Is the vinyl LP an open music format? (Opensource.com) [LWN.net]

“a well-recorded vinyl LP”

Posted Feb 12, 2016 21:26 UTC (Fri) by ldo (guest, #40946) [Link] (84 responses)

He forgot to mention “well-pressed” as well. The true audiophiles would carefully check all the stampings in the store, looking for the best-sounding one.

Which would only sound that good on the first playing. Every other playing after that would degrade the quality just that little bit...

I feel no regret to see analog formats like vinyl and tape consigned to museums. That’s where they belong.

“a well-recorded vinyl LP”

Posted Feb 12, 2016 21:50 UTC (Fri) by Wol (subscriber, #4433) [Link] (55 responses)

So you're quite happy with TV that suddenly degrades to large pixellated blocks? And where you suddenly have no sound and picture for a moment?

So you're quite happy with music that has gaps in it? Or clicks as artifacts from digitisation?

Oh - and can you hear a digital loudspeaker - indeed, is there even such a thing as a digital loudspeaker?

Analog formats have a LOT of advantages. Not least that the original format is analog so we don't get the damage and entropy of a format conversion.

Okay, so we can COPY digital without adding entropy, but digital is not a magical panacea ... (and too many people DO think digital is "magic", so those of us who actually understand its limitations actually find it hard to enjoy its advantages).

I gather that quite a lot of modern CDs, nowadays, aren't made from WAV or FLAC master recordings, but MP3!!! How stupid is that? To create a "lossless" recording from a lossy master!!!

Cheers,
Wol

“a well-recorded vinyl LP”

Posted Feb 12, 2016 22:28 UTC (Fri) by pizza (subscriber, #46) [Link] (4 responses)

> I gather that quite a lot of modern CDs, nowadays, aren't made from WAV or FLAC master recordings, but MP3!!! How stupid is that? To create a "lossless" recording from a lossy master!!!

Got a citation for that?

(And I'm referring to actual pressed CDs, not the "modern" equivalent of a mixtape that someone burned for a friend..)

“a well-recorded vinyl LP”

Posted Feb 12, 2016 23:17 UTC (Fri) by bronson (subscriber, #4806) [Link] (3 responses)

A friend and I found one in 2003 or 2004, an actual production pressed CD. It was some hip hop collection released a few years before, don't remember which one.

We were LAMEing his CDs with VBR on reasonably high quality, ending up with 160-220 kbit average mp3 files. This album, with the exact same encode settings, ended up at a rock-solid 128kbit start to finish. Pretty funny in retrospect.

I bet I could find it again if I really dug, should be on a backup drive somewhere. Probably not worth the time though.

“a well-recorded vinyl LP”

Posted Feb 12, 2016 23:25 UTC (Fri) by bronson (subscriber, #4806) [Link] (1 responses)

That said, I believe Wol's statement ("quite a lot of modern CDs, nowadays, aren't made from WAV or FLAC but MP3") is demonstrably wrong. Out of the 2500+ CDs we ripped between 1999 and 2005, we only found one.

So, I agree: citation please!

“a well-recorded vinyl LP”

Posted Feb 14, 2016 15:11 UTC (Sun) by Wol (subscriber, #4433) [Link]

Sorry, no citation. Like so many things, it's something I've picked up that I can't verify (I don't have the audiophile skills to check).

But if your data is at least 11 years old, how do you know things haven't changed?

And I don't know about you, but we get a lot of freebie CDs over here in newspapers. It wouldn't surprise me in the slightest if a newspaper gets a blanket licence, a sub downloads a bunch of mp3's to burn a CD, and there you are ... a lot of mp3-grade pressed CDs in circulation.

I have come across a fair few such CDs, however, that CDDB recognises as commercial CDs so those are probably full quality.

At the end of the day, the claim sounds plausible. Especially at the cheap/free end of the market, it wouldn't surprise me if there are a lot of people involved who wouldn't understand the difference between FLAC/WAV and mp3, and they're exactly the sort of people who are likely to be producing these bottom-end CDs :-(

Cheers,
Wol

“a well-recorded vinyl LP”

Posted Feb 13, 2016 0:33 UTC (Sat) by tialaramex (subscriber, #21167) [Link]

Because of how the MPEG audio layer III compression scheme works, I'm not sure your story makes any sense.

It's trivially true that a 128kbit MP3 file is one valid compression output of a PCM signal that sounds exactly like that 128kbit MP3 file but the format isn't designed in a way that would be expected to actually give that result.

MPEG defined everything in terms of the decoder. Given this series of bits, here is what your decoder should output. The output is defined in terms of floating point audio, and decoder compliance is measured against the ideal output. Encoders aren't defined at all, leaving implementers to come up with any strategy they liked to try to make their encodings sound more like the original or use less bits to do so. Obviously your CD doesn't use floating point, the CD is instead 16-bit integer PCM. So that means there will be significant (to a machine doing a comparison, not to a human listener) quantisation noise which arguably wasn't present in the "original" MP3. Most PC software decoders output 16-bit PCM by default, but doing so bakes in the quantisation.

When an encoder like LAME handles the PCM data it too has to quantise, but now it's quantising a transformed signal, because it wants to perform a compression that doesn't work on linear PCM. So further noise is introduced.

Thus, every time you go around the loop from PCM to MP3 and then MP3 to PCM, you introduce noise and other distortions. It's not impossible for it to all cancel out, but it isn't very likely, and it needs to happen for every chunk the MP3 encoder handles, if even one goes "wrong" the knock-on effects accumulate. Probably at some point this settles down to produce some strange attractor of distorted noise, but it's not at all obvious to me that it should, let alone did ten years ago, happen on the first time around the loop or that it would be a recognisable hip hop album.

Of course I defer to you on the story, I wasn't there, but well, I'd find this a whole lot more believable if someone had the PCM data and the appropriate version of LAME to let me see this really happen for myself.

Re: TV that suddenly degrades

Posted Feb 12, 2016 22:45 UTC (Fri) by ldo (guest, #40946) [Link] (4 responses)

Strawman, much?

Temporary digital glitches can be fixed, inexorable analog degradation cannot.

Re: TV that suddenly degrades

Posted Feb 15, 2016 21:20 UTC (Mon) by Wol (subscriber, #4433) [Link] (3 responses)

Temporary digital glitches don't get fixed - if you're listening to the radio :-(

I guess the back end of pretty much every radio station is digital nowadays, so why is the music quality so much worse nowadays? That was behind my complaint about dropouts, gaps in recordings (in the middle of a track - not between tracks!), etc etc.

I get everybody's saying that digital is so much better at the producer end - it's just that, at the consumer end, from my point of view it's a lot worse ... :-(

Probably cost cutting etc :-( in the old days you had a small team of engineers, now you have one person spending 5 minutes to press a switch and set of an hour or two's programming - and when it hiccups no-one is there to notice :-(

Cheers,
Wol

Re: TV that suddenly degrades

Posted Feb 16, 2016 18:04 UTC (Tue) by farnz (subscriber, #17727) [Link]

This was all going on back in the analogue days, too, on the lower budget radio stations. All of the thing you're complaining about are, IME, consequences of cutting budgets, not of digital versus analogue.

At the very least, I encountered reel to reel tape-based stations with 20 second gaps in recordings mid-track - the tape was damaged, and had been spliced in the middle, which happened to be mid-track when someone careless recorded onto the tape ready for unattended playout.

Of course, digital makes unattended playout cheaper and more attractive - a £2,000 machine does 24 hours of playout without trouble, rather than needing lots of tapes.

Re: TV that suddenly degrades

Posted Feb 17, 2016 16:59 UTC (Wed) by drag (guest, #31333) [Link]

> I guess the back end of pretty much every radio station is digital nowadays, so why is the music quality so much worse nowadays?

Pure laziness combined with cost cutting.

The end result of how the FCC serves the American public by preserving the wireless spectrum is that a small handful of companies have control over the entire radio industry. When you own the #1, #2, #4, #5, #8, and #9 most popular radio stations in a region you really have very little desire to compete with anybody or improve. You will just end up spending money for no purpose.

My guess is that they standardized on a low-bandwidth compression for their equipment. Most radio stations, especially music ones, are no longer local. Even for stations that have local content, like traffic reports, are still going to be mostly pre-recorded canned stuff except for the 'live' stuff (which often is recorded anyways). So I expect they just stream most of their stuff over the internet or beam it down from satellite and need to make sure they use something that can be used in relatively remote areas with lousy network access.

Basically, if you want to have a good radio experience now you have to go internet.

It's trivial for any digital signal to out-do the quality of analog recordings for practical purposes. Sure, if you are looking at old analog recordings and have perfect setup then the audio quality will be the best you can get, but that is extremely unlikely to be the case. So it seems obvious to me if the digital quality is lower then analog for consumer products or broadcasting then the problem is a human one or political issue, not technical.

This is a bit of aside but:

When you look at all the shit they have to do to LP's to make them work properly with needle-and-groove technology you realize that they are FAR from a lossless recording. Stereo support is a hack. They have to make low-frequencies mono-only because it'll use up too much space in the groove. They have to squash loudness/extreme excursions to prevent needle skipping. All sorts of stuff like that is needed to produce something that can be widely consumed by customers.

Re: TV that suddenly degrades

Posted Feb 18, 2016 2:22 UTC (Thu) by excors (subscriber, #95769) [Link]

> I guess the back end of pretty much every radio station is digital nowadays

I remember listening to BBC 6 Music a few years ago, which is only broadcast on digital radio, and they occasionally played vinyl records live (mostly from modern indie bands being retro), so at least they still had some analogue recorded inputs into their digital systems. Sometimes they accidentally played them at the wrong speed, then realised the music sounded better that way and left it running. One time a band had sent them a chocolate LP, so they played it live on air then ate it. Not great audio quality but more fun than a CD.

“a well-recorded vinyl LP”

Posted Feb 13, 2016 0:07 UTC (Sat) by tialaramex (subscriber, #21167) [Link] (1 responses)

For audio at least, I'd argue digital actually should be seen as a panacea. Digital audio makes "good enough" astoundingly cheap, so that you don't even need to worry about it any more. It would be theoretically possible to make a CD player for example that didn't sound very good because the digital stage is crappy, but it would save so fantastically little money that it's not even worth the cheapest suppliers bothering.

Digital does that by raising the goal posts so much that even a "near miss" hugely out-performs what's needed for transparency to a human listener. 16-bit 44.1kHz PCM is ridiculously better than anything people were doing with storing analogue waveforms on vinyl or tapes, and yet it's seen as a low bar, you can't even buy PCs that "only" offer 16-bit 44.1kHZ any more.

You talk about the "original format" but what "original format"? If you weren't actually in the concert hall, listening to the orchestra perform Beethoven's Ninth, you can only ever hear a recording of the performance. So the recording must use a recording medium. Before any mixing, editing, anything like that it has to be recorded. If you choose tape, big open reel high quality magnetic tape with analogue signals, sorry, the noise floor of the tape is worse than 14-bit digital audio. Not different, not "arguably less pleasant" just straight up worse. Today you would (of course) use a 24-bit digital recorder. Even if you get the levels completely wrong, say you're off by 10dB because you never did any sort of sound check, the 24-bit recorder will produce a massively better result than the best achievable with the analogue tape after painstaking attention to detail with the levels bang on.

Like I said, an actual panacea.

Over in video things are a bit more complicated, mostly because of the truly immense quantities of data involved. A digital cinema for example doesn't actually just have a huge pile of pixmaps to project in order, the images are compressed although they don't use inter-frame compression like consumer video formats do. It's still possible to make a reasonable argument that the result is better for the audience, but it's not as much a no brainer as for audio.

“a well-recorded vinyl LP”

Posted Feb 14, 2016 9:22 UTC (Sun) by farnz (subscriber, #17727) [Link]

Note that with a noise shaping ADC, you can make direct SNR comparisons to analogue formats; high end analogue formats could get the equivalent of 13 bits per sample.

“a well-recorded vinyl LP”

Posted Feb 13, 2016 0:22 UTC (Sat) by tao (subscriber, #17563) [Link] (1 responses)

Not all original format is analogue though. Quite a lot of the music I listen to is pretty much digital from start to finish; instrumental synthesizer music (with an analogue vocal track that's been heavily distorted by a vocoder) recorded in digital format, mixed digitally and mastered to a digital format.

Though obviously this is mostly for newer music. My #1 favourites, Pink Floyd, are TTBOMK all analogue :)

“a well-recorded vinyl LP”

Posted Feb 16, 2016 21:05 UTC (Tue) by jwarnica (subscriber, #27492) [Link]

If you are recording audio, by definition, the message is a waveform. Possibly a waveform which (happens) be be easily definable by a digital signal.

It gets back to the whole distinction of sound *pro*duction and sound *repro*duction. Your favorite guitar player may have a tube amp, but he doesn't play a "guitar", he plays a "guitar", cable, amplifier, speaker. The tube amp in that produces a particular sound, not a good or bad one. And at home, reproducing it, a (today) even bad DAC will more faithfully reproduce that recorded sound then even a really good tube amp.

The musician (and engineers) producing the sound chose a particular distortion.

“a well-recorded vinyl LP”

Posted Feb 13, 2016 5:01 UTC (Sat) by rqosa (subscriber, #24136) [Link] (8 responses)

> So you're quite happy with TV that suddenly degrades to large pixellated blocks? And where you suddenly have no sound and picture for a moment?

I've seldom seen that happen except with terrestrial-broadcast TV; and as far as I can remember about analog broadcast TV, the picture was always so snowy as to make it unwatchable. (It wasn't like that everywhere, but if every transmitter tower is on the opposite side of one hill or another from where you are…)

As for cable / satellite / tape / disc / Internet streaming video formats: the interlacing/de-interlacing artifacts and composite-video artifacts (rainbowing and mosquito-noise) that you get with analog video look far worse to me than lossy-compression artifacts do.

“a well-recorded vinyl LP”

Posted Feb 13, 2016 6:10 UTC (Sat) by Sesse (subscriber, #53779) [Link] (1 responses)

Compression artifacts are a real phenomenon (and some people prefer snow and blurriness over blockiness; it's subjective). However, it's an apples and oranges comparison: If an analog channel were to be broadcast over as much bandwidth as a digital one uses today, you simply wouldn't get a thing through. Or conversely, if you were to send a single digital signal using the bandwidth that used to be allocated to every analog channel, you'd end up with about 80 Mbit/sec bitrate (6 MHz, 40 dB SNR). I do believe even the poster would be happy with the (lack of) compression artifacts from an 80 Mbit/sec H.264 SD signal :-)

So you can complain that TV today gives you ten times as many channels, and that you occasionally get artifacts as a tradeoff for that. But those artifacts have nothing to do with digital vs. analog, beyond the fact that going digital allowed the broadcasters to choose that tradeoff in the first place.

/* Steinar */

Re: Compression artifacts are a real phenomenon

Posted Feb 14, 2016 0:35 UTC (Sun) by ldo (guest, #40946) [Link]

True, but different compression techniques produce very different compression artifacts. JPEG and its DCT ilk are (in)famous for blockiness, but other techniques, like JPEG 2000, produce much less objectionable degradation in the opinion of many.

“a well-recorded vinyl LP”

Posted Feb 13, 2016 10:02 UTC (Sat) by tialaramex (subscriber, #21167) [Link] (5 responses)

You can see it happen for satellite broadcast TV too. Snow storms will sometimes do it, enough water interposed between the transmitter and receiver is just as good as a hilltop. Again in the analogue days you had the visible "snow" (broadband noise) instead and it's a tradeoff between error correction factors versus number of channels crammed into the signal which determines whether your digital signal seems better or worse.

Interlaced video looks OK played back on devices intended for it (thus mainly televisions) and crappy when reconstructed as a full frame video feed.

Neither interlacing nor composite video is analogue's fault per se. Nothing prevents you from storing component separately in an analogue system, it's just that it's not an attractive trade in terms of extra cost versus apparent improvement in quality. Indeed the exact same understanding of how little humans notice what colour things are informs the concept of chroma sub-sampling in digital video, you can get "uncompressed" video that has 4:1:0 sub-sampling, thus even though nothing was "compressed" most of the colour detail was never captured, only 4:4:4 video does what those of us from a PC world would expect and stores colour information corresponding to every individual pixel and hardly anybody bothers doing that.

“a well-recorded vinyl LP”

Posted Feb 14, 2016 7:44 UTC (Sun) by rqosa (subscriber, #24136) [Link] (4 responses)

> Neither interlacing nor composite video is analogue's fault per se.

True, but were there ever any non-interlaced and/or component-separated analogue video storage or broadcasting systems in widespread use?

> Indeed the exact same understanding of how little humans notice what colour things are informs the concept of chroma sub-sampling in digital video

However, only analogue composite video suffers from the kinds of artifacts caused by chroma interference/noise, such as rainbowing and dot crawl.

> Interlaced video looks OK played back on devices intended for it

To me, it doesn't. I remember the mid-'90s SVGA chipsets that could use interlaced video modes — and it was irritating, even headache-inducing, to keep your eyes on the screen for any significant amount of time when doing so. And with NTSC or PAL CRT television sets, I bet it would have seemed at least as bad if you were as close to the screen as you usually are with a computer monitor.

(Also, in my experience, when using SVGA video — no matter whether the monitor is a trinitron CRT, dot-mask CRT, LCD panel, or whatever else — text will look nowhere near as sharp as it does when using DVI/HDMI to an LCD panel. To me, that's the greatest reason why I don't miss CRT monitors, regardless of whatever advantages they might have over LCDs in terms of contrast ratio, color gamut, etc.)

“a well-recorded vinyl LP”

Posted Feb 14, 2016 8:55 UTC (Sun) by rqosa (subscriber, #24136) [Link]

> were there ever any [...]

Well, except for movie film stock, of course. (On a related note: notice how much better digital film scans look compared to analog-telecine film transfers.)

“a well-recorded vinyl LP”

Posted Feb 14, 2016 9:19 UTC (Sun) by farnz (subscriber, #17727) [Link]

At the broadcaster end, component video was very common, on tape formats like Betacam SP. Broadcasting-wise, B-MAC was popular in parts of Africa and Asia.

reconstructing VGA

Posted Feb 14, 2016 18:14 UTC (Sun) by tialaramex (subscriber, #21167) [Link] (1 responses)

Actually a decent LCD panel can do 100% reconstruction of the digital data behind typical SVGA signals e.g. 1280x1024

Because this is an analogue signal you need to keep the distance short, and use good quality cables and connectors but it's do-able.

The display is receiving three, parallel, signals RGB which were generated from 8-bit data. It needs to figure out what +256 and 0 will look like, get the sync right to know where each pixel will appear in the signal, and then it can turn those signals back into bits and get back the picture for crystal clear display. Good (thus expensive) panels from the end of the CRT era often incorporate electronics that do this well. You will need to display a "test card" image, including checkerboard patterns in white and each colour, and a scale, then run the "calibrate" step on the display.

On a good panel, after those steps, the picture looks exactly the same with VGA vs DVI or a more modern digital connection. Pixel perfect.

reconstructing VGA

Posted Feb 14, 2016 18:16 UTC (Sun) by tialaramex (subscriber, #21167) [Link]

s/+256/+255/ doh

“a well-recorded vinyl LP”

Posted Feb 13, 2016 12:28 UTC (Sat) by eternaleye (guest, #67051) [Link] (8 responses)

Okay, let's go point by point.

> So you're quite happy with TV that suddenly degrades to large pixellated blocks? And where you suddenly have no sound and picture for a moment?

No, but that's not the fault of digital in general - that's the fault of:
- Macroblock-based codecs (wavelet-based, among other options, degrade as blur)
- Insufficient redundancy (FEC) in the transmission stream
- Overly long GOP settings, making the TV wait longer before resync on an intra-only frame than is really sensible
- Poorly-implemented TVs tying "picture shown" to "audio stream"
- MPEG

...and I'll take it over the noise introduced in digital any day, that gave me migraines when trying to read text on anything transmitted.

> So you're quite happy with music that has gaps in it? Or clicks as artifacts from digitisation?

No, but that's not the fault of digital in general - that's the fault of:
- People who don't author properly (there are plenty of gapless CDs)
- People who don't rip properly (take a look at tools like Morituri, an open-source, commandline tool to generate exact rips
- People who don't configure their audio players for gapless properly - though many come properly configured out of the box
- People who don't understand the Shannon-Nyquist theorem, and that it proves that digital *perfectly* captures any signal with wavelength less than half the sampling period (there are some issues if the signal even *contains* higher frequencies, but... that's what band-pass filters are for.)

> Oh - and can you hear a digital loudspeaker - indeed, is there even such a thing as a digital loudspeaker?

I don't know, but the analog ones don't defend themselves well. To quote an old Usenet post:

> bing-bong. Brimish Rull regret that mumble maz bem dermumble a mir mumble mumble bimble late. Passengers mizzing to mumble rimble mumble are advised to momble mar at murmble. Thank you mor mumble mimbling Brimble mum. bing-bong.

> Analog formats have a LOT of advantages. Not least that the original format is analog so we don't get the damage and entropy of a format conversion.

No, instead we get damage and entropy simply as a function of time spent in existence, and we can't even use a checksum to detect it.

> Okay, so we can COPY digital without adding entropy, but digital is not a magical panacea ... (and too many people DO think digital is "magic", so those of us who actually understand its limitations actually find it hard to enjoy its advantages).

Oh, you can do _many_ things with digital without adding entropy. Any reversible function over the sample format (float or int), in fact, which includes a great many things indeed - including mixing tracs, so long as the summation function is perfect (stick to int-based formats, for that) and you don't hit the limit of the sample value. And that last can be ensured by... adding one bit to the depth, as summing two numbers cannot be more than twice the maximum value of the arguments.

> I gather that quite a lot of modern CDs, nowadays, aren't made from WAV or FLAC master recordings, but MP3!!! How stupid is that? To create a "lossless" recording from a lossy master!!!

Pretty stupid. Stupid enough to screw up mastering vinyl in similar ways.

Why is it that on Vinyl, the pops are "warm" and the sound is "natural", but on digital everything is an affront? I truly fail to comprehend this mindset.

“a well-recorded vinyl LP”

Posted Feb 13, 2016 12:34 UTC (Sat) by eternaleye (guest, #67051) [Link]

Er, "I'll take it over the noise introduced in analog"

(To make this more than a typo correction: Track down an old analog episode of Jeopardy sometime. It was nearly the only thing I watched on TV.)

“a well-recorded vinyl LP”

Posted Feb 13, 2016 16:56 UTC (Sat) by tialaramex (subscriber, #21167) [Link] (4 responses)

One small thing you missed, which I'll point out just in case you didn't know

Sampling does introduce quantisation noise. When we do digital sampling we take a measurement, but the measurement cannot be infinitely precise. So e.g. on a cheap 16-bit PCM ADC maybe the exact measurement would have been +10383.42961 but what's actually read is +10383. The effect of this imprecision on its own is to introduce a small amount of signal dependent noise.

A more sophisticated ADC adds dither to the signal, this allow us to either whiten the noise (white noise is less offensive to hear than the signal dependent noise) or to create a shaped dither where the worst noise is moved to a higher frequency where human hearing is less sensitive. However there's no way to avoid noise altogether.

At 8-bit PCM quantisation noise is quite noticeable in an otherwise good quality recording, it doesn't leap out at most people (some people really don't like white noise), but it's certainly there if you listen.

While consumers have proved happy with such poor quality at a good price in the distant past, we don't need to compromise because today 16-bit PCM means the noise floor is almost 50dB lower, if you turned up the volume enough to hear the noise now in an otherwise silent room (think, very quiet reading room in a library in the countryside), the loudest passages of the music recording the noise is on would be as loud as a power mower. Now rock and roll music is played much louder than this live, but that doesn't matter because the people listening to it are being deafened (this is called "temporary threshold shift" and is why everything seems really really quiet when you leave the concert) so they can't hear any low level noise anyway.

For 24-bit PCM, this is all academic because the levels involved for practical audio systems are such that quantisation noise is swamped by an inherent property of the universe called thermal noise. Thermal noise is yet another white noise source, but this time one you can't dodge with any practical audio system, whether analogue or digital. Scientists who do need to reduce thermal noise for highly sensitive instruments use liquid nitrogen cooling which makes laser record players look like bargain bin technology. Digital audio actually got so good we ran into a limit set by the laws of physics. In practice thermal noise limits our music recording or playback systems to somewhere around 20 bits of precision even assuming all the analogue gear (e.g. microphones, speakers, amplifier) was somehow perfect. But of course in software no such limit applies during synthesis, mixing, or effects.

“a well-recorded vinyl LP”

Posted Feb 13, 2016 18:35 UTC (Sat) by Sesse (subscriber, #53779) [Link] (1 responses)

And more importantly, the signal-to-noise ratio of even a great vinyl record is nowhere near that of 16-bit PCM.

/* Steinar */

“a well-recorded vinyl LP”

Posted Feb 15, 2016 10:09 UTC (Mon) by ms (subscriber, #41272) [Link]

Not only that, the LP is pretty much awful at channel cross talk (~30dB IIRC?).

“a well-recorded vinyl LP”

Posted Feb 13, 2016 21:34 UTC (Sat) by Cyberax (✭ supporter ✭, #52523) [Link] (1 responses)

> Scientists who do need to reduce thermal noise for highly sensitive instruments use liquid nitrogen cooling which makes laser record players look like bargain bin technology. Digital audio actually got so good we ran into a limit set by the laws of physics.
/me runs off to a patent office to file a patent for "nitrogen-cooled high-fidelity audio LP-player".

I bet it'll be popular among the audiophile crowd.

“a well-recorded vinyl LP”

Posted Feb 19, 2016 20:55 UTC (Fri) by smurf (subscriber, #17840) [Link]

… until one of them lifts an LP off that supercooled player too quickly.

It'll shatter quite nicely.

“a well-recorded vinyl LP”

Posted Feb 15, 2016 10:14 UTC (Mon) by ms (subscriber, #41272) [Link] (1 responses)

> Pretty stupid. Stupid enough to screw up mastering vinyl in similar ways.

Exactly. I do wonder if a lot of this "LP stuff sounds much nicer" is because of the idiotic ways in which most music is produced. If you have a niche format which is inaccessible to most people and carries a much higher price tag, then you may well, at least initially, get people involved who on average know a bit more about how to get the best out of it. But as soon as everyone realises there's money to be made, it'll be awful again.

An awful lot of what sounds awful isn't to do with the recording or the medium. It's because you're listening to clipped waveforms because everything has been compressed badly to ensure the entire track is at 0dBFS throughout the entire track. It's that that is fatiguing. If you listen to CDs produced in the 1970s, when most of the sound engineers had grown up with LPs and the trend to use every single bit in every single sample didn't exist yet, the results were excellent.

“a well-recorded vinyl LP”

Posted Feb 19, 2016 21:03 UTC (Fri) by smurf (subscriber, #17840) [Link]

Actually, I beg to differ. Sound engineers were used to LPs softening almost any sound you throw at them, so they compensated (as much as possible). The first CDs were mixed the same way PLUS there's no physical limits to what you can do WRT sound levels (as opposed to LPs where you can easily record a loud bass in a way that'll cause players to skip), but there's no softening at the listener's end, so they sounded unusually harsh, which probably started the "CDs sound unnatural" myth that was true once, but persists to this day … primarily among the crowd that also buys gold-plated hyper-fine-stranded HDMI cables because they reduce the lag between picture and sound, y'know.

“a well-recorded vinyl LP”

Posted Feb 13, 2016 21:14 UTC (Sat) by malor (guest, #2973) [Link] (16 responses)

>So you're quite happy with TV that suddenly degrades to large pixellated blocks?

The nice thing about digital is that, when it fails, you *know* it failed. It's not subtle. It's either exactly what it's supposed to be, or it's obviously wrong. This is pretty powerful advantage; you immediately know if you're not getting what you're supposed to. You don't get gradual degradation, but you shouldn't have any degradation.

And if you're referring to the video formats that are used in digital television, those aren't problematic in lossy situations because they're digital, but because they're highly compressed and have no redundancy. If they wanted to spend more bandwidth, they could make failures much less obvious, or even entirely invisible.

In any case, those problems aren't related at all to CD format, which is entirely different. About the only thing that WAV and MP4 files have in common is that they're represented digitally.

>So you're quite happy with music that has gaps in it? Or clicks as artifacts from digitisation?

Gapless playback has been a thing in reasonable music players for a long time. Clicks are artifacts of bad cuts, where the digital file doesn't end on a 0. Neither of these things should happen with competent playback and ripping. You can always find someone or something doing it wrong, I'm sure, but these are easily avoidable, and have been for, hmm... at least a decade? (digital pops have been easily avoidable since the beginning of digital files, but gapless MP3 playback came along more recently. And gapless has never been an issue with WAV or FLAC, just MP3.)

>Oh - and can you hear a digital loudspeaker - indeed, is there even such a thing as a digital loudspeaker?

I'm not sure what this question even means?

>Analog formats have a LOT of advantages.

Not really. They do have one: they can't be abused in the Loudness Wars. Mix engineers love them some clipping, and it's very, very common for digital music to be mixed well past 100%. I find this extremely frustrating, and this is a major advantage of vinyl: if engineers tried that kind of shit with a record, the needle would jump right out of the groove. They're forced by the format itself to create good mixes.

But the abuse of digital files is a testament to their power. When not abused, a correctly mixed digital file will be superior in every way to an analog one. The noise floor will be far lower and the frequency response will be literally perfect. If an original waveform can't be represented *exactly* by a CD, then it has components above 20KHz, full stop.

With the shitty mixes that are often on CD, you may very well be better off with a relatively lousy reproduction of a superior analog mix. But that's humans messing up the sound, not the format.

>but digital is not a magical panacea

If it's done right, with audio anyway, it's damn close. A 44.1KHz CD signal will produce literally perfect sound up to 20KHz. There are some people who get antsy about the ultrasonic ranges, but as someone who could hear up to about 25KHz in my youth: that noise up there is just annoying. It sucks. It's not music, it's just nasty.

If you're totally anal about it, you can go to a 96KHz signal, but in all honesty, that's dumb. And there's no need for 24-bit files, either. They're nice for mastering and mixing, but once that process is finished, a final mixdown to 16 bits is gorgeous.

Sometimes, they get the engineering so perfect that there is no need for another solution. CD format is an excellent example. For any situation where stereo sound is adequate, there will never need to be anything else. They got it in one, and as long as humans have the same ears that we have now, it will not need replacement.

Adding more channels is certainly an option, but those channels can just be more of the same thing.

“a well-recorded vinyl LP”

Posted Feb 15, 2016 11:49 UTC (Mon) by nye (guest, #51576) [Link] (15 responses)

>If an original waveform can't be represented *exactly* by a CD, then it has components above 20KHz, full stop.

This is true for the final result, but getting it exactly right does require some degree of competence in the digitisation process. The following is largely academic, and you probably know it anyway, but as a colossal nerd I'm compelled to point it out.

In order to avoid the possibility of aliasing artefacts, you need to ensure that there are no components higher than 22kHz, so you need to filter them out. In reality, no analogue low-pass filter is exactly perfect, so you end up attenuating some frequencies that are below the filter cut-off - how bad this effect is will depend on whether you spent 5 pence on your filter or splashed out and went up to 50 (well, I made these numbers up, but I'm sure we're all familiar with manufacturers who will inexplicably choose to cripple some product in order to save pennies per unit).

So there are two options then. The first is not to care - the result won't be a perfect representation of the original sound up to the sampling rate of the format, but a tiny amount of attenuation at the very top end will be almost certainly inaudible to basically everyone. The other is to do all of your filtering and digitisation at some frequency far higher than you care about and then downsample and filter it digitally. In practice people making digital artworks always like working at a higher resolution than the final result will ever need so that they never need to worry about artefacts introduced during the process, so you're probably going to want to digitise at a higher frequency anyway.

>Gapless playback has been a thing in reasonable music players for a long time

Do you know of any cross-platform FOSS audio players that can do this reliably? I remember getting gapless playback about 15 years ago, but it's something that I've not seen working in a long time.

>And there's no need for 24-bit files, either. They're nice for mastering and mixing, but once that process is finished, a final mixdown to 16 bits is gorgeous.

I think people should be asking for 24-bit audio for the same reason I think they should be asking for source code - not because it improves the final result in any way, but because it is the preferred format for modification.

preferred format

Posted Feb 15, 2016 20:02 UTC (Mon) by tialaramex (subscriber, #21167) [Link]

The "preferred format" for further modification of almost any modern audio recording, even including an orchestral work that's notionally a "live" recording is a multi-track with a bunch of automation and mixing metadata.

Ideally the multitrack would be 24-bit. Sure. But if you want to do something other than just listen to it, you are MUCH better off having two dozen tracks of 16-bit PCM and a pile of EQ settings than a stereo 24-bit PCM version of the "finished" recording from which you've got to separate out vocals, lead guitar, drums and so on to actually get anything else done.

Trent Reznor's bands "Nine Inch Nails" and "How To Destroy Angels" release semi-final tracks (so e.g. "vocals", "drums" but not e.g. "Take #4 of Trent yelling a word") explicitly so that you can re-mix his work. You could definitely make "Guns by Computer" from "Hyperpower" just using a 24-bit PCM FLAC, or indeed a mediocre MP3. But Paul Epworth's remix of "Capital G" really takes the original to pieces and builds something quite different which would be much harder without all the component tracks.

Of course hobbyist remixers often assemble things that nobody would have authorized, and they're usually not working from a 24-bit audiophile edition of anything. Isosine's Shake It Off (the Perfect Drug) mashes Taylor Swift into an older NiN track from before there were official multitracks of their CDs. Unofficial multitracks of The Perfect Drug are widely available. Trent doubtless has no problem with that, but as to the other half of the equation it's hard to imagine Taylor's label authorised this inspired bit of tampering with her pop hit, so presumably Isosine had to pull that to bits by hand which sucks.

“a well-recorded vinyl LP”

Posted Feb 16, 2016 9:36 UTC (Tue) by jezuch (subscriber, #52988) [Link] (12 responses)

> Do you know of any cross-platform FOSS audio players that can do this reliably? I remember getting gapless playback about 15 years ago, but it's something that I've not seen working in a long time.

Clementine works fine for me, even when the directory it plays from is sshfs-mounted over a not-very-snappy connection :) (Don't ask why I do this.)

But I know what you mean. At one time I stoped expecting gapless playback to work, so now when it works fine it's a pleasant surprise ;)

“a well-recorded vinyl LP”

Posted Feb 16, 2016 13:14 UTC (Tue) by nye (guest, #51576) [Link] (11 responses)

>Clementine works fine for me, even when the directory it plays from is sshfs-mounted over a not-very-snappy connection :) (Don't ask why I do this.)

Damn, that's what I'm using (except for the sshfs part) and it's not working for me :(.

On the other hand, that does show that it *could* work. I need to investigate this further...

“a well-recorded vinyl LP”

Posted Feb 16, 2016 14:59 UTC (Tue) by malor (guest, #2973) [Link] (10 responses)

As an aside, which may not be useful for you: I understand that the typical "gap problem" is an artifact of how the MP3 format works. I normally use FLAC, which doesn't seem to have the problem (I rarely listen closely enough to be sure), and I see on Wikipedia that Vorbis has the inherent ability to eliminate gaps entirely. (Ogg is the container format, Vorbis is the music compression algorithm.)

It may not be worth re-ripping and re-encoding for this one problem, but there's also a nice secondary effect: once you've generated known-good CUE/FLACs, and preferably some PAR2s for redundancy, you can then generate any other format you want without ever having to rip again. I've used this repeatedly over the years, most recently to convert my FLACs to 160KMP3 for a Sansa player.

I used Foobar on Windows, which is not open source, but damn, it is fantastic software. It feels very Unixy. Watching it automatically spool up eight threads because it detected eight processors, and then chewing through the whole library at something like 500x playback speed, complete with perfect directory structure and tag conversion, was impressive as hell.

“a well-recorded vinyl LP”

Posted Feb 17, 2016 8:32 UTC (Wed) by jezuch (subscriber, #52988) [Link]

> As an aside, which may not be useful for you: I understand that the typical "gap problem" is an artifact of how the MP3 format works.

Right. Forgot about that. Maybe because the only non-Vorbis/Opus files I have are the ones that I haven't ripped myself. And if I think something's worth listening to, I usually have the CD of it :)

“a well-recorded vinyl LP”

Posted Feb 17, 2016 11:46 UTC (Wed) by nye (guest, #51576) [Link] (8 responses)

>As an aside, which may not be useful for you: I understand that the typical "gap problem" is an artifact of how the MP3 format works. I normally use FLAC, which doesn't seem to have the problem (I rarely listen closely enough to be sure), and I see on Wikipedia that Vorbis has the inherent ability to eliminate gaps entirely. (Ogg is the container format, Vorbis is the music compression algorithm.)

Almost all of my music is already FLAC anyway. I think Clementine for Windows has some problems, and I'm kind of hopeful that it might be due to the old version of Gstreamer (and other libs) that's still used in the latest release build - hopeful because it looks like the dev version can be easily linked against newer versions, so if I can get it to build maybe it will fix itself (and I can go out and buy that lottery ticket).

>I used Foobar on Windows, which is not open source, but damn, it is fantastic software

I have used it in the past, but I get frustrated having to switch applications whenever I switch platforms so I try hard to use cross-platform applications whenever there are decent options.

I've never found anything that handled transcoding well enough to satisfy me, so I ended up writing my own tool (well, wrapper around the various format-specific tools).

“a well-recorded vinyl LP”

Posted Feb 17, 2016 15:13 UTC (Wed) by malor (guest, #2973) [Link] (7 responses)

>I've never found anything that handled transcoding well enough to satisfy me, so I ended up writing my own tool (well, wrapper around the various format-specific tools).

Foobar2K is just amazing at this. It typically needs the EXE or DLL associated with the format you want to convert to (lame.exe for MP3, for instance), and then can drive a multi-threaded conversion effortlessly, including getting all the tags right. Even if you don't want to use it as a player, if you have easy access to Windows, it's an extraordinarily good tool for converting files around. It knows many of the flags for LAME, and has a number of smart presets, but can drive it with custom flags if you prefer.

It's also superb at manipulating and tagging files; you can easily reorganize huge libraries in single-digit seconds based on their tag information, or tag huge libraries in double-digit seconds based on their directory/path location. It knows multiple tag formats, and lets you specify which one you want to use.

Even if you never touch it for playback, it is the best tool I know for bulk manipulation of audio files. It's really weird that a tool that feels so very Unixy is Windows-only.

It's a damn shame it's not open source. I suspect it would be the flag standard everywhere if it were. But even closed-source, it doesn't have a lot of potential for abuse or lock-in by the author, so it doesn't seem at all dangerous to use.

It sounds like it won't work as a player for you, but you may really appreciate having it in the toolbox.

“a well-recorded vinyl LP”

Posted Feb 20, 2016 15:39 UTC (Sat) by paulj (subscriber, #341) [Link] (6 responses)

A tool I'd love is something to detect and manage duplicates in my Mp3 library (e.g. family who copy my MP3s to their collection, add some of her own, then copy it all back to my collection in different places with different names thanks to their MP3 player mangling stuff; etc). There's some acoustic-ID stuff out there, and a GUI, but it's not brilliant at detecting dupes and even when it does the UI is pretty bad for dealing with them.

Sigh... 1st world problems.

“a well-recorded vinyl LP”

Posted Feb 22, 2016 9:45 UTC (Mon) by jezuch (subscriber, #52988) [Link] (5 responses)

> A tool I'd love is something to detect and manage duplicates in my Mp3 library

If MusicBrainz is like Shazaam but not non-free, then I guess that would be the answer...

“a well-recorded vinyl LP”

Posted Feb 23, 2016 2:38 UTC (Tue) by mathstuf (subscriber, #69389) [Link] (4 responses)

MusicBrainz is a community-driven database of, usually released, music. So it has UUIDs for recordings, artists, albums, etc. However, these are usually manually maintained (and tags used to help) rather than based on the actual audio content. There are FOSS audio fingerprinters though (Amarok had a plugin at least).

“a well-recorded vinyl LP”

Posted Feb 23, 2016 10:04 UTC (Tue) by robbe (guest, #16131) [Link] (3 responses)

MusicBrainz works closely with AcoustID
https://acoustid.org/

“a well-recorded vinyl LP”

Posted Feb 23, 2016 13:34 UTC (Tue) by mathstuf (subscriber, #69389) [Link] (2 responses)

Ah, the feature is not exposed in Picard (my usual MusicBrainz interaction point). Or I haven't found it.

acoustid & picard

Posted Feb 24, 2016 10:22 UTC (Wed) by robbe (guest, #16131) [Link] (1 responses)

In picard, go to Options > Options > Fingerprinting, tell it to
»Use AcoustID«. You’ll need whatever package your distribution puts /usr/bin/fpcalc in (libchromaprint-tools in Debian), and an API key for the webservice.

Then functionality is behind the Tools > Scan or the Scan button.

acoustid & picard

Posted Feb 26, 2016 9:26 UTC (Fri) by paulj (subscriber, #341) [Link]

Yeah, that's very interesting. Though, last I used it, it is not easy to then do something about duplicates it might find. I see there's now "beets" (http://beets.io/), a music manager with support for musicbrainz and acoustid. Will have to try that sometime.

“a well-recorded vinyl LP”

Posted Feb 16, 2016 18:15 UTC (Tue) by farnz (subscriber, #17727) [Link]

Note, of course, that my "44.1 kHz" audio ADC can be implemented as an analogue -18 dB per decade low pass filter, -3 dB point at 22 kHz, a 22 MHz unfiltered ADC, and a digital filter to turn the 22 MHz signal into a 44.1 kHz signal. Thus, while there's reasons to have the raw ADC sample at a huge sample rate, there's no reason to have output from it at that high sample rate.

“a well-recorded vinyl LP”

Posted Feb 13, 2016 21:38 UTC (Sat) by Cyberax (✭ supporter ✭, #52523) [Link]

> So you're quite happy with music that has gaps in it?
FWIW, all conforming audio CD players must support gapless playback. It's a part of the Red Book standard.

“a well-recorded vinyl LP”

Posted Feb 13, 2016 22:35 UTC (Sat) by MattJD (subscriber, #91390) [Link] (4 responses)

> Analog formats have a LOT of advantages. Not least that the original format is analog so we don't get the damage and entropy of a format conversion.

The only reason I know of why analog is better (that isn't beaten by math) is that analog mediums tend to distort the music in a pleasant way. When people talk about digital being harsh, versus analog being warm, it isn't a problem with digital. It's that digital outputs almost exactly what is recorded (better then analog at least). Analog's distortion tends to sound pleasant to people, which causes people to prefer it.

Of course, with digital one could always apply a filter that mimics the distortion of analog. It's just that people haven't. Or you get an audiophile who claims you can never mimic it.

Re: analog mediums tend to distort the music in a pleasant way.

Posted Feb 14, 2016 8:33 UTC (Sun) by ldo (guest, #40946) [Link] (3 responses)

No they don’t. At least, not in general.

Go to some audiophile chatroom, just ask whether anybody prefers “transistor sound” over “vacuum-tube sound”, then stand well back...

Re: analog mediums tend to distort the music in a pleasant way.

Posted Feb 15, 2016 12:17 UTC (Mon) by nye (guest, #51576) [Link] (2 responses)

>Go to some audiophile chatroom, just ask whether anybody prefers “transistor sound” over “vacuum-tube sound”, then stand well back...

You can replicate "vacuum-tube sound" by designing your transistor amp to have a high output impedance, but you probably shouldn't.

Speaker impedance changes with frequency, and as the amplifier's output impedence rises, the speaker's impedance has an increasing effect on the voltage across it. In general, an idealised amplifier with a goal of maximising fidelity would have an output impedance of zero, so as to minimise the non-linearity in the speaker's frequency response.

Because of this, a goal in designing valve amplifiers is to get the output impedance as low as possible; solid state amplifiers represent a major breakthrough here. It's ironic that the same people who were chasing low output impedance for years, suddenly decided they didn't like it when they got it - presumably because the breakthrough was so sudden that they never got the chance to slowly progress through it and just don't like that it "sounded different".

Different people like the sound of different things, and subjectively it's entirely valid for someone to prefer the sound of a valve amplifier, but what they're liking is the way in which the sound is distorted. Objectively, assuming your metric is output fidelity, transistor amplifiers are better.

Re: Objectively, assuming your metric is output fidelity, transistor amplifiers are better.

Posted Feb 15, 2016 19:49 UTC (Mon) by ldo (guest, #40946) [Link]

And objectively, digital is better than analog.

Yet there are still people who feel this irrational attachment to the old technology.

Re: analog mediums tend to distort the music in a pleasant way.

Posted Feb 16, 2016 18:21 UTC (Tue) by malor (guest, #2973) [Link]

I remember reading somewhere that there's a thing with harmonics, where transistors produce even-numbered harmonics and tubes produce odd, and some people prefer the sound of the odd harmonics.

It could be the other way around too, that transistors do odd and tubes do even.... I just know they were opposites, not which was which.

“a well-recorded vinyl LP”

Posted Feb 13, 2016 17:54 UTC (Sat) by rsidd (subscriber, #2582) [Link] (27 responses)

I feel no regret to see analog formats like vinyl and tape consigned to museums. That’s where they belong.

Tape, maybe. Vinyl... well, I have a stack of CDs dating back 20-25 years. I've ripped most of them but was recently ripping the remainder. Quite a few were unplayable, including most of the CD-Rs.

I also have a stack of LPs obtained from my parents. Most of them are still playable and sound quite good.

An LP pressed in the 1950s and rarely played will sound as good as new. All digital formats degrade quickly to the point of unplayability. Having ripped my CDs I will need to keep at least two copies and periodically check the media for errors, making another copy if necessary.

Plus the point of open format is very valid. The two Voyager spacecraft in the 1970s carried gold LPs carrying the sounds of the earth. There was at least some hope that any alien civilization discovering them would figure out how to convert the medium into sound. With CDs and other digital media, not only do they degrade, but there is no hope of instructing aliens how to decode them (at least without teaching them our language).

Re: All digital formats degrade quickly to the point of unplayability

Posted Feb 14, 2016 0:33 UTC (Sun) by ldo (guest, #40946) [Link] (23 responses)

Make that “all formats degrade eventually to the point of unplayability”, and I will agree with you. I have had lots of optical discs go bad, but they were invariably from a brand that I learned to avoid.

The beauty of digital is that you can make 100% perfect copies before you get to that point.

Re: All digital formats degrade quickly to the point of unplayability

Posted Feb 14, 2016 3:41 UTC (Sun) by rsidd (subscriber, #2582) [Link] (22 responses)

The beauty of digital is that you can make 100% perfect copies before you get to that point.

I don't dispute that. But someone has to make the copies. We can read manuscripts from 1000 years ago, sometimes even older. Will today's literature be available, in any form, 1000 years from now? Even when it is printed on paper, the quality of the paper is so terrible it falls apart in 50 or 60 years.

In the case of LPs, there are pristine copies of much important work available in, say, the Library of Congress. These can easily last hundreds of years. Digital media can, too, but only if the curators take care to maintain and update them.

For other analog media like tape and film, yes, their lifetime is limited and digitisation is a huge benefit. But again, films from 100 years ago are still recoverable. When we can barely read word processor documents from the 1980s, are we sure that today's h.264 films will be readable and decodable in the year 2116? Not just the famous films, but the chance stuff that someone finds in a box in a warehouse?

Re: All digital formats degrade quickly to the point of unplayability

Posted Feb 14, 2016 3:42 UTC (Sun) by Cyberax (✭ supporter ✭, #52523) [Link]

Why not? h264 is not something terribly esoteric. We can read EBCDIC files, after all.

Re: there are pristine copies of much important work available in, say, the Library of Congress

Posted Feb 14, 2016 4:07 UTC (Sun) by ldo (guest, #40946) [Link] (2 responses)

“Pristine”, did you say?

Re: there are pristine copies of much important work available in, say, the Library of Congress

Posted Feb 14, 2016 6:04 UTC (Sun) by rsidd (subscriber, #2582) [Link] (1 responses)

Yes, pristine I said, about LPs. Your link is about film and tape. Also, your link links to this which makes the same points about digital that I did.

Re: Yes, pristine I said, about LPs

Posted Feb 14, 2016 8:30 UTC (Sun) by ldo (guest, #40946) [Link]

Read the article again, where it says LPs can last maybe 150 years. Don’t forget some of them would be coming up to a century old.

What are we supposed to do with them in 50 years’ time?

Film - great example

Posted Feb 14, 2016 16:56 UTC (Sun) by tialaramex (subscriber, #21167) [Link] (6 responses)

Films from 100 years ago are NOT recoverable _unless_ somebody spent a lot of money and effort specially conserving them, for decades. Films from the 19th century used "nitrate film" which is chemically far from inert, and got a bad reputation because it's really, really enthusiastic about catching fire, a bad choice when the whole idea is to make something incredibly hot and shine the light from it through the film. Unless deliberately conserved by experts (are you an expert? No, you just put your CDs in a cupboard or left them on a shelf) the nitrate film will have long since crumbled into useless powder.

The next idea they had, at the start of the 20th century, was "safety film". Much like other safety products (e.g. CFCs, a safe coolant for refrigerators) the safety film turned out to have enormous problems that weren't immediately apparent. Specifically, unless kept under very carefully controlled conditions it experiences the "vinegar syndrome" a strong vinegar-like smell is the first sign, then it becomes brittle, shrinks and buckles. You now have a pile of very fragile pieces of film, good luck assembling them into a movie. Here's a thought: step one, digitise everything you've recovered.

Film - great example

Posted Feb 14, 2016 17:20 UTC (Sun) by rsidd (subscriber, #2582) [Link] (5 responses)

Look, digital media today won't do any better unless you keep backing it up to fresh media. Or unless your medium is this.

Film - great example

Posted Feb 14, 2016 17:49 UTC (Sun) by tialaramex (subscriber, #21167) [Link] (3 responses)

Right, but I mean this is Sunday, it's back-up day. My script said it's time for Wendy drive, so I plugged the drive with "Wendy" on the label and it runs. Next week it'll be "Sarah" apparently. I will take Wendy away and bring Sarah back.

These days I'm using cheap USB spinning rust for the backups. I wanted a pretty purple drive for "Wendy" but they only had boring silver colour in stock, that's why it's "Wendy" and "Sarah" and not "Green", "Purple", etc.

Ordinary people are too lazy for backups. But, they're also too lazy to care about privacy, so they're all backed up in the cloud anyway.

Film - great example

Posted Feb 14, 2016 17:52 UTC (Sun) by rsidd (subscriber, #2582) [Link] (1 responses)

And after you've been dead 50 years and your grandchildren are digging through the archives for family photographs?

Film - great example

Posted Feb 14, 2016 23:32 UTC (Sun) by tialaramex (subscriber, #21167) [Link]

All the drives are encrypted, it was never intended that anybody's grandchildren would "dig through" my archives. When I die the immediately valuable assets will be divided among a small group of people according to their preferences, (or if they cannot agree, equally between them all) and most likely the various encrypted drives and devices will be destroyed or re-purposed. It is quite enough of an imposition that I make them decide how to split up the real estate and the money, without asking that they inherit some bizarre emotional attachment to data they previously had no idea existed.

Anybody looking for "family photographs" would be disappointed anyway, I don't usually take photographs of people, nor of famous monuments and other tourist attractions. The world already has plenty of photographs of the Eiffel Tower and the Empire State Building as it is. When I take photographs that I think other people might want, I upload them to be shared. Likewise I upload software that's going to be re-usable to github and/or my web sites. But mostly I take photographs only as a visual note to myself, they're of no plausible interest to anybody else.

Not so very long ago my last grandmother died. Her remaining children fussed about the photographs, the little scraps of paper tucked into jewellery boxes, the cryptic hand-written annotation in books. They wondered who the old lady is in the picture with the babies, what the significance was of the nice big house in another photograph, they tried to put 2 + 2 together to make a number big enough to represent my grandmother's whole life, and they couldn't do it.

I'm not a sentimental person. A romantic yes, but not sentimental.

Film - great example

Posted Feb 14, 2016 17:54 UTC (Sun) by rsidd (subscriber, #2582) [Link]

ps -- if you're merely alternating between Wendy and Sarah, they'll both die about the same time and you're screwed. You should change partners, er, drives, every couple of years (ie, each drive gets turned over after 4-5 years, in a staggered manner).

Re: unless you keep backing it up to fresh media

Posted Feb 16, 2016 3:12 UTC (Tue) by ldo (guest, #40946) [Link]

Of course. How else are you supposed to keep your data? Ask any librarian or archivist, and they will tell you that maintaining any kind of records, from whatever era of technology, is an ongoing process—you don’t just leave them to moulder on a shelf or in a box.

If anything, digital makes the process even easier, since computers can automate much of the process.

Re: All digital formats degrade quickly to the point of unplayability

Posted Feb 14, 2016 17:41 UTC (Sun) by tialaramex (subscriber, #21167) [Link] (10 responses)

H.264 is documented as "ISO/IEC 14496-10 – MPEG-4 Part 10, Advanced Video Coding"

Is it your contention that not only ISO itself, but all copies of its standards will be destroyed or forgotten by 2116?

For reference ISO is approximately 70 years old, it is the successor institution to an earlier ISA which would be about 90 years old. It is not technically part of the UN, and so could survive the loss of that body although most scenarios that would destroy the UN (e.g. multi-lateral global nuclear war) would presumably also destroy ISO.

Re: All digital formats degrade quickly to the point of unplayability

Posted Feb 14, 2016 18:10 UTC (Sun) by rsidd (subscriber, #2582) [Link] (9 responses)

I said "Are we sure that h.264 films will be readable and decodable in the year 2016?" There is a big question mark on readability on current media that is not actively backed up on fresh media. As for decodability, I asked "are we sure?" Are you sure? There is simply no historical precedent for this situation. But yes, if the documentation of the format is preserved and the document is sufficiently important, someone will figure out how to read the file.

Picture this situation: by 2060, h.375 is the overwhelmingly dominant standard, and all actively maintained backups have been transcoded to this standard. This includes archives in the Library of Congress, MPPA, libraries around the world, etc. But in 2116, someone discovers this incredibly important film by some guy from 100 years ago called Christopher Nolan, which he had done on his home camera and saved to a CD-R that was somehow still readable after all those years. In h.264, which nobody had used for over 56 years. Would an archivist be bothered to write a decoder for that? Probably yes, given Nolan's importance. Now imagine it was not Nolan, but some unknown genius and this was random junk. Would anyone write a decoder for that? Probably no.

(I'm not even going into the question of filesystems here! ext3 is expected to be removed from near-future linux kernels. How do we know today's filesystems, even if readable, can be read in 2116? Our future archiver would need to write both a filesystem reader and a decoder for the codec. That filesystem may be iso9660, or udf, or ext4, or ntfs, or btrfs, or zfs, or hfs+, or something else. Can you imagine the effort involved in 2016 to write a program to read these systems?)

Finally, consider that J.S. Bach died in 1750, and was all but forgotten among the general public for decades after -- but a few enthusiasts (including Mozart and Beethoven) had copies of his scores and were influenced by them, and finally Mendelssohn performed the St Matthew Passion in 1829 and everybody rediscovered the old man, because the scores were still readable, because they were written/printed on good-quality paper. Consider that the solo cello works were pretty much forgotten until the 20th century (Casals), and much of the keyboard oeuvre likewise (Landowska and others).

Re: All digital formats degrade quickly to the point of unplayability

Posted Feb 14, 2016 20:47 UTC (Sun) by anselm (subscriber, #2796) [Link] (1 responses)

That filesystem may be iso9660, or udf, or ext4, or ntfs, or btrfs, or zfs, or hfs+, or something else. Can you imagine the effort involved in 2016 to write a program to read these systems?

I think you mean “2116”. But it doesn't really matter – people will run a 2016 Linux PC as a virtual machine on a 2046 computer that is really a virtual machine on a 2076 computer which is a virtual machine on a 2106 computer and so on. IBM has been doing that sort of thing for a very long time.

(Going back 30 years doesn't seem to be a real problem in practice. Last Saturday was the 30th anniversary of the first ROM-based TOS for the Atari ST, and I celebrated that by looking at some old Atari-based stuff I used to do at that time. There are various Atari ST emulators on Linux that run that code just fine, and way faster than the actual ST ever did.)

Re: All digital formats degrade quickly to the point of unplayability

Posted Feb 15, 2016 0:31 UTC (Mon) by tialaramex (subscriber, #21167) [Link]

The popular Amiga emulator (UAE) is actually so good that the "next generation" Amiga OS 4.1FE (designed to run on "new", yes really, PowerPC-based hardware manufactured this decade) is suspected to have sold far more copies to be run in the emulator than on the real hardware it's purportedly made for.

Why buy a $2000 computer when you could buy a $50 emulator package, a $30 OS CD image, and be up and running the same hour you were inspired rather than in 4-6 weeks plus shipping?

Re: All digital formats degrade quickly to the point of unplayability

Posted Feb 14, 2016 22:10 UTC (Sun) by Cyberax (✭ supporter ✭, #52523) [Link] (5 responses)

It's software! An archivist won't need to _write_ a decoder, it's quite likely to be preserved in some kind of an archive somewhere.

And even if all copies of the h264 decoder are somehow lost, an archivist would need to write only ONE copy of transcoder and then publish it on their BrainBook page. After that anybody will be able to use it.

Now, what are you going to do with a crumbling cinema tape reel from 1950? Are you sure you're going to spend money/time to read it?

Re: All digital formats degrade quickly to the point of unplayability

Posted Feb 15, 2016 3:54 UTC (Mon) by rgmoore (✭ supporter ✭, #75) [Link] (4 responses)

An archivist won't need to _write_ a decoder, it's quite likely to be preserved in some kind of an archive somewhere.

Especially because it isn't something that's going to be suddenly coming up in 2116 after not having been an issue for a really long time. In reality, there will be a constant, if possibly low-level, need to go back and look at old media, so it's likely there will be an on-going effort to maintain decoders. If there's a problem, it will be with files stored on obsolete hardware that nobody can interface to, and that will probably be solved for most people by cloud storage.

And that's assuming that we give up on current formats, which is by no means assured. For many kinds of data, we have standardized formats that don't seem to be going anywhere. The JPEG format is well over 20 years old, and it's still the dominant format for photographs. Everyone wants a camera that takes JPEG pictures because that's what the picture viewing software understands, and all the picture viewing software needs to understand JPEG because that's what all the cameras use. And now there are literally trillions of JPEG pictures out there, so any competing format has a substantial barrier to overcome in getting established. I wouldn't be at all surprised if people 50 and 100 years from now were still using JPEG as their main format for photographic images. We aren't quite as far along in standardizing for other kinds of media, like movies, but at some point there will be a "good enough" format that everyone will stick with.

Re: All digital formats degrade quickly to the point of unplayability

Posted Feb 15, 2016 14:33 UTC (Mon) by tzafrir (subscriber, #11501) [Link]

Furthermore, we have many legacy formats. For many of them there are mostly just converters to other formats. WAV is likely to stay around long enough (due to its simplicity) and there are plenty of converters from arcane formats to WAV (and vice versa).

JPEG

Posted Feb 15, 2016 21:19 UTC (Mon) by tialaramex (subscriber, #21167) [Link] (2 responses)

JPEG is an interesting example; the file format on disk that we all recognise is JFIF, because the Joint Photographic Experts Group (and the standard named for them) didn't consider storing the compressed images in a computer file to be a key element of the problem and so the brief mention of how to do this in the standard itself is entirely inadequate.

When JPEG compression became available most early attempts to store it on disk used TIFF. The resulting mess still exists (short summary: Do not try to write JPEG data to a TIFF, you'll regret it) but very quickly a simpler, easy to implement solution was needed and JFIF specifies basically a way to write out a JPEG bitstream to a disk, with a bunch of constraints that mean (if everybody does this) the decoder can get back the intended picture without needing to guess any parameters.

But, if some hobbyist with the JPEG documentation and (somehow) nothing else to work from in 2116 finds a 1995 JPEG file and wants to decode it, the lack of any JFIF documentation isn't likely to mean they fail. Worst case they get the colours wrong which is annoying but hardly a fatal or unrecognisable error unless the one JPEG they've found is of an abstract pattern. The correct transformation may be a bit subtle, if too much documentation has been destroyed / forgotten / isn't available without paying a colossal fee to the World Disney Government they won't even get the colours quite perfect, but that would be more of a tragedy if we, today, were usually seeing the colours exactly as they were intended anyway, which we're not because we don't calibrate our hardware.

JPEG

Posted Feb 17, 2016 15:18 UTC (Wed) by mathstuf (subscriber, #69389) [Link] (1 responses)

> World Disney Government

So is this Mickey Mouse governence taken to its extreme or did you miss a ™ in there?

JPEG

Posted Feb 17, 2016 17:44 UTC (Wed) by gioele (subscriber, #61675) [Link]

> > World Disney Government
>
> So is this Mickey Mouse governence taken to its extreme or did you miss a ™ in there?

Or that was a citation from Garbageland by Juan Abreu. [1]

[1] <http://www.barcelonareview.com/24/s_ent_ja.htm>

Re: All digital formats degrade quickly to the point of unplayability

Posted Feb 15, 2016 15:53 UTC (Mon) by bfields (subscriber, #19510) [Link]

"ext3 is expected to be removed from near-future linux kernels."

Linux kernels are expected to continue to support ext3 (and ext2) indefinitely, with a shared codebase that also supports ext4. All that's being removed is the old code that supports ext3 only: https://lwn.net/Articles/651645/

“a well-recorded vinyl LP”

Posted Feb 14, 2016 15:45 UTC (Sun) by JanC_ (guest, #34940) [Link] (2 responses)

My (mostly unscratched, despite having been played a lot) CDs from the late 1980s still play perfectly.

Things are very different for CD-Rs obviously, although there are huge differences in CD-R archivability between manufacturers (≠ the brand printed on the label!).

“a well-recorded vinyl LP”

Posted Feb 14, 2016 16:17 UTC (Sun) by pizza (subscriber, #46) [Link] (1 responses)

The biggest weakness of CD-Rs (and CDs too) is that the recording surface is exposed -- it's actually the top of the disc.

With CDs this isn't that big of a deal as the surface is reasonably robust against mechanical damage, but CD-Rs tend to use softer/more fragile surfaces.

However, the bigger problem with CD-Rs is that they are dye-based, and this dye degrades when exposed to UV and/or oxygen. If you leave it out in light, it'll degrade faster. If you slightly nick the sealing layer, or it wasn't all that well sealed to begin with, it'll degrade faster.

DVDs (including the recordable variety) are actually far more robust than CDs because their recording surface is sandwiched in the middle of two polycarbonate layers, making them far less susceptible to mechanical damage and problems due to poor sealing. You should still keep 'em in a dark place though. :)

“a well-recorded vinyl LP”

Posted Feb 15, 2016 10:56 UTC (Mon) by tao (subscriber, #17563) [Link]

And if I remember correctly blu-ray are even more robust.

Longevity over technology changes

Posted Feb 13, 2016 17:16 UTC (Sat) by eru (subscriber, #2753) [Link] (10 responses)

One advantage (possibly the only real one) of analogue formats is the playback device is simple, and in some cases could be designed from scratch just by looking at the media. For example, you can play (sort of) LP:s with a sewing needle attached to a small cardboard box, and some kind of turntable that might be constructed from a motor from a toy and others such household objects. Did this as a kid, as an experiment. Not recommended to try this with any records you value, though. I attacked a free promotional single from Reader's Digest.

As another example, some monts ago I scanned some 35mm colour negatives as old as I am (53 years), showing me shortly after I entered this world. The result looked fairly good even without any fixing, and after some minor enhancements they were about as good as new. I suspect that if my granny had had a digital camera, and the memory card had been left lying in a box for decades (like these negatives were) instead of having been diligently recopied and format-converted, the images would have been lost, unless physical prints (DA conversion!) had been made of them.

Longevity over technology changes

Posted Feb 13, 2016 21:07 UTC (Sat) by ibukanov (subscriber, #3942) [Link] (3 responses)

The longevity of analog formats is a direct consequence of their relatively low information density. LP is nowhere near in quality of 16-bit digital sound on CD. I suspect that recording, say, 12-bit audio on a CD the size of LP would result in a media that lasts decades and that could be read with a magnifying glass.

With current information densities and technologies it is just cheaper to use an active maintenance with check-sums and copies than to create a media that survive a century in a shoe box. I doubt that this change in future as even in nature the cells use precisely such active maintenance with check-sums and copies for their DNA.

Longevity over technology changes

Posted Feb 13, 2016 21:26 UTC (Sat) by malor (guest, #2973) [Link] (2 responses)

For some time, they were selling 'archival CDs' that were supposed to last a hundred years. The dyes in regular CD-Rs just don't last that long, sometimes fading within a few years. The archival ones are supposed to be better, but how good they actually are, I dunno.

Pressed CDs should last longer; I've got a CD from the 80s that still reads just fine. But I've archived the whole collection to hard drive long since, along with 20% PAR2 files, just in case. I can generate new, bitperfect CDs for as long as I can still buy media and a recorder.

The problem there is that it needs active management, manually bringing the library forward to newer media. If I ever forget, the data may all be lost.

Interestingly, magnetic tape may be one of the best archival methods easily available to normal people. It's not cheap, and it's not easy to use, but it's extremely robust. Thirty-year lifespan is pretty common for tapes.

Longevity over technology changes

Posted Feb 13, 2016 21:44 UTC (Sat) by ibukanov (subscriber, #3942) [Link] (1 responses)

As I need to manage a family archive with 5TB of photo/video I once looked at tapes especially given that media is much cheaper than hard drives and their longevity makes long-term storage less problematic.

However the price for devices are insane. I even looked for a service where one can upload an archive and get a tape by post until I realized that I would have no idea that the tape actually worked. So I stick to copies on a hard drives stored at different locations that I periodically check-sums.

Longevity over technology changes

Posted Feb 14, 2016 13:16 UTC (Sun) by malor (guest, #2973) [Link]

Yeah, I got curious after posting that, and popped over to Newegg to see what tape drives cost. Looks like the current state of the art is the Ultrium 6, with 2.5TB raw and about 6TB compressed capacity. Even single drives, not autoloaders, are at least $1500.

The tapes are reasonable, at about $40, but wow, the drives are expensive as hell.

Re: could be designed from scratch just by looking at the media

Posted Feb 15, 2016 5:50 UTC (Mon) by ldo (guest, #40946) [Link] (5 responses)

You think you could deduce the existence of the RIAA equalization curve just from looking at the grooves on a vinyl disc?

Re: could be designed from scratch just by looking at the media

Posted Feb 15, 2016 10:24 UTC (Mon) by ms (subscriber, #41272) [Link] (1 responses)

Quite. Or take a cassette tape which had been processed with Dolby B, C or S dehiss. Good luck identifying that from looking at the tape.

Re: could be designed from scratch just by looking at the media

Posted Feb 16, 2016 5:18 UTC (Tue) by eru (subscriber, #2753) [Link]

No, but RIAA or Dolby do not transform the sound beyond recognition. The researcher (maybe a monk in the order of St Leibowitz?) will realize what the undulated grooves or magnetic patterns mean, and manages to turn them into sound. Later he (or his successors) realizes the frequencies are distorted by comparing to known references, like the human voice, and creates a compensating filter.

Re: could be designed from scratch just by looking at the media

Posted Feb 19, 2016 20:27 UTC (Fri) by jmspeex (guest, #51639) [Link] (2 responses)

Yes. The RIAA equalization curve has to be simple enough to be implementable with simple/old analog components. That restricts the domain of possible curves. Based on that, you'd only need to look at the recording of a known signal (e.g. you know what the spectrum of a real violin should look like) and figure it out pretty accurately. Even without that, the human auditory system is pretty good at equalizing by itself, since between different audio systems (speakers, headphones or different models) in different acoustic environments (car, living room, with different building material), the spectrum of the same recording can change by sometimes more than 20 dB.

Re: could be designed from scratch just by looking at the media

Posted Feb 19, 2016 21:25 UTC (Fri) by ldo (guest, #40946) [Link] (1 responses)

Don’t forget that, for stereo discs, the groove moves in two dimensions. How would you deduce the orientation of the axes for the left and right stereo channels?

Re: could be designed from scratch just by looking at the media

Posted Feb 20, 2016 10:45 UTC (Sat) by smurf (subscriber, #17840) [Link]

Easy. You only have four possible combinations. Take a spatially distinct recording (like an audio drama with stage directions) and check which sounds most correct ("exeunt stage left" should have the door banging on the left speaker, center sound should work when you play a mono LP).

Is the vinyl LP an open music format? (Opensource.com)

Posted Feb 16, 2016 16:18 UTC (Tue) by ccchips (subscriber, #3222) [Link]

I have some vinyl records left around here, and I use gramofile to rip them. That program is tricky to use, but I love it!