Vinyl preference would truly be off topic. There are so many reasons for vinyl preference over CD, and the reverse preference. Nothing to do with capture for archiving though.
Absolutely true. Presumably the OP would be satisfied with a solution which archives LP as digital files that he cannot audibly distinguish from the original direct LP sound in his system. What does it matter if he archives at 192/24 rather than 48/16? It probably won't be significantly more expensive.
SSD storage is cheap, hard drives are even cheaper and prices will continue to fall. This thread has been hijacked with very few genuine attempts to answer the OP queries. It has become a farcical point scoring exercise.
In theory yes, but I read somewhere that in practise, as counter intuitive as it is, many AD converter circuits work in a way that the frequency response usually gets a little better when sample rate is increased but at the same time the timing may get worse. So, with higher clock rate the sample intervals variate more but their values are more accurate. Again, I'm only repeating what I have read so I'm not willing to argue about this.
I give you an example.
Sub 24kHz wave digitized with 96kHz@infinitebit:
0.997 0.998 0.999 1.000 1.001 1.002 1.003 1.004
Same wave digitized with 48kHz@infinitebit:
0.997 0.999 1.001 1.003
Same wave digitized with 48kHz@16bit:
0011
Same wave digitized with 96kHz@16bit:
00011111
Every time we change a setting, we get different representation of the wave. Because of Nyquist, we can reproduce original wave from 48kHz@infinitebit data so we don't need 96kHz.
But the question was: do we get better wave approximation from 96kHz@16bit data than 48kHz@16bit data? My guess is still "YES", because 96kHz data has more information (the red bit), so it is closer to the original. You say "NO" so where does my logic go wrong?
You owe it to yourself to read the right stuff. None of the above is true.
I'm sorry, I must ignore the above as it makes no sense at all.
While I agree with 2, it contradicts 1.
Your logic goes wrong because you're ignoring what goes on during reconstruction. If you start with a perfect sine wave (convenient for analysis), after reconstruction you end up with the same perfect sine wave without regard to sampling frequency (except for the different Nyquist upper limit) and without regard to bit depth. Only one thing changes when you change bit depth: noise. Only one thing changes when you change sampling frequency: the upper maximum frequency of the system.
Reconstruction is not smoothing of a stair-stepped digital signal, reconstruction is connecting the dots. Quantization produces a string of numerically represented dots that exist for that tiny moment in time. The reconstruction filter connects these dots and removes all traces of the sampling frequency. The result is a perfect sine wave.
This video will help.
SSDs have an unpredictable failure rate. I personally have changed 4 dead SSD drives within the calendar 2019, which were all less than 2 years old.
No archive project should be undertaken without a solid plan for backup. If photographers don't consider their digital images "save" unless they exist in 3 separate devices and locations, there's no reason an audio archive wouldn't demand at least a plan that approaches that.
HDDs are not long-term storage devices. Data will eventually self-corrupt with the drive just sitting on a shelf un-powered. SSDs, at least in my world, have had a failure rate that was disappointing at best. So your only option is to save your archived audio files on more than one device, probably 2 or 3. How's that storage cost looking now? 6 times the storage times 3? And you have to manage all that data too.
The only thing that seems to make sense is to digitize at the maximum sampling frequency that provides an audible benefit. That would be 44.1kHz. But since we're all OCD, let's go up a bit beyond the limits of human hearing, and sample at 48khz. And, since records noise floor is higher than 16 bits at all points of the spectrum, 16/48 is the sweet spot, for performance, storage cost, and data manageability.
I'm not personally looking for lowering SSD costs, I'm looking for higher reliability over time and capacity over 4tB that doesn't break the bank.
I recently picked up a TI PCM4222 eval board to do the ADC duties. This is a SD ADC so natively spits out DSD If I have read the spec sheet properly and downsamples to PCM on chip. So bascially the rate to sample at is fixed and really the question is what resolution to process at and what to store at. My miniDSP does 24/96 so for me its a no brainer. I claim no audible advantages.
You go to buy gas. One station has it for $3/gal, another up the street has it for $18/gal., and claims their gas is better because your engine runs smoother. Your engine already runs smooth, you can't perceive it to run any smoother with the expensive gas. There is no scientific evidence of the expensive gas advantages. Which gas do you buy?
Your WAVCar needs 5 gallons to go 100 miles, but your FLACCar only needs 3. Does that change what gas you buy?
Who stores uncompressed audio? Professionals, and OCD audiophiles. FLAC is no friend to the Apple biosphere, where ALC is their friend. You know, all them unpopular iPhones and such. You can get apps that work with FLAC, but you can't get the basic biosphere to cooperate.
Regardless, still no reason to by the expensive gas, unless spending money needlessly makes you feel great. And for some, that's what counts.
But here's another question...What if the $18/gal gas harmed your car? What if it actually made the car perform worse? Welcome to the world of high sampling rates and ultrasonic content! The truth is, you don't actually want ultrasonic material hitting your transducers or analog amplifiers because at frequency extremes both become progressively nonlinear. That means they cause distortion more easily. Distortion creates new unwanted signals in your "pure" music, some of which can (and have been proven to) become quite audible, and not in a good way.
Again, 16/48 is the sweet spot, no need to do more, no benefit, and a strong potential for several detriments. And for CD rips, 16/44.1 is the sweet spot. Never resample if you don't have to.
If anyone is curios, and at this point, I doubt that as we are just volleying, you can take a look at this article.
I'd buy the gas for 3$/gal... Only a complete fool would waste 18$/gal on foolish claims that can not logically be worth the extra expense so I would never know if it "damaged" my car, not using it...
IDGAF about Apple or the people who are sheeple enough to drink their kool-aid since I've run Linux or Android for 10 years... ALAC also exists for those stuck in the "walled garden".
The fact that a FREE LOSSLESS AUDIO CODEC is somehow hard to implement on Apple speaks volumes about them.
I agree about not resampling of course.
At least with 24 bit you get another 24 db of dynamic range... Especially nice when you are manipulating the audio many times in the digital domain.
24/96 is MY sweet spot though, although 24/192 only results in an average 20MB increase in a 5 minute song as FLAC (or ALAC if you're on iCrap). More and more people are using DSD or even PCM at 32 bit 768kbit. THIS I see very much past the point of diminishing returns. In your own post: "An engineer also requires more than 16 bits during mixing and mastering. Modern work flows may involve literally thousands of effects and operations. The quantization noise and noise floor of a 16 bit sample may be undetectable during playback, but multiplying that noise by a few thousand times eventually becomes noticeable. 24 bits keeps the accumulated noise at a very low level. Once the music is ready to distribute, there's no reason to keep more than 16 bits." Except, if you have a higher quality file, why downgrade it? HDD space? But a bigger drive, cheap batârd
When I sample stereo records, I record them at 24/96, run some "clip/pop elimator" type plugins, maybe hand draw a bad scratch out of the recording, normalise it, noise reduction depending on the noise etc. When I sample mono records though, it's as simple as recording in stereo, and then going mono since almost all clicks/pops are common mode. When I listen to music on the go, in the car or on the subway, I export and downsample it to OPUS (90% smaller files, hardly noticeable artifacts).
My real point is, if you use lossless compression, there is ZERO reason not to keep high quality files based on their size. And even without compression, the average "CD" is under 700MB. You can buy an 8 TB hdd for about 200 dollars if you look hard enough. That's 39 cents per 1024MB. If this cost is excessive, I suggest sticking to radio.
IDGAF about Apple or the people who are sheeple enough to drink their kool-aid since I've run Linux or Android for 10 years... ALAC also exists for those stuck in the "walled garden".
The fact that a FREE LOSSLESS AUDIO CODEC is somehow hard to implement on Apple speaks volumes about them.
I agree about not resampling of course.
At least with 24 bit you get another 24 db of dynamic range... Especially nice when you are manipulating the audio many times in the digital domain.
24/96 is MY sweet spot though, although 24/192 only results in an average 20MB increase in a 5 minute song as FLAC (or ALAC if you're on iCrap). More and more people are using DSD or even PCM at 32 bit 768kbit. THIS I see very much past the point of diminishing returns. In your own post: "An engineer also requires more than 16 bits during mixing and mastering. Modern work flows may involve literally thousands of effects and operations. The quantization noise and noise floor of a 16 bit sample may be undetectable during playback, but multiplying that noise by a few thousand times eventually becomes noticeable. 24 bits keeps the accumulated noise at a very low level. Once the music is ready to distribute, there's no reason to keep more than 16 bits." Except, if you have a higher quality file, why downgrade it? HDD space? But a bigger drive, cheap batârd
When I sample stereo records, I record them at 24/96, run some "clip/pop elimator" type plugins, maybe hand draw a bad scratch out of the recording, normalise it, noise reduction depending on the noise etc. When I sample mono records though, it's as simple as recording in stereo, and then going mono since almost all clicks/pops are common mode. When I listen to music on the go, in the car or on the subway, I export and downsample it to OPUS (90% smaller files, hardly noticeable artifacts).
My real point is, if you use lossless compression, there is ZERO reason not to keep high quality files based on their size. And even without compression, the average "CD" is under 700MB. You can buy an 8 TB hdd for about 200 dollars if you look hard enough. That's 39 cents per 1024MB. If this cost is excessive, I suggest sticking to radio.
Last edited:
Completely agreed. But the Apple world exists, and dominates portable devices.
Another fallacy. The 24 bit word has a possible DR of 144dB, with dithering a bit more than that. But there are no ADCs with true 24 bit performance, not even close. That 24 bit data you are slinging around has a true DR of about 20 bits. There are no studios capable of true 24 bit DR either, most are well under 16 bits. You can't reproduce 24 bit DR without dropping well below the threshold of hearing and well above the threshold of pain and permanent hearing damage. Typical listening room has a noise floor at 30dB SPL A, if you put -144dBFS at 30dB SPL, then 0dBFS is at 174dB SPL, not only impossible to achieve, it's deadly. 24 bits makes no sense in any application but the original recording. Doesn't matter in post processing either because all DSP is done in either 32 bit floating point or 64 bit floating point.
Ok.
We should probably not go down the DSD road. Long since been debunked as inferior in many ways to PCM, except, of course, by it's own proponents.
I stand by the quote. It is correct. In production, you benefit from more bit depth. It's the unpredictable nature of recording, a bit more headroom can save the take. But not in release to the public. You don't downgrade the result by releasing it in 16 bit.In your own post: "An engineer also requires more than 16 bits during mixing and mastering. Modern work flows may involve literally thousands of effects and operations. The quantization noise and noise floor of a 16 bit sample may be undetectable during playback, but multiplying that noise by a few thousand times eventually becomes noticeable. 24 bits keeps the accumulated noise at a very low level. Once the music is ready to distribute, there's no reason to keep more than 16 bits." Except, if you have a higher quality file, why downgrade it? HDD space? But a bigger drive, cheap batârd
OK.
All valid viewpoints. None support 24/96, except that its a personal preference. It is an investment in HDD real estate and cost, though obviously for most its not significant. It gains the user absolutely nothing, however, and may actually cause problems (read my last link).
Everyone is of course free to make whatever choices make them feel good. Record at 8/15 or 32/768, or anywhere in between. People make their choices mostly based on emotion without real knowlege, and that's just the way it is.
Yes, I am now giving up. Points made, points ignored.
I use the Furutech ADL GT40 Alpha and have found it cost effective and transparent in my system when used as a line input A/D converter at 192/24. It has –6 dB, and –12 dB gain settings which prove useful. The RIAA input stage has a high quality potentiometer but I use a separate phono stage.
I would suggest that for the purposes of digitizing vinyl the energy and attention should be focussed on the analog portion; the cartridge, stylus, arm, and preamp. In particular the performance of the total system and its ability to precisely track the RIAA record characteristic. Errors there far exceed any that could possibly encountered in the digital part of the system. Any ADC is adequate for precise digitization of vinyl, but the analog side, including the preamp, are where all the errors that are audible lie.
The only ADCs that may not be adequate are found integrated onto logic boards, though even those have gotten really good lately.
The only ADCs that may not be adequate are found integrated onto logic boards, though even those have gotten really good lately.
wow... not one mention of what TT or cart the OP is using. no point digitising music from something that was found in a thrift store or belongs on top of a stacking system.
i have a Schiit Mani and its superb for the money. just take a line out from the Schiit to your comps IN using a phono to mini jack, download Audacity set the sample rate to the highest you can. once you have a track in Audacity transfer it to FLAC. if you record in high bit rate you can always convert it to lower rate if needed.
i have a Schiit Mani and its superb for the money. just take a line out from the Schiit to your comps IN using a phono to mini jack, download Audacity set the sample rate to the highest you can. once you have a track in Audacity transfer it to FLAC. if you record in high bit rate you can always convert it to lower rate if needed.
The OP has been a member of this forum longer than anyone else in this thread. I think it is only reasonable to assume (s)he is an experienced vinyl user. Bringing in LP rigs will likely start another p***ing contest.
Whatever the front end, the target is still the same: capture the LP sound indistinguishably from the original.
