Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Vinyl -> 24/96: Rip once, rip right, never rip again (Read 44207 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Vinyl -> 24/96: Rip once, rip right, never rip again

Now that I'm half-way through ripping my 1600 CDs to disk, my thoughts are turning to the next big project: ripping my vinyl.  In addition to ripping stuff that I don't have on CD, I'm expecting that my vinyl rips will often sound better than my CD rips, so I'm planning to rerip some of my very favorite stuff, like the Beatles.

Ideally, I'd like all my LPs to be available digitally, but that isn't going to happen in this lifetime.  One thing's for sure: once I rip an album, I never want to do it again.

I have a very high-end turntable/cartridge/equalization stage, and I want to archive the LPs with the maximum audio quality that they have to offer.  I basically don't care about the price of disk storage: it's cheap compared to the cost of my time to rip the vinyl.

First off, my impression is that the sonic gain from going beyond 24/96 is unimportant.  Any thoughts on this? 

Second, I'm going to wait until 24/96 codecs are available (at least), even if that's several years. I'm guessing the MPEG4 standard will be the real-world basis for 24/96 support.

So my question is, when will products be available to support ripping vinyl at a resolution of 24/96 and playing 24/96 rips?

My wild guess is that I will be able to start ripping in about a year using boutique products like foobar2000. More general adoption of 24/96 by products like Squeezebox won't be for 2-4 years.

What do your crystal balls say?

Rob

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #1 –
Quote from: RobH on
First off, my impression is that the sonic gain from going beyond 24/96 is unimportant.  Any thoughts on this?
As far as can be determined by studies - and by most of the personal experiences on this forum - yes. In fact, such experience also shows that 16/44.1 is probably going to be just as good as 24/96 for vinyl. (Some people in particular believe that there is nothing whatsoever to be gained from recording over 16/44, and they have very good evidence to support it.)

That said, I usually record at 24/96 simply because I can. I give up about 5db of SNR by doing so, and it takes 1-2 more minutes of time to postprocess due to the added computational load.
Quote
Second, I'm going to wait until 24/96 codecs are available (at least), even if that's several years. I'm guessing the MPEG4 standard will be the real-world basis for 24/96 support.
Most codecs can do 24/96 out of the box, although developers do not often put as much effort into tuning those sample rates as they do for 44khz.

However, if you're that concerned about audio quality, you really should be sticking with lossless codecs, and those handle every sample rate equally well.
Quote
So my question is, when will products be available to support ripping vinyl at a resolution of 24/96 and playing 24/96 rips?
I've been doing just that for over a year; the software/hardware has been available for perhaps 5 years now.
Quote
My wild guess is that I will be able to start ripping in about a year using boutique products like foobar2000. More general adoption of 24/96 by products like Squeezebox won't be for 2-4 years.
foobar doesn't do "ripping" of LPs. You'll be wanting some semi-pro audio editing software, like Audacity, Adobe Audition, MAGIX, etc. foobar or Squeezebox can play 24/96 just fine.

Everything you're asking for is pretty old hat. I suspect that the big constraining factor for you, as you point out, is time. How much postprocessing do you want to do to the rips - do you want to normalize them, remove pops/ticks, cut the cue in/out time on the beginning and end of the recording? Have you cleaned your records? I tend to do all of those things and the resulting time requirements are pretty big.

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #2 –
Axon,

You're right: time is the issue for me.  As to postprocessing, I want the basics.  Track info and maybe cutting the cue-in, cue-out time. I agree 105% on the importance of record cleaning, it's huge.  (I've researched this: I clean my records with a Nitty Gritty using distilled water from the vending machines in super markets (super-pure!),  industrial-grade isopropyl alcohol, and a surfactant whose name I forget (Record Genii?).  All of this stored in glass jugs. Works great!)

Which codec do you use for 24/96 recording?  FLAC? I will absolutely be sticking with lossless compression. 

A quick search of the Squeezebox support site indicates that they downsample 24/96 feeds to 24/48, but that's OK for now.  24/96 feeds are recognized, which is the main thing. I'm in it for the long run!

Do you have a recommendation for audio editing software among the choices you mentioned?

Thanks for your helpful reply,
Rob

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #3 –
Off the top of my head, FLAC, wavpack and WMA Lossless can handle 24/96

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #4 –
I generally use FLAC for lossless compression, but I've switched to/from APE from time to time. A lot of it just depends on personal preference. FLAC -5 works just fine, -8 works slightly better and much slower.

About the only "gotcha" with lossless codecs is that very few of them (only LA, to be exact) support floating point input. And some audio editors will want to record in floating point instead of 24- or 32-bit fixed point. But 32-bit floats only have 24-bit mantissas anyway, so you lose no real information by downconverting from float to 24-bit.

I use KRISTAL Audio Engine for my recording, but I'm in a particularly weird situation. The ASIO drivers on my RME card seem to be buggy at 24 bit recording, and KRISTAL is the only free editor that seems to give decent results at that format. You're likely to get much better results than me. To evaluate an editor, just record some input noise on your sound card, save the signal to disk, load it back up and see that it "makes sense". Are tehre discontinuities in the waveform, where it "jumps" far more than it should, or it's periodic? Are there any zeroed out sections of the recording?

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #5 –
Quote from: Axon on
About the only "gotcha" with lossless codecs is that very few of them (only LA, to be exact) support floating point input.

WavPack also supports floating point data.

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #6 –
Wavpack supports 32 bit float as well.

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #7 –
Quote from: RobH on
So my question is, when will products be available to support ripping vinyl at a resolution of 24/96 and playing 24/96 rips?

My wild guess is that I will be able to start ripping in about a year using boutique products like foobar2000. More general adoption of 24/96 by products like Squeezebox won't be for 2-4 years.

What do your crystal balls say?


Recording at 24 bit is a good idea.  It'll save you a lot of trouble in adjusting levels.  For playback, it really doesn't matter.  If you use replaygain with your files, you basically lose nothing from 16 bit playback.

96k doesn't get you anything, but it doesn't cost you anything either, aside from disk space.  I'd probably just do 48k for efficiency.

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #8 –
I digitize analog (SACDs in this case) at 88/24 rather than 96, only because 88-->44 would seem a simpler transform, and uses somewhat less space/processing time than 96.  Am I being superstitious?

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #9 –
Resampling to a even factor is much faster according to my tests in CoolEdit (e.g 88.2 to 44.1 instead of 96 to 44.1). I've seen quite a few opinions as to what's involved, enough to be reasonably sure that few people actually know and those who do probably don't bother with such trivia. I haven't been able to tell any difference in the way the results sounds.

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #10 –
Quote from: krabapple on
I digitize analog (SACDs in this case) at 88/24 rather than 96, only because 88-->44 would seem a simpler transform, and uses somewhat less space/processing time than 96.  Am I being superstitious?


88 to 44 is no simplier then 96 to 48.  If you're planning on using 44, then 88 makes sense, but given that everything supports 48k these days, and a lot of cards don't support 44.1, I don't see the need to worry about 44.1 for an analog source.

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #11 –
Quote from: RobH on
Now that I'm half-way through ripping my 1600 CDs to disk, my thoughts are turning to the next big project: ripping my vinyl.  In addition to ripping stuff that I don't have on CD, I'm expecting that my vinyl rips will often sound better than my CD rips, so I'm planning to rerip some of my very favorite stuff, like the Beatles.


I am constantly amazed that there are people who still believe that LP's sound better than CD's.  They sound different.  They are less accurate as analog introduces its own distortions which some perceive as "better"  Why  eludes me.   
Nov schmoz kapop.

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #12 –
Various factors are involved (such as 'I already have the LPs so I don't want to spend 20 bucks * 500 albums (or whatever) for another copy of each'), but the fact is that music originally released on LP is frequently re-mastered to obnoxious standards for CD. Sure, the media is better, but the content is significantly poorer. And yes, I know some CDs are essentially impossible to tell from the original LPs (aside from the lack of clicks and pops and surface noise), but many are quite different.

I know someone who is doing his rather battered LP collection right now for exactly the reason that he never liked the CD versions, although he bought most of the CDs over the years because his LPs were wearing out and the CDs were much more convenient. He has no illusions that vinyl is better, its all about the music. For me it is that I enjoy discovering old music, I have some fun cleaning the recordings, and I get most albums for fifty cents.
 

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #13 –
yeah... the Nyquist theorem is enough. We don't need LP's, SACD's, or DVD-Audio, and ripping anything higher than the standards it's only placebo. A perfectly mastered CD sounds just fine.

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #14 –
My main audio source is a PC with 24-bit and up to 96 kHz capability, but if I decide to record my vinyls I guess I would like to burn some of them on audio CDs too. So perhaps 88.2/44.1 kHz would be a good sample rate choice.

However, I don't quite understand what benefit a 96 or 88.2 kHz sample rate offers over 48 or 44.1 kHz. I understand that it is good to record at 24-bit instead of 16-bit if the files are going to be post processed, but does a higher sample rate offer any better audio quality as an end result if the target audience does not consist of bats and especially if the source does not contain frequencies higher than about 20 kHz? Does post processing introduce some distortion in the audible frequency range that is lower if the sample rate is higher?

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #15 –
Quote from: Alex B on
However, I don't quite understand what benefit a 96 or 88.2 kHz sample rate offers over 48 or 44.1 kHz. I understand that it is good to record at 24-bit instead of 16-bit if the files are going to be post processed, but does a higher sample rate offer any better audio quality as an end result if the target audience does not consist of bats and especially if the source does not contain frequencies higher than about 20 kHz?


As I said above, theres no advantage.

Quote from: Alex B on
Does post processing introduce some distortion in the audible frequency range that is lower if the sample rate is higher?


None that I know of.

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #16 –
Yes, I am convinced that if the recordings are going to be used as they are 44.1 kHz is enough for human hearing. I just wonder where the trend to record at a higher sample rate and down sample after processing came from (I mean, if the purpose is not to produce new hi-res end formats for audio industry).

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #17 –
Quote from: Alex B on
My main audio source is a PC with 24-bit and up to 96 kHz capability, but if I decide to record my vinyls I guess I would like to burn some of them on audio CDs too. So perhaps 88.2/44.1 kHz would be a good sample rate choice.

However, I don't quite understand what benefit a 96 or 88.2 kHz sample rate offers over 48 or 44.1 kHz. I understand that it is good to record at 24-bit instead of 16-bit if the files are going to be post processed, but does a higher sample rate offer any better audio quality as an end result if the target audience does not consist of bats and especially if the source does not contain frequencies higher than about 20 kHz? Does post processing introduce some distortion in the audible frequency range that is lower if the sample rate is higher?


On a perfectly clean, brand-new vinyl, I've noticed substantial content up to about 24kHz (only for things like cymbal crashes, though - most other stuff seems to drop off after 20kHz, although I've noticed some harmonics/distortion up to about 30kHz on a few 180-gram jazz pressings). Although this is obviously beyond the range of human hearing, if one really wanted to capture the full range of frequencies on a new LP, they would only need 48kHz to do so. 88 or 96Khz is just not necessary, as you've said. That's why I record at 48kHz, but I use 24-bits of resolution so I have more than enough headroom for manipulating/editing the recordings later. If you're going to CD (I don't), 44.1/24 would be better, as you'll avoid a sampling rate conversion.

I agree with one of the earlier posters - collecting vinyl isn't done for the sound quality; it's the fun of finding music for mere pocket change. The big album art and ritualistic aspects of playing back a vinyl LP (carefully holding it, cleaning it, dropping the needle oh-so-gently in the groove) make collecting and playing records a different experience than CD, but not a better or more musically accurate one. Personally, I don't understand how audiophiles can blissfully ignore the constant groove rubbing, rumble, and surface noise that I hear even on new, clean records played back on a good Music Hall turntable. CD's are definitely a more accurate medium!

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #18 –
Quote from: Cygnus X1 on
Quote from: Alex B on

My main audio source is a PC with 24-bit and up to 96 kHz capability, but if I decide to record my vinyls I guess I would like to burn some of them on audio CDs too. So perhaps 88.2/44.1 kHz would be a good sample rate choice.

However, I don't quite understand what benefit a 96 or 88.2 kHz sample rate offers over 48 or 44.1 kHz. I understand that it is good to record at 24-bit instead of 16-bit if the files are going to be post processed, but does a higher sample rate offer any better audio quality as an end result if the target audience does not consist of bats and especially if the source does not contain frequencies higher than about 20 kHz? Does post processing introduce some distortion in the audible frequency range that is lower if the sample rate is higher?


On a perfectly clean, brand-new vinyl, I've noticed substantial content up to about 24kHz (only for things like cymbal crashes, though - most other stuff seems to drop off after 20kHz, although I've noticed some harmonics/distortion up to about 30kHz on a few 180-gram jazz pressings). Although this is obviously beyond the range of human hearing, if one really wanted to capture the full range of frequencies on a new LP, they would only need 48kHz to do so.


Thats probably just noise.  Since LP output isn't hard limited like a DAC, you get lots of energy at higher frequencies, but that doesn't imply that theres any information there.  A lot of it will just be mechanical vibration in the system, and nonlinear distortion from sounds in the higher part of the audiable spectrum (which may or may not seem correlated with higher but audiable frequency sound).

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #19 –
All this has been covered quite a few times in this forum, as well as dozens of others. I see no point in going into details, but will remind people of a couple of the high points.

While little music on LPs goes above the CD frequency limit (and its debatable that any of that is audible), the media itself is capable of going much higher. I can record the 30Hz to 30kHz sweep tone that is cut into the Cardas test/burn-in LP with no difficulty. Using a 96kHz sample rate, I record quite clear harmonics of that sweep tone all the way to the 48kHz limit. Are those harmonics in the vinyl or an artifice of my playback system?

It probably doesn't matter in any practical sense, as in being audible except under extreme conditions, but probably every soundcard on the market lets through readily measured alias images when recording at 44.1 (if the input contains frequencies beyond the Nyquist limit). I've posted detailed information and documentation here in the past.

If one records at 88.2 or higher, and then resamples to 44.1, there is (1) less energy in the alias images to begin with and (2) very little energy indeed by the time it reaches back to audible frequencies. Therefore, recording at 44.1 will contain this distortion (even if only with HF harmonic distortion as the source), and recording at a higher sample rate will eliminate most of it. A fair number of professionals who record live at higher sample rates have made statements to the effect that its all about alias distortion, not capturing higher frequency content.

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #20 –
Quote from: AndyH-ha on
If one records at 88.2 or higher, and then resamples to 44.1, there is (1) less energy in the alias images to begin with and (2) very little energy indeed by the time it reaches back to audible frequencies. Therefore, recording at 44.1 will contain this distortion (even if only with HF harmonic distortion as the source), and recording at a higher sample rate will eliminate most of it. A fair number of professionals who record live at higher sample rates have made statements to the effect that its all about alias distortion, not capturing higher frequency content.


I'd strongly agree. For proper sampling you simply must not have present any frequencies above the Nyquist limit in the analogue signal that is passed to the sampler. This requires an analogue filter of some kind, and if you wish to preserve audible frequencies only about 10% below the Nyquist limit it's very difficult to filter well.

Otherwise, if you sample at 44.1 kSa/s (kilosamples per second) your Nyquist limit being 44.1/2 = 22.05 kHz, a frequency above 22.05 kHz would be mirrored to a frequency below 22.05 kHz. Just imagine a 30 kHz sine wave that wasn't filtered being sampled at 44.1 kSa/s. At 7.95 kHz above the Nyquist limit it can be shown that it is indistinguishable from a sinusoid 7.95 kHz below the Nyquist limit. When playing back the waveform, the reconstruction filter has to assume it's below 22.05 kHz, so it comes out as 14.1 kHz (a mirrored alias tone), which is an audible frequency.

The filter in a soundcard capable of 88.2 or 96.0 kSa/s is likely to be a good filter with fair response well above 22.05 kHz but very little at 40 kHz or more. I don't think it's likely that many or any soundcards would implement a different pre-sampling electronic filter for each sampling rate, so only the higher sampling rates are likely to be sampled properly according to Nyquist's theorem.

You want the best quality and to do it only once, so respect Nyquist diligently and allow ample headroom to avoid clipping by using 24-bit sampling and not worrying if the peaks in your waveform are even only 1/4 or 1/2 of full-scale. You can use Replygain's Album Gain mode to achieve normal loudness for playback after you've made your recording.

When you later downsample to 44.1 kHz, a brickwall pre-filter must be applied before generating the samples at the lower rate, as part of the algorithm. SSRC and the equivalent downsampler in Foobar2000 will certainly do this, as will good quality audio editing software.

Foobar2000 has excellent DSP resampler and dithering algorithms for optimum conversion of your lossless high-sampling rate, high bit-depth files to 16-bit, 44.1 kSa/s for burning to CD. With vinyl sources, I'd say that "no noise shaping" dither is ideal when converting to 16-bit.
Dynamic – the artist formerly known as DickD

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #21 –
Even with a (good) vinyl source, noise shaping is necessary to avoid adding to the audible noise background.

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #22 –
Quote from: AndyH-ha on
Even with a (good) vinyl source, noise shaping is necessary to avoid adding to the audible noise background.

I don't think so. Vinyl background noise is much higher than 16-bit background noise, so you don't need to use any kind of dithering when converting it to 16 bit. Anyway, in case you still want to use dither, use flat dither, it won't do anything to the source, it's faster and is, as its name implies, spectrally flat. ATH Noise shaping will probably add some HF noise.

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #23 –
Its true that dither isn't often really needed. The primary results of dither/no dither aren't audible even in the purest live recordings until the levels get very low. However, with a good vinyl transfer and judicious NR, fades can get down to about the level where it starts to make a difference.

If you are using dither I guess you can argue for whtever appeals to you. I like haveing it mostly above 18kHz where I'm never going to hear it.

Vinyl -> 24/96: Rip once, rip right, never rip again

Reply #24 –
Quote from: Mike Giacomelli on
Quote from: krabapple on

I digitize analog (SACDs in this case) at 88/24 rather than 96, only because 88-->44 would seem a simpler transform, and uses somewhat less space/processing time than 96.  Am I being superstitious?


88 to 44 is no simplier then 96 to 48.  If you're planning on using 44, then 88 makes sense, but given that everything supports 48k these days, and a lot of cards don't support 44.1, I don't see the need to worry about 44.1 for an analog source.


I do resample to 44.1 as a last step, as much out of habit as out of the rare instance where I actually burn the .wavs back to CDR for playback in a CD or DVD player.  FWIW, I rarely digitize LPs, and when I do, I don't use 88.2, I use plain old 44.1. 



Quote from: Bourne on
yeah... the Nyquist theorem is enough. We don't need LP's, SACD's, or DVD-Audio, and ripping anything higher than the standards it's only placebo. A perfectly mastered CD sounds just fine.



Correct. 

The 'need' for higher sampling rates and bitdepths *for home playback delivery formats* has never been established in fact.  The utility of 24 bits for recording / production is not in doubt.  The Nyquist theorem has been extremely well established to be 'enough' if by that you mean, enough to determine the parameters for sampling needed to capture whatever desired bandwith.