USB vs. TOSLINK optical?

[KeithL]

First of all, I agree with you in "the absolute sense", that the best way is to start out with a good image...
But, even with photography, there are starting to be rather wide grey areas... and they keep getting wider.
Photo editors are now smart enough to automatically mask things like the subject, or the sky, and usually get it right...
And to "intelligently add detail" (partly by sharpening and partly buy guessing what "should" be there.)
And the "content aware fill" in the latest version of Photoshop is downright scary...

And then there are the specialized apps...
For example, if I want a picture of a building, and there are always people walking back and forth on the sidewalk, I can still get my prefect picture.
I set up a tripod, and snap a dozen pictures from the exact same spot.
I can now have software analyze those dozen frames...
Recognize that the parts that never change are the building...
While the parts that change from image to image are people...
And automagically splice together a perfect image of the entire building without any of the people...
And, if I'm at least a little bit careful, it will be dead-on perfect.

Similarly, if I had a digital microscope, which has a ridiculously shallow depth of field...
I can take a large series of images, moving the carriage vertically a tiny amount between each one...
So now I have a whole bunch of photos where only a small area in each one is in sharp focus...
Then I have "focus stacking software" detect the area on each image that is in sharpest focus and "stack them" into one sharp composite image.
(One click and the software does all the work... even making minor adjustments if the registration is a bit off.)

And modern "AI software" is getting downright ridiculous...
With some of it you can literally feed in a blurry photo of a woman wearing glasses... and get back what appears to be that same image... only now it's sharp.
When you examine it carefully, you may find that certain parts have been "Frankensteined" from the software's massive collection of body parts and accessories, but it usually looks pretty good.
However, some of the software is a little less extreme, but can still "remaster" things like cement and grass textures quite convincingly to cover missing spots or flaws or to add detail.

HOWEVER, as for your question about "fake stereo"...

In the old days this was pretty well limited to applying some sort of "comb filter" that literally "sifted" certain frequency ranges between left and right.
Or, in certain situations, if someone was willing to exert the manual effort, actually use EQ to isolate specific instruments or voices by selecting their frequency range.
So, if you had a singer, a bass guitar, and a cymbal.... put the midrange in the center, pan the high notes to the right, and pan the low notes to the left. 
And, in certain situations, this could be quite convincing.

Lately, as you might expect, "things have changed a bit".

We now have software that can isolate specific voices or instruments or sounds extremely accurately... and put each into a new separate track.
So, for example, it will pick out the bass guitar, and set filters to isolate both it, and all of the harmonics of the note it is playing...
Or it recognizes the cymbal as characterized by "a burst of random high frequency noise" and isolates the range of associated frequencies (but only while it detects all of them being present).
The result is NOT to have perfectly clean isolated tracks for each instrument...
But even if, for example, the voice is 12 dB down in the guitar track, and the guitar is 12 dB down in the voice track, there is enough separation that you can pan the voice and the guitar to different locations.
It doesn't always work, or work perfectly, but it can be incredibly effective most of the time.
One of the popular programs for that sort of thing is this one:  www.steinberg.net/spectralayers/
(When they talk about "unmixing tracks into stems" what they're talking about is basically separating the individual instruments in a track into separate tracks - which is what you do to make mono into stereo .)

This same technology can be used to remove both hum and buzz...

With buzz, I click on the 60 Hz fundamental, and the software automatically selects the fundamental and all of the harmonics that go with it...

I can then hit the Delete button to delete the buzz... or just turn it down to where it isn't noticeable... but not leave a "spectral gap" in the remaining content.

Now, for restoring, repairing, and things like removing noise and clicks... nothing beats Izotope RX.
It can fix clicks, pops, distortion, noise, hum, clipping, and all sorts of other problems you might have with an audio sample.
(You've seen apps where you can move a slider to add reverb... Izotope RX has a slider that can REMOVE reverb from a track.) 

Dec 9, 2022 14:44:52 GMT -5 leonski said:

Dec 9, 2022 10:43:07 GMT -5 KeithL said:

I'm a major fan of remastering things to sound better...
And, considering the availability of really good software, which can run on an ordinary computer, and doesn't really cost all that much, it's something anyone can learn to do these days.
For example, for restoring and repairing things like old recordings of records, check out Izotope RX...
  (The "lite" version is only about $99 on sale... and, compared to the cost of expensive audio gear, even $800 for the "full on studio version" is not too bad.)  

NLE, which is "just how digital editing works", is one of the single biggest advances of all time in audio and video editing.
NLE... non-linear editing... refers to the fact that, unlike old analog "mixing consoles", modern audio and video editing software doesn't have to work in real-time. 
This means that, within reason, while your edit may take a lot longer to process on a slower computer, when it's finished you will end up with exactly the same high quality results. 
With modern editing software editing or repairing an audio track has much the same "feel" to it as editing a picture in Photoshop... 
For example, if there's an annoying chair squeak in that concert recording, or a tick on that recording of your favorite album...
  You can literally punch up the spectral view, draw a circle around the noise, and hit the Delete key to get rid of it, or even just carefully turn it down a bit to make it less annoying.  

I also absolutely agree with your "priority list of things to improve".... especially with putting speakers and rooms at the very top.
(And especially that far too many people still seem to believe that "room correction software can fix anything and, if it can't, you need to find better room correction software".

However... I have mixed feelings about using specially remastered copies of tracks to demo audio equipment at shows.

Let me explain my sentiments there with an example...
Imagine that you were at a high-end store and saw a projector or TV that looked incredibly good with several movies they showed you on it.
But, when you got it home, you found that you just couldn't get it to look that way, no matter how hard you tried...
And, when you played those exact same movies at home, they didn't look nearly as good as they did at the store...
And you later found out that what you'd seen in the store were "special copies" of those movies "adjusted" to look good on that particular set.
Personally I feel that, if you're going to use "special content" to demo your products, you should at least mention it, or be careful not to mislead people.
I have no problem with showing off gear with cool demos it - as long as they don't leave the WalMart jewel case for that movie lying on the table and let me assume that's the version I'm watching... 

Coming from the viewpoint of a digital photographer?

Remastering is only as good as the data you have.     (I'll just call it data, for now).     If you start with a 16bit / 44.1khz?  You really can't get much better......
If you start with some huge file and bit depth?   You can easily throw away data and STILL improve.

That's why I shoot RAW images.   14bit color and very large (in comparson to sensor size) files......My 24meg sensor produces 35meg files......and no less than 30meg.....depending on image.

Some original movie soundtracks were on an optical strip and can be de-noised.   And still can sound excellent......

I'm curious about how you turn mono into good stereo?   

[KeithL]

Our new DAC is going to be just that - a DAC.
We have no plans to include any of the other functionality of a Roon endpoint into it.
Therefore, since Roon works with DACs, it will work with ours - so I'm not sure I see much value in having them "test" or "certify" it.
(Since there isn't any Roon-specific functionality involved you're basically talking about a "partnership agreement" and "a trading of logos".)
I wouldn't rule out the possibility that we might establish a relationship with them somewhere along the line.
(But that's probably a question to bring up after the product is actually released  .)

Dec 9, 2022 13:00:54 GMT -5 audiobill said:

Dec 8, 2022 16:21:32 GMT -5 KeithL said:

Just for the record there was a DC-2.
It was a prototype, mounted on a wooden base, and we had it on display at the last EmoFest several years ago.
It turned out to have various issues and never became a product.

Our new DAC will be quite different. 

Keith, I suppose this effort will be certified by Roon, having been tested by them?

[Boomzilla]

Dec 9, 2022 14:44:52 GMT -5 leonski said:

Remastering is only as good as the data you have.     (I'll just call it data, for now).     If you start with a 16bit / 44.1khz?  You really can't get much better......
If you start with some huge file and bit depth?   You can easily throw away data and STILL improve...I'm curious about how you turn mono into good stereo?   

In theory, you may be right. In practice, however...

Although you can interpolate resolution that wasn't there in the original data, you're still just guessing. But with frequency response, you can improve quite a bit. Was the microphone farther from the drum kit than the guitar speaker? You can modify phase. Is the second singer too low in the mix? You can boost them up. Etc.

The extrapolation of stereo from mono is somewhat similar. You can move some instruments right, some left. Change the phase and mix some out-of-phase right into the left channel, etc. while maintaining a solid center. It's definitely not true stereo, but you'd be surprised what can be accomplished.

[leonski]

Keith makes my point 'small' by referring to focus stacking. If you took ONE image of a still subject at some ridiculous small aperture to ensure great depth of field, you'd get an OK image of whatever a SINGLE image in your camera provides.
Focal length of lens, distance to subject and 'range' from near to far of the required 'good' focus all play into what you need to do and potential result.
I can set my camera up to be essentially a No Focus needed camera. Short focal length. Higher ISO and maybe f11 or smaller? Leave it focused at 10 or 12 feet and essentially everything from 3 feet to infinity will be fine.....

But if you stack? You might get 3 to 5 images the same size which total a LOT more data.....I don't know the upper limit in this kind of manipulation.
Experiment sometime with HDR......Similar idea of multiple images yielding a single result....

You can even 'Stitch' multiple images together to make very large images with very large files. My Prof....teaching photo business class showed an image of his which was composed of maybe 80 or 100 seperate and stitched
images. Excellent work.....and very time consuming. You should try to get all the images at one sitting since light changes and background MAY also.....

[Boomzilla]

Your professor is now obsolete, leonski. Computers & software can layer hundreds (maybe thousands) of big images in the blink of an eye. Really!

[marcl]

Dec 9, 2022 19:39:19 GMT -5 leonski said:

Keith makes my point 'small' by referring to focus stacking. If you took ONE image of a still subject at some ridiculous small aperture to ensure great depth of field, you'd get an OK image of whatever a SINGLE image in your camera provides.
Focal length of lens, distance to subject and 'range' from near to far of the required 'good' focus all play into what you need to do and potential result.
I can set my camera up to be essentially a No Focus needed camera. Short focal length. Higher ISO and maybe f11 or smaller? Leave it focused at 10 or 12 feet and essentially everything from 3 feet to infinity will be fine.....

But if you stack? You might get 3 to 5 images the same size which total a LOT more data.....I don't know the upper limit in this kind of manipulation.
Experiment sometime with HDR......Similar idea of multiple images yielding a single result....

You can even 'Stitch' multiple images together to make very large images with very large files. My Prof....teaching photo business class showed an image of his which was composed of maybe 80 or 100 seperate and stitched
images. Excellent work.....and very time consuming. You should try to get all the images at one sitting since light changes and background MAY also.....

This was 75 images, HDR ... and it was 10 years ago with a 22Mp camera.  Today I'm shooting 50Mp ... and that presents a new issue.  With very high Mp cameras you need to use a smaller circle of confusion.  The old hyperfocal calculations and DoF calculations are not good enough.  So I'm doing more focus stacking now, and especially not shooting smaller aperture than f/11 to avoid diffraction.  It's a whole new world 


At another extreme, it's possible to get 20fps 50Mp hand-held flight shots of a hummingbird at ISO 12800 and use DxO noise reduction to get very detailed images.

[leonski]

Nice Pano, Marci.

Now for the FINAL INSULT! Go to a frame shop and get an estimate.

I had about a 15x60 Pano and brought my OWN UV proof art-glass (some acrylic) along with having already had the print mounted on hardboard.

The bill was over 400$ and I felt abused. And didnt' save a PENNY by doing some legwork myself.....


As for Circle Of Confusion? Most don't know what that means. But it DOES boil down to needing REAL good glass......The standard drawing of focus generally does NOT detail RGB.....but just makes the

dangerous and incorrect call tha all colors focus together. I think that's Apochromatic? I'm digging WAY back. Looking at some images, you'll see color fringing. Maybe a purpleISH shade?

But the point is that Better glass TENDS to focus all colors together at the same point. ZOOM lenses are their own particular bit of hell for that reason AND focus breathing. My Nikkor 70-200 'S' lens? Has so much

alphabet soup glass, I can't even keep track. And the claims? NO focus breathing. And points made about all sorts of Chromatic / pincushion / barrel and anything else you've ever hear of.

One thing it is NOT? SMALLER....

That is a VERY WORTHY shot of a Hummer. Nasty little creatures. DxO is a favorite of mine and YES.....the noise reduction is 2nd to NONE......My 25mp Nikon has excellent dynamic range and at even a reasonable ISO3200,

I can see stars while photographing christmas lights.....