring differences that aren’t audible, listeners are misdirected from the actual cause of whatever deficiencies they perceive in their music playback experience, ultimately getting confused by the ever shifting experience of listening.
I’m always bemused by the argument that digital recording is better than analog because of the vastly superior dynamic range. I used to record on analog tape and it was a superior recording medium in a lot of ways because of the dynamic range compression it has.
Decent audio gear and digital recording will actually capture and playback the electric signal generated by a mic faithfully. And in most cases this is not what you want! Recording on digital mediums requires the substitution of other forms of compression.
Dynamic range in music is paradoxical. If we had super high dynamic range recording and playback systems, we’d have to put limiters in the chain to make it a meaningful musical experience for all but a very niche type of music and recording.
Even then, you would need a very quiet room to enjoy it, and playback levels on par with the original signal.
The communication of the perceptual dynamic range of a live performance requires dynamic range compression on a playback system.
There is a lot involved in the activity of recording and listening to music, but the most important thing to me is expression.
The resolution of the current audio recording system as a whole from a technical point of view is good enough. There are more fruitful avenues of improvements to be pursued, at lower cost.
On balance, I think the focus on minute differences in signal transmission technologies are at best a mild salve for the anxiety of trying to create a good music playback experience.
I agree that there are sound arguments for higher sample rate and bit depth recording on the production side, within reason.
In the playback side, whatever psychological benefit this brings comes at a cost.
It is in fact hard to have a good music listening experience if one has high expectations for sound quality and musical expression.
There are a myriad of reasons for this, but whether sample rate is 44.1 or 96khz, or whether this amp has .001 THD vs that other with a mere .01 are not among them!
On the question of hires audio as is being discussed here, it’s like being worried about whether the hasta in the front yard is getting enough water when the house is burning down.
We’re talking about music, which isn’t on the same level as climate change debates, or alternative medicine trends.
But to the extent it matters, a bigger contribution to the preservation of valuable cultural legacies would be the adoption of a trustable metadata system that would give the listener clear information about what version of a particular song they are hearing. I tried to do a quick ab of song on Tidal vs Spotify the other day, and the mixes were totally different. I liked the Spotify version a lot more, so the resolution is irrelevant.
High quality remasters of classic albums would be vastly more “impactful” than allegedly hires versions.
Having attention brought to the ongoing loss of actual musicians in the world would easily justify living with 320 kbit mp3 forever.
Even though I love the resource, my guess is a site like ASR is ironically reinforcing a subjectivist mindset. By focusing on measuring differences that aren’t audible, listeners are misdirected from the actual cause of whatever deficiencies they perceive in their music playback experience, ultimately getting confused by the ever shifting experience of listening.
I’m always bemused by the argument that digital recording is better than analog because of the vastly superior dynamic range. I used to record on analog tape and it was a superior recording medium in a lot of ways because of the dynamic range compression it has.
Decent audio gear and digital recording will actually capture and playback the electric signal generated by a mic faithfully. And in most cases this is not what you want! Recording on digital mediums requires the substitution of other forms of compression.
Dynamic range in music is paradoxical. If we had super high dynamic range recording and playback systems, we’d have to put limiters in the chain to make it a meaningful musical experience for all but a very niche type of music and recording.
Even then, you would need a very quiet room to enjoy it, and playback levels on par with the original signal.
The communication of the perceptual dynamic range of a live performance requires dynamic range compression on a playback system.
There is a lot involved in the activity of recording and listening to music, but the most important thing to me is expression.
The resolution of the current audio recording system as a whole from a technical point of view is good enough. There are more fruitful avenues of improvements to be pursued, at lower cost.
A sinc(x) filter does not have in-band ripples.
MRC01
Major Contributor
True. According to W-S theory, sinc(x) perfectly reconstructs the bandwidth-limited analog signal that was encoded. The ripple is at Nyquist which is above the passband.
However, DACs in the real world don't use sinc(x). They use alternative implementations that fall short of theoretical perfection. And the AD side is just as critical. If the AD filter falls short of perfection, it allows some ripple or energy above Nyquist to leak through, that gets aliased into the passband, in which case even theoretically ideal perfect DA will reconstruct a slightly different wave.
PS: from the article linked above: Ringing from oversampling filters in DACs is eliminated entirely if the input signal has a little margin between its highest frequency component and the Nyquist limit of half the sample rate.
In my view, this is the primary justification for higher sampling rates: increasing that margin.
Last edited:
MRC01
Major Contributor
Exactly. When we listen to an acoustic music recording, more than 90% of what it sounds like comes from the room they were in, the position of the musicians and mics, and the number and type of mics. Much of that last 10% is what the engineers did to it after recording: EQ, compression, whatever. In that sense, debating how close to theoretically ideal performance real DACs can get may be as meaningful as debating how many angels fit on the head of a pin.
That said, it's still interesting to me for 2 reasons. First because how to engineer something ever closer to the theoretical limits is inherently interesting, and second because some people can hear the limitations under ideal conditions. Which gives at least a thread (however thin) of pragmatism to the first reason.
Compression, like any tool, can be used and abused. I have some recordings that use it well with subtlety to artfully enhance the recreation of the actual musical event. The 1960s RCA Victors come to mind, among others. I don't prefer this euphonic portrayal, but it at least can be enjoyable, not offensive. I have other recordings where it is abused and squashes the life out of the music. The latter seems to be the "standard" with modern rock, pop and, sadly, this disease is spreading into other genres like jazz, blues, folk and bluegrass. Fortunately, classical music has resisted this.
I understand the desire for a compressed version that sounds as loud as possible. Most people listen on cheap stereos, earbuds, cars and noisy environments. But why not provide a "studio master" or "faithful" version without all that artificial processing that squashes the life out of the music, and make it available on formats that audiophiles use, like CD, DVD, BluRay or as a download? The irony is, many of the "high-res" downloads from places like HDTracks are so compressed they could be transferred to 8-bit without any audible difference!
Last edited:
64Khz, please. Based on the assumption one has to use an analog filter. 50, maybe, if one can use a constant-delay digital filter.
768kHz is ludicrous.
A sinc filter is infinite in length. Not sure where you're headed there.
Now, nobody uses that kind of filter, even windowed. There are better ways to design an anti-aliasing/anti-imaging filter that have quite finite length. While there is an argument about whether to use equiripple (firpm) or mean squares (fircls), both are better than an equal-length windowed sinc.
The ripple, in either case, is at very high frequency, in a sinc fs/2, in an equiripple a combination of the cutoff frequency and fs/2.
Now, people who have optimized their filters for minimum length can create problems by designing filters that have large tap weights near the beginning and end of the filter. (this generally also translates to passband ripples) Somewhere I have some plots that show all this, but its sunny outside, I can't find the matlab script to make them, and I don't feel like working on a nice sunny day with blue skies. Sorry about that.
If I draw a waveform in Audacity, and export the .wav data to Excel, and smunge it into a format that the miniDSP will digest as an FIR filter via copy/paste, I wonder what kind of horribly mutated sound I would come up with...
We’re actually getting hit with “double whammy” with overly compressed/limited masters released by commercial labels. Not only are the dynamics squashed far beyond what would sound good musically, the extreme limiting or even deliberate clipping of the peaks is introducing nasty distortion.
Dynamic range compression is about changing the ratio of soft to loud sounds. To represent a signal that originally came from a “high perceptual dynamic range” into the average playback environments requires compression. If done well, it actually allows the dynamics to be perceived more faithfully.
I think the arguments made by Dan Lavry about an optimum sample rate being around 60khz make sense. Since that is unlikely to happen, an argument could be made that 88.2 is the best compromise.
But really the state of modern recording is so dreadful sounding...well it should be a scandal!
There is perhaps a ray if hope with the advent of level normalization by streaming media companies. Under most of the algorithms used, a mix that is massively squashed to increase average relative level backfires, and the track sounds weaker when the levels are scaled back by that same parameter.
But I would be thrilled with high quality masters being available on Spotify, at a mere 320kbit.
Somehere on here are some plots that prove the "clipping", etc. I know. I posted them.
Some modern production techniques are best described as "How to master for maximum fail".
That's the quest we have to make, the quest for good masters. Hunting down all the BS.
MRC01
Major Contributor
Except for modern recordings of classical music, which thankfully seem to have been unaffected by the loudness wars. They seem to use improving technology to incrementally improve. Every few years, another recording that I used to consider the best SOTA is edged out by a new one that is even better. Companies like Hyperion, Telarc, Alpha, Ondine are putting out some fantastic realistic sounding stuff. There are some duds even in classical music, so it's still a bit of a crap-shoot, but in this genre the overall odds are in your favor to get a better recording.
MRC01
Major Contributor
Sadly, true. I listen mostly to classical which doesn't have this problem. But recently I got Rival Sons' latest album Feral Roots based on a recommendation from a friend. The first few seconds revealed it was so distorted and crunched it was practically unlistenable. So I got curious and ran some tools. Turns out measures a paltry DR5, with massive clipping all over the place. If I were in that band I'd be angry at the recording engineers for screwing up my work and art!
The irony is, they have a 96-24 download for an album that has so little dynamic range it could be transferred to 8-bit! I didn't want to unfairly judge it by classical music standards, so I measured Pink Floyd Wish You Were Here for comparison, which measures DR14 with no clipping.
Similar threads
