Crosstalk: digital more like Vinyl?

[jkeny]

did they compare a direct-to-disc mastered LP, to the tape copy and then to the RBCD mastered from the same tape as well as an LP? Sheffield Lab has offered some titles this way, and a child could hear the superiority of D2D (ask Doug Sax what he prefers). imo, the limitations of the LP cutting/pressing/playback process are overstated and D2D mastered LP may well be the highest resolution format of any kind in existence today.

I don't know - the quote is from AVS forum but I would have to find the post.

I see Atmasphere has said exactly the same thing & I've seen it mentioned before. How does it square with the other reports that I hera which is the R2R tape is even better sounding than vinyl? Again, all my questions on here are not to challenge but simply to learn.

 

[Atmasphere]

^^ That has to do with the fact that the LP mastering operation is lower noise, wider bandwidth and lower distortion than analog tape.

But- if the master of the recording is a tape and you can get a good copy of it, its likely to sound better than an LP that was made without a lot of care.

 

[jkeny]

This is the link to J-J's post.

He goes on to explain the loudness/growth issues as:

Well, as you recall, then, loudness grows as a compressed function (approximatley 1/3.5 power law) inside a critical band (or better ERB, but I am old and keep calling ERB's critical bands even though I know the difference), but *adds* across critical bands.

So, if I take a signal, and double the energy, the loudness grows by a factor of 2^(1/3.5).

If I double the energy by adding it in critical bands that did not previously have substantial energy, though, I'll get DOUBLE the loudness.

Alternatively, if we keep the same energy, but spread the signal bandwidth, the loudness will go up. In the loudness tutorial .ppt I already referenced a while ago is a graph of roughly by how much that will happen. (I suspect you understand why I'm being approximate when dealing with the auditory system?)

LP's distort more (percentagewise, whateverwise) at higher levels. A polynomial model, in fact, seems to do a good representation of this until outright mistracking happens. The model is different for M and S (not L and R). This also turns out to be important, but not for the reason we're discussing here.

This means that at low levels, the distortion is not very noticible. At high levels, with the usual kind of lowpassy audio signal one sees most often,it means that the spectrum will spread approximately two octaves higher, with some substantial energy there. This means that even though the energy will grow a bit, the loudness grows quite a bit.

This creates the ILLUSION of more dynamic range.

 

[esldude]

This is the link to J-J's post.

He goes on to explain the loudness/growth issues as:

This fits with some results I have just messing with stereo mike setups. I had the idea to mix in low though audible levels of noise around small numbers of test signals. I delayed the noise between channels and altered interchannel levels of the noise so the noise imaged in the location of the test sounds. Each signal got its own little cushion of noise around it. Doing so with varying strength signals results in a more dynamic lively imaged result upon playback. I had in mind trying to improve the seeming realness of multi-miked sounds beyond simple pan-pot stereo. Though my preference is not to do multi-miking.

 

[esldude]

The phantom source problem described by Uli should vary from system to system. In my case, it's still unclear to me how the added crosstalk helps. When I do the test signals, the closest to the 1khz is with the settings at 0 for both HF and LF. Maybe this explains why the "flow" setting isn't that helpful with some music. I will play with it a little more. But I wouldn't say it's not a huge difference one way or the other in my system. I am sure this varies from system to system.

Maybe long term listening would be different.

Let me take a shot at this again though Uli already has.

Let us imagine you are mixing an acoustic guitar into a track. It was close miked in mono for a later mix with other instruments. You decide you want to position it halfway to the left speaker. Typically you put the sound in both channels and make the left channel say 4 db louder than the right. It will sound somewhere about where you want it. But this is somewhat unnatural.

The guitar would have a resonance somewhere just either side of 200 hz and some notes in that range. In that frequency range your ear mostly uses time delay between channel. A .4 millisecond time delay in that range would place the sound halfway to the left even if the notes are equally loud in both channels. Your ear does respond weakly in that range to level differences between channels. So for that part of things with your typical pan pot mixing instead of halfway left it sounds very slightly left barely off center.

Now overtones and harmonics will reach into the 1000 to 1500 hz range. Here your ear responds only a little to time delay, but there isn't any in the mix. It responds better to level differences and those sounds might appear to be 1/3 of the way left to you.

Now at 1600 hz and above level differences are about all you hear. And those will sound halfway to the left as intended.

So with one instrument intended to be halfway to the left speaker you have conflicting cues. Your brain will put it together and hear it somewhere around 1/3 to 1/2 half of the way to the left speaker. But it won't be a sharp natural image (presumably it won't sound natural like real guitar in this sense). The lack of integration will probably interfere with a sense of any natural space around it as well.

Now these numbers may be off, but illustrate the idea I hope. You can pan the left channel to be 10 db higher which will move the lower band to sound halfway to the left speaker. You might need 6 db for the mid band, and 4 db for everything above 1600 hz. If you panned 10 db and had a crossfeed filter that progressively crossfed more at higher frequencies you might reach a point where the imaging would be more coherent, integrated and sound better. The crossfeed might be nothing around the lower band, a bit to reduce interchannel level in the midband and a bit more to reduce level differences yet more at higher frequencies.

So for pan-potted multi-miked recordings, for crossed figure eights, or coincident X-Y stereo pairs this would likely be an improvement. For spaced pairs and spaced omnis the cross feeding will not be a benefit. So it will vary on its effectiveness depending on how something was recorded and mixed.

Much of this could be avoided if DAW's combined an interchannel delay to match interchannel level differences. But to my knowledge most do not work this way. One could add the delay manually, but it is almost never done. And there is some issue that how much delay matches how much level difference for good imaging is different for speakers vs headphones. Along with likely varying somewhat with particular system setup in the case of speakers. It sounds like the setup in Uli's software allows you to compensate for your speaker positioning to some extent.

 

[jkeny]

Another complexity is that we are not all calibrated in the same way with regard to ILD & ITD so I guess it has to be done at playback, rather than at mastering (I'm correcting what I said about this earlier).

Here's an ILD ITD practical experiment http://auditoryneuroscience.com/book/export/html/122

And another one from the same site http://auditoryneuroscience.com/topics/time-intensity-trading

 

[Atmasphere]

This is the link to J-J's post.

He goes on to explain the loudness/growth issues as:

-And you don't think he is really talking about superposition of the waveform? Actually in most cutterheads as you push the system, the distortion in the amplifiers actually goes down. This is because most cutter amplifiers have about 10x or more power than needed to drive the cutterhead. The cutterhead itself can easily cut grooves that no cartridge or arm combination can track, and it can do so without significant distortion. The limitations of LP are mostly in playback.

 

[jkeny]

-And you don't think he is really talking about superposition of the waveform?

Sorry, I don't know you mean by this.

Actually in most cutterheads as you push the system, the distortion in the amplifiers actually goes down. This is because most cutter amplifiers have about 10x or more power than needed to drive the cutterhead. The cutterhead itself can easily cut grooves that no cartridge or arm combination can track, and it can do so without significant distortion.

Yes, I believe that is correct

The limitations of LP are mostly in playback.

I believe this is also correct. J_J has a habit of being ambiguous in his posts - it seems to be his style, IMO

 

[Atmasphere]

Superposition:

https://en.wikipedia.org/wiki/Superposition_principle

 

[jkeny]

Superposition:

https://en.wikipedia.org/wiki/Superposition_principle

Yea, I know what it means but how does it relate to what J_J said - sorry, maybe I'm being thick?

 

[Atmasphere]

It was this:

So, if I take a signal, and double the energy, the loudness grows by a factor of 2^(1/3.5).

If I double the energy by adding it in critical bands that did not previously have substantial energy, though, I'll get DOUBLE the loudness.

 

[RogerD]

I am a digital dude. But I can't deny that sometimes vinyl systems can offer something digital struggles to deliver. We often hear folks pay a compliment to digital when they say it sounds "analog."

I love using DSP in my system. Recently, I switched over to Acourate.
http://www.audiovero.de/en/acourate.html

Since then I've been introduced to a setting in the Acourate Convolver called "flow." The basic idea is to set the parameters to permit some small frequency dependent crosstalk. The author's words explain it well:
http://digitalroomcorrection.hk/http___www.digitalroomcorrection.hk_/AcourateFlow.html

In summary, Uli's theory is that vinyl's perceived technical weakness (high cartridge crosstalk) can actually be very beneficial to better stereo reproduction.

I've tried it out and I really like it. It's not a huge difference. However, the music does sound a little more natural without any loss of resolution. It's cool!

Can DSP really be used to make digital playback sound more like vinyl?

Michael.

This sounds like a subtle way to sum information from R & L channels. There is some research done which is termed Binaral or 3D imaging. David Hafler back in the 1970's developed a circuit to increase ambient information reproduced. My experience is that Analog is a far better format for reproducing low level ques then digital. I use a circuit that sums information and is a time domain correction circuit. The main benefit of that technology is to enhance ambient information reproduction which does in fact enhance digital formats. The main qualities of such a circuit is that it only comes into play if "the" information is present and therefore not all recordings seem to be enhanced and it will expand the "sweet spot" or increase the size of the phantom center. Low level detail recognition is enhanced as well as the imaging or soundstage. I have expanded the circuit by adopting the Cinerama speaker araingement devolped by Hazard Reeves and Robert Fine. It does indeed make digital sound like Analog.

I do not believe that it is a distortion in the vinyl playback. Uli is just summing information digitaly that's all.

 

[esldude]

It was this:

While superposition will occur that isn't really j_j's point. The point is adding it in a band already occupied by signal doubling signal level results in a perceived increase that is smaller than double. Adding signal in other critical bands that were formerly empty result in a perceived increase in volume which does double for doubled energy. So added noise with increased signal level or distortion starting to get high enough to be perceived would cause something like LP to be perceived as more dynamics than the same energy in a low distortion quiet system. Because the energy increase as volume goes up spills into other critical bands with something like LP and would not if cleanly reproduced.

 

[Atmasphere]

OK.

However, the LP *is* a low distortion and quiet system- 'cleanly reproduced' is what its all about...

 

[jkeny]

While superposition will occur that isn't really j_j's point. The point is adding it in a band already occupied by signal doubling signal level results in a perceived increase that is smaller than double. Adding signal in other critical bands that were formerly empty result in a perceived increase in volume which does double for doubled energy. So added noise with increased signal level or distortion starting to get high enough to be perceived would cause something like LP to be perceived as more dynamics than the same energy in a low distortion quiet system. Because the energy increase as volume goes up spills into other critical bands with something like LP and would not if cleanly reproduced.

Exactly - it's why I couldn't understand what Atmas was getting at with his superpositioning?

 

[RogerD]

While superposition will occur that isn't really j_j's point. The point is adding it in a band already occupied by signal doubling signal level results in a perceived increase that is smaller than double. Adding signal in other critical bands that were formerly empty result in a perceived increase in volume which does double for doubled energy. So added noise with increased signal level or distortion starting to get high enough to be perceived would cause something like LP to be perceived as more dynamics than the same energy in a low distortion quiet system. Because the energy increase as volume goes up spills into other critical bands with something like LP and would not if cleanly reproduced.

Now that is a novelty.

 

[jkeny]

Now that is a novelty.

Well, it is based on J_J's psychoacoustic testing & his findings which I presume have not been overturned or challenged?
But it does not explain the low level superiority in detail retrieval that vinyl has over digital, which you mentioned - J_J's statement only applies to this occurring when volume is increased. I know in digital there are fewer bits involved in encoding low level detail but I don't believe that fully explains the perceptual difference, does it?

 

[RogerD]

Well, it is based on J_J's psychoacoustic testing & his findings which I presume have not been overturned or challenged?
But it does not explain the low level superiority in detail retrieval that vinyl has over digital, which you mentioned - J_J's statement only applies to this occurring when volume is increased. I know in digital there are fewer bits involved in the encoding low level detail but I don't believe that fully explains the perceptual difference, does it?

edited

I have rethought this and the only thing I can attribute the difference is bandwith capacity or it's the process itself. Both can achieve great levels on their reproduction, but the difference is in information delivery.

 

[esldude]

Can't speak for j_j, but it would be my opinion that low level details are more apparent due to some slight compression needed for making a pressed record playable. Small amounts of compression elevate moderately low level details and increase apparent dynamics without being noticed as such. Not the whole answer mind you just a part of it.

 

[jkeny]

Can't speak for j_j, but it would be my opinion that low level details are more apparent due to some slight compression needed for making a pressed record playable. Small amounts of compression elevate moderately low level details and increase apparent dynamics without being noticed as such. Not the whole answer mind you just a part of it.

Are you saying that a small amount of compression that slightly squashes dynamics has the effect of perceptually increasing the dynamics? That's an interesting idea - is there any background support for it?