Digital Needle: Ripping vinyl records with a scanner (2013)
cs.huji.ac.il
cs.huji.ac.il
Edit: mono is radial only, not depth.
Following this animation I'm left thinking that you could do a bit of filtering to better extract the wave(s) from an image, though you'd require a particular setup for the lighting conditions to get the best possible quality out of it. And then back to my first idea: probably a laser-tip would be the best tool here too.
Unfortunately I don't have vinyl records around to test any of this!
You mean horizontal only?
Old Edison disc records encoded single-channel audio in the vertical axis (varying the depth), but regular 78s and single-channel vinyl records encode the audio in the horizontal axis.
"The method has been used to successfully play several recordings for the first time, including an 1860 phonautogram, the oldest known sound recording of a human voice, and the only known recording of Alexander Graham Bell’s voice."
[1] https://www.macfound.org/fellows/class-of-2013/carl-haber
http://recherche.ina.fr/eng/Details-projets/saphir
https://irene.lbl.gov/I actually got to hear one once. Extremely neutral and high-res, but the record has to be super clean because with the laser, you will hear every little speck of dust. Also it doesn't work with colored vinyl, iirc.
edit: they use 5 different lasers to get the data out of the groove, go read the section on their site, it's actually really intricate and cool
> the format has some funky EQ and other things to make a physical needle work better
Yes, in order to fit the high bandwidth music signal into a groove with a fixed width, the bass is EQed down and treble EQed up, since for a constant velocity sine wave, a low frequency wave would take up more physical width on the record than a high frequency wave. Without this EQ, you would only be able to store a small amount of music on a record. When the record is played, your phono preamp applies an inverted EQ curve to restore the signal to its original frequency response.
In the past, there were lots of different vinyl equalization curves, but the RIAA curve has been standard for decades now.
Audiophiles likely know of this, but I'm still surprised.
Clean vinyl records with glue...
https://everyrecordtellsastory.com/2015/05/25/how-to-deep-cl...
If you want to thoroughly clean a vinyl record though, then an ultrasonic cleaner is the way to go (it is expensive though).
Silly putty can also work in a pinch but some of them leave an oily residue.
Spin clean works but the cheap machines can easily break the record and the expensive ones are expensive as fuck and can still result in damage due to user error.
Almost all the noise on a record is high frequency...so by applying a strong treble rolloff most of it just disappears.
Because the music has the exact inverse transform applied it before it's cut to the master, you don't lose any signal, just noise.
It does not work great.
The problem is dirt, which sounds like a minor issue, but it renders the concept almost completely useless.
A stylus pushes dirt out of the groove as the record turns. A laser cannot do so, and it's virtually impossible to fully clean a record.
The only remaining advantage is that a laser turntable does not wear out the disc, and it's an incredibly minor advantage, compared to simply playing a record once on a traditional player, and recording the output.
But I did hear it, and it did work great.
EDIT: and as soon as I submitted I realize film scanners scan for light and maybe use the IR channel for physical scanning. with a record everything is physical.
or maybe the IR channel focuses on a different plane? could they use different wavelengths for distance sensing/separation?
It's trivial to invert in software, and as another response points out, there is indeed a working laser record player - although the record needs to be extensively cleaned before playback or it doesn't work very well.
What people "hear" regarding LPs is generally their biases + their particular analog setup. My setup is going to sound totally different than yours, regardless if it's from a CD or an LP.
There are some cases where an LP had a better -master- than a CD for the same release, but that is due to the publisher/owner of the master tracks and not the actual medium itself.
44.1kHz is all that one needs for -perfect fidelity-, given that humans cannot hear higher than ~20kHz. See: Nyquist-Shannon Sampling Theorem. [0] 16 bits of quantization gives a Signal to Noise Ratio (SNR) of 96dB, and noise dithering increases that [1]. More bits are not necessary unless one is doing music production with that audio and warrants the extra headroom, before mixing back to 44.1kHz/16bit for distribution. [1]
The only thing that higher bits/sampling rates do for a listener is waste CD and drive space, so it is odd to lament 44.1kHz/16bit as being "really poor". Even if your speaker setup can actually output extremely high frequencies, only your dog will be able to hear it...
[0]: https://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampli...
[1] https://web.archive.org/web/20181109203430/https://people.xi...
CD's are certainly good enough for now.
But that is for a pure tone under perfect laboratory conditions. Also, they report that the minimum hearing threshold at those frequencies was above 90 dB. So I don't think any human in this planet can actually hear anything above 20 kHz under normal circumstances, much less for music.
cf musopen. https://musopen.org/music/
They've got some lovely high bitrate and bandwidth recordings there, and what they've caught is room and audience noise. They'd have benefited from more mics and mixing and less "throw bits at it"
No, that doesn't follow. It just means that recording at those bit rates and sampling frequencies isn't the "bottleneck" for the sound in the system.
>But there really is a difference between a phone with a music app playing a 16bit/44.1kHz song on a medium quality headphones and a dedicated player with a good DAC chip playing a 96kHz song on a good quality headphones.
When everything you compare to is better quality then of course it sounds better. But it's not due to the 96kHz vs 44.1kHz. Nyquist-Shannon is mathematical reality.
You have to compare the same high quality headphones, the amp with the same DAC, with the same recording, being played through a high quality filter bringing everything down to 44.1kHz/16bit.
I agree. If there is a methodologically sound study showing that people can actually distinguish (>50%) between 44.1kHz/16bit and 192kHz/24bit on the same hardware [at exactly the same dB SPL], then I will be really surprised. It's not physically possible to hear those differences, as far as I understand it.
https://xiph.org/video/vid2.shtml
(part 1 at https://xiph.org/video/vid1.shtml is also very well worth watching.)
Most likely, CDs just don't get mastered right. The music is compressed and so even though the format is better, the music is worse. Vinyl is sold to people who care about sound, in their living rooms so they get better mastering. Digital is for people out and about, in your car you need compression even though it destroyes the music you can't even hear 3/4ths of the sound without it. This isn't the fault of the format, but the result remains, digital music is objectively worse.
Vinyl is able to get a slightly better S/N ratio in perfect conditions. So if the record is new it might be better (every time you play the record you wear it just a bit). Odds are it isn't, as almost nothing has enough dynamic range for this to matter.
They are used to worse sound and prefer it. CDs are perfect - vinyl often introduces imperfections that people like to hear. By objective measurements the music isn't perfect, but they like that imperfection.
Most likely it is the first. It is by far the most significant difference.
No. Compact Discs have an SNR of 96dB. Vinyl typically has an SNR of 60-70dB. The physics of these mediums aside, CDs are a newer technology... Why would we have gone for inferior distribution mediums?
"Most likely, CDs just don't get mastered right. The music is compressed and so even though the format is better, the music is worse"
Compression like that has more to do with radio play + the genre of music. Thankfully, the loudness wars are mostly over, but generally speaking the actual distributed master would've been the same.
"This isn't the fault of the format, but the result remains, digital music is objectively worse."
Please provide such objective citations. Mixing and mastering engineers do their best to make music that sounds good on a variety of audio systems. When I worked in a studio, we would mix on $8k genelec monitors, some cheaper monitors, and also test the mix in our cars and on our phones. Part of the art and difficulties in that was producing a mix that sounded good on great to average systems, and passable/decent on like mono phone speakers. Music and music hardware has never been more accessible than today, so I find it hard to believe that everyone a few decades ago had such amazing systems compared to today.
"Vinyl is sold to people who care about sound"
Citation needed. There are people who just want vinyl because it looks cool (especially the colored limited releases), and they go play it on a $50 Victrola all-in-one. There are CD enthusiasts who have systems worth > $200k.
"CDs are perfect - vinyl often introduces imperfections that people like to hear."
This is totally subjective. A CD is a bit-for-bit perfect representation of what the author wanted to distribute. The actual things like "warmth" that people talk about depend much more on the pre-amplifier, amplifier, and speakers.
There is nothing wrong with liking LPs (I collect them myself), but they are in no way, shape, or form better than the bit-for-bit perfect distribution of CDs (that also don't degrade every time you play one).
Yes! When i was producing an album a year, i'd write in monitor headphones, master on bookshelf monitors + a smallish subwoofer, then computer speakers, my car stereo, and a friend's street comp A car stereo. It's the only way to be sure that my ears in the "studio" weren't being biased by the overall quality of the sound damping and accoustics of the room.
And objectively, i know that CDs are "pure". I also really can't hear the difference between a mastered 96khz 24 bit raw audio file, FLAC, or ~300ABR Mp3. And before it's mentioned, i have much better hearing than anyone i know, even if i have lost the ability to distinguish words when someone talks to me off-axis. the other night i heard my nearest neighbor playing music, and no one else could, here. Neighbor is like 1/4 mile away, from inside their house to inside my house. I brought my tascam to the back door and recorded[0]. Sure enough, the recording has a high noise floor but you can see the bass hits.
I still prefer live electronica shows where the DJ has pennies taped to their stylus, but i watch DJs on twitch using laptops and DJ-I digital turntables. All this to say, i've never made fun of someone spending thousands on pre-amps, amps, speakers, decks, whatever. If someone likes the way a Marantz sounds, or a Pioneer, or JBL or klipsch speakers - it doesn't affect me at all.
We're not using skippy CD players and realaudio/56kbit mp3s anymore!
An MP3 file at around 300kbps should be pretty much indistinguishable from FLAC for a human, so that's not surprising at all :)
The reason why I keep my music library as FLAC is not because it sounds better - it's because it's a perfect lossless copy of the original source that can be encoded to something lossy should a need arise. If I had to make 128kbps Opus files for some mobile use case, I'd rather do that from FLACs than MP3s. It also avoids double lossy compression when using things like A2DP.
Higher data density, easier to manufacture, portable.
While we often phrase it as "vinyl vs CD" it was more a "tape vs CD". And CD is arguably the better technology there.
In https://www.soundguys.com/vinyl-better-than-streaming-20654/ - there is a chart of "Music sales by media" and you will see that CD - it replaced cassette tapes. Another view of similar data - https://blogs.sas.com/content/graphicallyspeaking/2019/11/11...
And even with that, "the best technology wins" isn't always the case. One could make a reasonable argument laser disks were better than video tapes and that Beta was better than VHS.
There are forces beyond the pure technology comparison at work in market dominance.
you are confusing typical with what it is capable of in lab like conditions. I agree in real use CDs have a better SNR.
> Thankfully, the loudness wars are mostly over, but generally speaking the actual distributed master would've been the same.
The loudness wars being over helps a lot. However vinyl often is mastered different. Just because you didn't do that doesn't mean others are not. For the few who bother with vinyl of course.
> they are in no way, shape, or form better than the bit-for-bit perfect distribution of CDs
I didn't mean to imply that. I only stated that in ideal conditions they can be better, and that often they get a better music going in. The first of course assumes always fails when put into the real world. The second is very much YMMV, but happens just often enough to be aware of.
The data has to be processed very quickly. I believe you should think in terms of 0.3m groove per second.
If you put a camera above the groove you also need something like a stepper motor to control the arm.
But I believe the biggest problem is dust. A needle will move small particles out of the way, but a camera does not.
This is also why laser turntables need 'clipping' filters. Dust will create all kinds of noise you don't want to hear.
How about designing an arm that used more than one laser 'needle'?
Every other record plays fine, except some new records play slow and/or some carve out a thin hair of lacquer/enamel the first time. This always makes me think some sort of damage is happening from a misaligned cartridge. But, other good quality and used records don't do this and work fine.
It's an AT microline cartridge.
If records are playing slow make sure if your table is belt driven your belts are still in good condition. Also make sure your tonearm height and anti-skate are properly adjusted.
Cleaning your records before you play them (even brand new records) is important.
I always clean them with a Zerostat and Discwasher. And have a Japanese stylus cleaner too (the stuff that looks like a breast implant for mice porn stars.)
The local record store has the same record, maybe they'd compare them?
This also reminds me of the idea to use paper for archival. This is based on two factors: that modern office paper is extremely durable (and is likely to outlast most of the other physical media out there), and that scanners are (and have been for years now) extremely high-resolution; so, you can simply print the bits (say, of a zip file) as dots, and then you could use a scanner to read the file back!
https://archive.org/details/stereo-review-presents-stereo-bu...
Wayback Machine does indeed has archives of the page from 2013 onwards, but while googling "Vinyl-ROM" (inspired by comments in this thread) I found a reference to this exact huji.ac.il page in a comment thread on Halfbakery dating back to 2003:
https://www.halfbakery.com/idea/Vinyl-ROM
In fact the parent URL does have WayBack archives going back to the early 2000s, which include this content. dang: I nominate the date for this item should be listed as 2003.
http://web.archive.org/web/20030221050553/http://www.cs.huji...
https://twitter.com/internetarchive/status/13864235128107212...
And here's all the discussion from a day ago:
PVC grains are typically 100s of um in size, but the monomers themselves are ~10nm. Note that with 120dB claimed dynamic range this would be 37nm of deflection (at 45deg or 25nm lateral) for full resolution. That is close (3x off) to molecular Atomic Force Microscopy.
Of course what is encoded is actually force/velocity which in turn is inductively coupled out, but it is no surprise that simple optical scanning gives you mediocre results. Even 10x lower dynamic range requires 250nm optical resolution. It's doable, but not easy.
Here's a good place to look for vinyl LP info:
Nice experiment.
previous discussion: https://news.ycombinator.com/item?id=2098115