Back

This isn't zero sum. It isn't "streaming or vinyl." The two things are complementary.

If a great indie band can survive the lean early years while they develop their sound, they might actually make it to your Spotify or YouTube Music station. You might have heard the streaming version of your favorite song, but do you have the vinyl record of the live performance of that song that you saw in a dive bar outside Nashville?

Vinyl sounds different. Not everything needs to be super high-fi.

@mekkaokereke I thought the point of vinyl was that, being analogue, with its infinite bitrate and zero sampling loss, it effectively /was/ super hi-fi?

@conniptions

You have "software engineer" and "musician" in your bio? That means I probably shouldn't try to 'splain to you how (I think) vinyl works. I've learned from previous (gentle!) spankings that this is generally not a good idea.

That's how I find out which of my friends worked on CoreAudio, who owns a modular synth, and who was in a rock band and played with bands like Life House before they suddenly said "C++ is cool, I wanna work on Spanner, tell me about the atomic clocks again."

@conniptions

Instead I'll politely wait for someone else with "software engineer" and "musician" in their bio to show up, and they can tell both of us how it really works!

My (probably wrong?) understanding: the noise of vinyl gives it an effective bitrate. As in, there's minimal audibly detectable loss relative to vinyl if digitizing above bitrate X.

@mekkaokereke I'll answer this with my 'music fan' hat on rather than the 'musician' or 'software engineer' hats.

I am lucky enough to have copies of several albums both on vinyl and digital.

Repeated A-B testing reveals that I prefer the vinyl playback every time without exception.

I don't really care if - as Adrian has pointed out - "digital audio is more accurate to its input". Maybe so.

In that case, perhaps accuracy-to-input isn't so important after all. Something else is going on.

@conniptions @mekkaokereke
I’ll take this one.

Vinyl does not have “effectively infinite bitrate.” All sound reproductions introduce error, including analog. The question is: What •kind• of error? How much?

In the case of digital, much of the error comes in the form of •quantization noise•: the difference between the “stair step” shape of digital and the actual signal. (Good image of it here: http://davidswiston.blogspot.com/2014/11/adcs-dithering-when-adding-noise-is.html) There’s also some error that comes from circuitry etc. 1/

@conniptions @mekkaokereke
Vinyl lives in a whole different universe of error: harmonic distortion, noise, surface defects, etc. No quantization noise!

But here’s the thing: the noise vinyl introduces is •orders of magnitude• larger than the noise CD-quality digital audio introduces. Like 20-30dB more.

There is basically no perceptible sound an LP can produce that a CD can’t reproduce (much less higher-quality digital audio). There reverse is not true.

Yet LPs •do• often sound better! Why?
2/

@conniptions @mekkaokereke
I owe this next part to my brilliant friend Greg Reierson (https://www.rareformmastering.com):

Mastering folks use compression and limiting and “sonic maximizing” and crap to make CDs sound louder. Unfortunately, it also makes the audio sound worse! Turn the volume on “maximized” audio down to match, and you’ll probably prefer the original.

Here’s the thing: those tricks mostly •don’t work• with analog audio. So people don’t use them.

LPs = worse audio of better masters.
3/

@inthehands @conniptions @mekkaokereke yes, exactly this. listening to digital masters that weren’t subject to the loudness wars (for instance, mfsl) is a very good experience, often sounding as good or better than vinyl. streaming services’ lufs-based loudness leveling renders the loudness wars moot, and people listen in noise canceling headphones, so these days, music tends to be mastered with much less of that horrible compression

@inthehands @conniptions @mekkaokereke and due to the fact that cd quality audio is outside the bounds of human hearing, we’ll never actually notice that quantization noise. 16-bit audio has a 96dB dynamic range (well enough to blow your eardrums out), and 44.1kHz accounts for nyquist-shannon at the maximum of human hearing plus a bit for the low pass rolloff at the top. audiophiles who claim otherwise are victims of pseudoscience and placebo marketing

@exchgr @conniptions @mekkaokereke When I was younger and had better hearing, my right ear could hear up to 24 kHz. I could easily distinguish 48kHz from 44.1 in a blind test, and could pick out 96 kHz and 24-bit samples in blind tests •sometimes• (low bass in quiet sections, usually). And there are some good theoretical arguments for why that should be. But basically yes, CD quality is at the edge of human perception; 96x24 is comfortably beyond it.

@inthehands @conniptions @mekkaokereke i can understand 48, but what’s the argument for 96?

@exchgr @inthehands @mekkaokereke I believe the idea is if you record at 96, mixing and processing there brings things into the perceptible range which remain present when downsampled later on, giving a fuller and richer sound than if you just started at 48/44.1. Not sure how true that is.

@exchgr @conniptions @mekkaokereke The argument is that process error accumulates: reduction at every individual processing step introduces more cumulative error than computing at higher precision and then reducing at the end — just like rounding floats to int and then summing ≠ summing floats and then rounding to int.

This is trivially true for bit depth, and almost certainly true for sample rate.

@inthehands @conniptions @mekkaokereke just so i understand what you’re saying, for instance: 24/96 would be useful for mixing and mastering, and then only at the final bounce step would it be wise to bounce to 16/44.1? i would tend to agree, since in the daw, one is constantly fiddling with levels, effects, panning, and even slowing/speeding, operations that would require higher precision than the end listener

@exchgr @conniptions @mekkaokereke Yes, exactly. DAWs these days even use 64-bit floats in their internal busses, which is probably overkill, but why not? But certainly having source recording at a quality that far exceeds human perception is a reasonable precaution.

@inthehands @conniptions @mekkaokereke yep, that makes a lot of sense. but i think distributing the end result at anything higher than approximately cd quality is just kind of a waste of bandwidth

@exchgr There’s a very solid case for 48 kHz, and also a solid case for 24-bit just for high-dynamic-range (e.g. classical) music. There •might• be a case for 96 kHz for listeners with young ears on high-end equipment. Beyond that, I agree.