Hacker News Reader

Comment by renlo | Hacker News Reader

My very limited understanding after talking to someone who was interested in this stuff, is that people are concerned with quantizing of the signal, which will happen with that laser solution. I’ve heard this is why audiophiles also use magnetic tape. To be clear I heard this second hand and am not directly familiar.

View on HN

A completely analog laser scanning solution is entirely practical. There's no reason to sample it digitally.

But honestly, this seems like insane overkill. Vinyl has a ridiculously limited dynamic range, and digital recording at 24/48 is practically transparent even with low-end prosumer equipment. At the high end you have to work quite hard to spend more than $4k on a spectacularly clean and neutral studio-grade ADC/DAC. (You can spend more on madly expensive super-fi, but it's not going to sound any cleaner than a good Burl, Lynx, or Metric box.)

So if you spend $4000 on the turntable/arm/pickup another $4000 on studio grade converters, you can capture recordings that will be absolutely indistinguishable from the vinyl source. And your precious plastic disks can stay in their wrappers.

As far as DACs are concerned, you’d be surprised: A few hundred dollars (e.g. Topping D50) can get you drastically better performance than many multi-thousand dollar “audiophile DACs”:

https://www.audiosciencereview.com/forum/index.php?threads/m...

Now as far as speakers are concerned, yes, you’ll need to spend at least a few thousand for a stereo pair capable of doing justice to this level of precision and dynamic range.

Ah yes, ASR. The place where a DAC or amp having a -100db SINAD means it's a terrible piece of hardware despite it being well below the threshold of human listening and the noise floor of the best recording and tube amps, who have significantly higher distortion, are considered terrible because of it despite the vast majority of people who actually listen to them liking them far more than some of the most beloved hardware at ASR (I'm looking at you THX and Topping hardware).

I belived in their swansong and bought into it up until recently, then I tried "terrible" hardware and dear God it sounds so much better.

For a good introduction to signal quantization and some of the myths surrounding that, A Digital Media Primer for Geeks, and especially the follow-up, Digital Show & Tell, are well worth watching. https://xiph.org/video/

Frankly there’s no saving the people concerned about quantization issues with digital vs vinyl. They clearly don’t understand the basic physics and that mechanical of PVC severely limits dynamic range such that with about 10 bits of signal and some noise you can quite comfortably transparently recreate vinyl. 16 bits is more than enough to record vinyl.

There’s no reasoning somebody out of something they weren’t reasoned into the first place, so usually this isn’t an argument worth having.

A modern laser interferometer could easily get well past the resolution that is limited by the vinyl structure itself.

It should be possible to do a laser system that is analog. As the distance to the groove varies continuously, that would cause a corresponding continuous variation in the intensity of the recombined beam in an interferometer.

I don't think the correspondence would be linear. There's probably something like cos(2 pi d / W) in there somewhere, where d is the distance to the grove and W is the wavelength of the laser. So you'd need some sort of equalization circuit on the output, but audiophiles are OK with that--they are used to RIAA equilization on vinyl already.

I assume that the variations in groove distance are much larger than the wavelength of the laser, which would mean that there are several plausible grove distances that give the same intensity output of the interferometer. I'm not quite sure how to handle that. I have a feeling an integrator would be involved, but details elude me at the moment.

Magnetic tape is still used in recording, but for the distortion it adds rather than concern about quantization. In fact, the audio is simply routed through the tape machine, then into a computer for storage.

Distortion here means both extreme distortion (e.g. clipping) which is what people often think of when they hear the word "distortion", or the more technical definition of any alteration of the input signal. The pre-amps, compressors, and so on will all add their own distortion to the signal. However this is desired distortion. Tape offers some of this, although I'm not familiar with what its exact characteristics are.

Clipping (when the input signal amplitude exceeds the limits of the storage medium or output device) is a type of distortion that can sound great or absolutely awful.

Digital clipping sound awful. The max amplitude is all 1's in the digital storage medium. When a signal exceeds this, for the duration of it exceeding every sample will have that same value of all 1's. The waveform will basically have the top lopped off. There will be sharp "corners" where it starts to clip, and where it stops clipping. Those corners create "pops". It makes that part of the recording unusable.

Analog clipping can be a very sought after sound though. The max amplitude on tape is when all of the poles of the magnetic material are aligned. This isn't perfect though and it's not instant, so you get a slower compression of the signal and some variation. The waveform will look like its top has been squeezed downward. This can sound amazing.

The quality on youtube is pretty bad (spotify or elsewhere is better), but this song has a great example. You can hear his voice distort in many places, but it sounds wonderful (particularly starting around 2:20 - 2:40 and 3:10). https://www.youtube.com/watch?v=wAGS8UL9b6A

If you play the sound through the air, which is an analog medium connected to your ears, haven't you un-quantized that nasty digital information before you can hear it?

These types of "audiophiles" don't believe in the Shannon-Nyquist-theorem, there's no reasoning with them.

Reply on news.ycombinator.com