Attached are several clips for vinyl rips, each in (24/48) and (16/44.1) -- I have many more but 7 I filled up my attachment space.
I used Izotope's SRC and MBit+ algorithms to do the conversion, I converted sample rate first and then bit rate.
The rips were done using Thorens Turntable, Lyra Delos cartride, Parasound phono pre, Burl B2 ADC, Lynx AES16 card, Pure Vinyl software.
If anyone has better conversion technology than Izotope I would be interested to hear those samples for my own comparison, as a practical matter I would like to have the best possible bit reduced versions in my library in addition to my 24 bit versions.
If anyone is interested in specific notes on what I'm hearing as the differences in my playback system then let me know.
The sources are:
Dire Straits "Romeo And Juliet"
Prefab Sprout "Faron Young"
Yes "Roundabout"
The National "Bloodbuzz Ohio"
Cat Stevens "Father And Son"
Roxy Music "Running Wild"
John Lennon "Mind Games"
Thanks for checking these out.
It shouldn´t be of any advantage to use iZotope dither here. It is found being good for having a noiseshape that prefers low-frequency noise + high shape above 15kHz and therefore sounds more comfortable. When dithering down a Vinyl rip there has to be so much noise in that you even most likely need NO dither at all, leave alone having an advanztage with high efficient dither.
Edit: removed quote
I gave this Faron Young sample a try and played it on my main system. So far it didn´t impress me. Sounds like some mediocore recording with not to much detail and even lack of fine highs.
I couldn´t find the 16bit version sounding different by simply switching between the 2 but i didn´t try much there.
Will be interesting what the others have to say.
An interesting experiment for Gretschguy to perform would be to convert the 16/44.1 files back to 24/48 and see how they sound. That could rule out problems with the sound card.
An interesting experiment for Gretschguy to perform would be to convert the 16/44.1 files back to 24/48 and see how they sound. That could rule out problems with the sound card.
pdq, that sounds like a really good suggestion -- so you are saying take the 16 bit files and convert them back up to 24 bit and if that once again sounds better to me then this suggests its a bias in how the card DAC is handling 24 as opposed to 16. Seems like a great test. I'm only confused that you mention soundcard, did you mean DAC?
From the website for the DAC, it appears that it does not natively support 44.1 kHz. That means that something else is doing the conversion, and perhaps doing a poor job of it (has happened before).
I gave this Faron Young sample a try and played it on my main system. So far it didn´t impress me. Sounds like some mediocore recording with not to much detail and even lack of fine highs.
I couldn´t find the 16bit version sounding different by simply switching between the 2 but i didn´t try much there.
Will be interesting what the others have to say.
Wombat, thanks for checking it out. I hear plenty of highs on Faron Young, make sure your tone controls are all neutral and no surround processing is set on. You also might want to check out "Romeo and Juliet" if you want to give another one a try.
On Faron Young specifically, take a close listen to the bass line -- do you hear the individual notes of the bass line exactly the same on the two versions. Does one version spell out the bass individual bass notes better? Do you hear all the instruments on the same plane in both versions or does one version have more front to back dimension in which the instruments widen out and take on more definition. Just a couple things to look for on a recording like this one that has a lot of going on at a fast pace. This one measures close to 20dB between LEQ and peak per AudioLeak. You might want to try compressing it down to 12 dB or so if you have compression software and see if that sounds more modern and perhaps to your liking.
From the website for the DAC, it appears that it does not natively support 44.1 kHz. That means that something else is doing the conversion, and perhaps doing a poor job of it (has happened before).
pdq, what DAC are you referring to? Burl is my ADC not my DAC.
Let there be no doubt that any audible differences will be due to resampling problems long before they will be due to deficiencies in bit depth. There is absolutely no empirical data indicating that vinyl will suffer any audible degradation when presented in a delivery format that makes use of 16 bits, if not less.
Yes and no -- I've certainly spent a lot of care and money on my vinyl setup -- my Burl ADC alone probably costs more than most folks whole sound systems. But I would say, in my experience, that vinyl quickly exceeds CD for the right material (I listen to much 70s and 80s vinyl) and is not just providing "good reproduction" but rather "stellar reproduction" that I can't seem to get from my CD playback and my 16 / 44.1 playback is done by Burmester which is a pretty good name in the business. Even with my more modest Denon cartridge before upgrading to my Delos, I felt that my vinyl playback exceeded my CD playback when both were compared on my Burmester audio system.
Have I misunderstood your setup?
Does one version spell out the bass individual bass notes better? Do you hear all the instruments on the same plane in both versions or does one version have more front to back dimension in which the instruments widen out and take on more definition. Just a couple things to look for on a recording like this one that has a lot of going on at a fast pace.
Is it just me or does this sound like the same old placebo-based audiophile subjectivism that upon which this forum deeply frowns?
Let's see some ABX test results before we go traipsing off into pseudo-scientific woo land, m'kay?
Let there be no doubt that any audible differences will be due to resampling problems long before they will be due to deficiencies in bit depth. There is absolutely no empirical data indicating that vinyl will suffer any audible degradation when presented in a delivery format that makes use of 16 bits, if not less.
Greynol, the proper test I think is to have the ADC sample at 16 bit / 44.1 and compare it to a 24 bit / 48 kHz sample which would then remove any of the resampling / conversion bias. My Burl only samples in 24 bit so I can't do that test but I think that's probably the best test actually.
It's readily accepted that best practice is to sample at 24 bits, using the same rate that will be used for playback unless there is a problem with the hardware at that rate or subsequent processing (prior to conversion to the desired delivery format) would benefit from sampling at a different rate from that used for playback.
Does one version spell out the bass individual bass notes better? Do you hear all the instruments on the same plane in both versions or does one version have more front to back dimension in which the instruments widen out and take on more definition. Just a couple things to look for on a recording like this one that has a lot of going on at a fast pace.
Is it just me or does this sound like the same old placebo-based audiophile subjectivism that upon which this forum deeply frowns?
Let's see some ABX test results before we go traipsing off into pseudo-scientific woo land, m'kay?
Greynol, I apologize, I don't mean to bring the forum down a few notches -- I suggested those as things to listen for -- it could very well be that sampling errors and conversion errors are causing that difference -- it could very well be that my DAC does something different with 16 than 24 bits. I thought it would be helpful to provide some context to what I'm hearing, maybe smart folks can pinpoint it and say "aha, this guy is hearing dithering noise.." or whatever...
I can try to setup some kind of AB test myself, sounds like you won't trust my results -- I use an entirely Burmester sound system not a computer card so I would have to sort out a way to do that. I think I can random shuffle the playback on my system and see how I score that way.
To be clear, you do not want me to reference the specifics of what I'm hearing -- even if I freely admit that what I'm hearing could be due to any number of distortions, bias', conversion problems, etc... You want me to just not express it at all right?
Wombat, thanks for checking it out. I hear plenty of highs on Faron Young, make sure your tone controls are all neutral and no surround processing is set on. You also might want to check out "Romeo and Juliet" if you want to give another one a try.
On Faron Young specifically, take a close listen to the bass line -- do you hear the individual notes of the bass line exactly the same on the two versions. Does one version spell out the bass individual bass notes better? Do you hear all the instruments on the same plane in both versions or does one version have more front to back dimension in which the instruments widen out and take on more definition. Just a couple things to look for on a recording like this one that has a lot of going on at a fast pace. This one measures close to 20dB between LEQ and peak per AudioLeak. You might want to try compressing it down to 12 dB or so if you have compression software and see if that sounds more modern and perhaps to your liking.
There isn´t much seperation in this recortding at all, i really don´t know what front to back dimension you talk about. This is played back with a Slimdevices Transporter with a build in AK4396 DAC, no tone-control.
It doesn´t matter to me how it measures on peaks. I listened Stevie Ray Vaughan´s - Couldn't Stand The Weather from MFSL, CD layer these days often and after that your sample here sounds mediocore at best, sorry.
If you cannot demonstrate that what you're hearing is not based on expectation bias, then you are not free to comment on the differences you appear to perceive, that is correct.
To be honest, I don't see the point in discussing possible hardware problems until placebo effect is ruled out.
Finally, to clarify this forum is not interested in AB tests, especially when ABX tests can be performed. If an AB test single-blind then the results are highly suspect. If an AB test is is sighted (which most seem to be) then it is completely useless.
Yes and no -- I've certainly spent a lot of care and money on my vinyl setup -- my Burl ADC alone probably costs more than most folks whole sound systems. But I would say, in my experience, that vinyl quickly exceeds CD for the right material (I listen to much 70s and 80s vinyl) and is not just providing "good reproduction" but rather "stellar reproduction" that I can't seem to get from my CD playback and my 16 / 44.1 playback is done by Burmester which is a pretty good name in the business. Even with my more modest Denon cartridge before upgrading to my Delos, I felt that my vinyl playback exceeded my CD playback when both were compared on my Burmester audio system.
Have I misunderstood your setup?
Burmester DAC yes -- do they not natively support 16 bit / 44.1? This may help explain the difference. I understand that they do. My ADC however only outputs 24 bit so this could be a source of some bias as I must convert to get to 16 bit / 44.1.
What are you asking for? Do you have a problem to overcome? It takes much too long to download even one such sample over my dial.-up line.
Unless you are doing something strange and unreasonable, or something is defective, there is no difference to be heard between before conversion and after conversion. This has been well established for a long time.
Wombat, thanks for checking it out. I hear plenty of highs on Faron Young, make sure your tone controls are all neutral and no surround processing is set on. You also might want to check out "Romeo and Juliet" if you want to give another one a try.
On Faron Young specifically, take a close listen to the bass line -- do you hear the individual notes of the bass line exactly the same on the two versions. Does one version spell out the bass individual bass notes better? Do you hear all the instruments on the same plane in both versions or does one version have more front to back dimension in which the instruments widen out and take on more definition. Just a couple things to look for on a recording like this one that has a lot of going on at a fast pace. This one measures close to 20dB between LEQ and peak per AudioLeak. You might want to try compressing it down to 12 dB or so if you have compression software and see if that sounds more modern and perhaps to your liking.
There isn´t much seperation in this recortding at all, i really don´t know what front to back dimension you talk about. This is played back with a Slimdevices Transporter with a build in AK4396 DAC, no tone-control.
It doesn´t matter to me how it measures on peaks. I listened Stevie Ray Vaughan´s - Couldn't Stand The Weather from MFSL, CD layer these days often and after that your sample here sounds mediocore at best, sorry.
Thanks Wombat, I appreciate you taking the time to listen.
If you cannot demonstrate that what you're hearing is not based on expectation bias, then you are not free to comment on the differences you appear to perceive, that is correct.
To be honest, I don't see the point in discussing possible hardware problems until placebo effect is ruled out.
Finally, to clarify this forum is not interested in AB tests, especially when ABX tests can be performed. If an AB test single-blind then the results are highly suspect. If an AB test is is sighted (which most seem to be) then it is completely useless.
Got it -- I'll see if I can sort out a way to do that.
Thanks
@Gretschguy, a couple questions:
1) You said earlier that you were using your Burl to capture at 24/96, but the samples you posted are 24/48. Are these files new captures performed with the Burl running @ 48 kHz, or downsampled from your original 96 kHz captures?
2) Regarding performing an ABX test with your system: What have you been using for playback of your 24-bit files (and related to that, exactly how have you been comparing the "hi-res" and downsampled versions)? The output of the Lynx card to your Burmester DAC? If so, and you're running Windows, the freeware foobar2000 (http://www.foobar2000.org) audio player has an ABX plugin (http://www.foobar2000.org/components/view/foo_abx) that would make performing an ABX test simple, especially as foobar can be set to always output 24-bit audio to your audio interface, regardless of whether the file being played is 16 or 24 bit (16-bit files are just padded with zeros).
Comparing different sample rates could be tricky, though, as although foobar has no problem outputting any sample rate known to man, your DAC will most likely mute while switching rates - or I should say, *hopefully* it mutes, and doesn't pop/click/etc - and depending on how long that mute lasts, it could make ABX'ing difficult.
He could just up-convert to the highest bit depth/sample rate in order to avoid all that. If ABX results are inconclusive (which will be the case in all likelihood, even taking into account the second conversion) then we're pretty much done. If the ABX results indicate a positive identification then we can start honing in on the cause of the difference, beginning with the processing then the hardware before concluding that 16/44.1 is inadequate as a delivery format for digitized vinyl.
@Gretschguy, a couple questions:
1) You said earlier that you were using your Burl to capture at 24/96, but the samples you posted are 24/48. Are these files new captures performed with the Burl running @ 48 kHz, or downsampled from your original 96 kHz captures?
2) Regarding performing an ABX test with your system: What have you been using for playback of your 24-bit files (and related to that, exactly how have you been comparing the "hi-res" and downsampled versions)? The output of the Lynx card to your Burmester DAC? If so, and you're running Windows, the freeware foobar2000 (http://www.foobar2000.org) audio player has an ABX plugin (http://www.foobar2000.org/components/view/foo_abx) that would make performing an ABX test simple, especially as foobar can be set to always output 24-bit audio to your audio interface, regardless of whether the file being played is 16 or 24 bit (16-bit files are just padded with zeros).
Comparing different sample rates could be tricky, though, as although foobar has no problem outputting any sample rate known to man, your DAC will most likely mute while switching rates - or I should say, *hopefully* it mutes, and doesn't pop/click/etc - and depending on how long that mute lasts, it could make ABX'ing difficult.
Mixminus:
(1) I capture at 24/96 and converted directly from there to 16 / 44.1, I also converted down to 24 / 48 and I compared on my Burmester system the 16 / 44/1 and 24 / 48. My Burmester DAC is limited to 48 kHz sample rates. I was going to upload all three samples but there was only 100MB available to me so I just uploaded exactly what I had compared on my system.
(2) My primary listening system is a full Burmester car audio system which includes Burmester DAC conversion, Burmester speakers, and Burmester amplification. I can't attach a soundcard to that system. I can however play back wav files of different sample rates and bit depths with a limit of 24 / 48. According to Burmester the DAC reclocks the wav files, perhaps thats a clue to something going on.
My studio machine uses an older MOTU DAC and is capable for 24 / 96, however, overall that system does not compare with the Burmester system and is not my target listening environment that I've been optimizing towards. I would not trust that DAC for much critical listening given that I experienced a major leap in sound quality when I upgraded the MOTU ACD to Burl last year -- I listen to the Burmester a couple hours each day often at loud volumes so I'm optimizing my vinyl replay to that system -- and I also rely upon the Burmester for my own studio work as I find it provides a great sound. For critical listening tests I listen without the car moving and without engine noise, it's a quiet environment and I play my music loud.
My studio machine uses an older MOTU DAC and is capable for 24 / 96, however, overall that system does not compare with the Burmester system and is not my target listening environment that I've been optimizing towards. I would not trust that DAC for much critical listening given that I experienced a major leap in sound quality when I upgraded the MOTU ACD to Burl last year
Once again, this is not compliant with #8 in our Terms of Service (http://www.hydrogenaudio.org/forums/index.php?showtopic=3974), to which you agreed to follow in order to participate here.
Thanks for the samples. A simple analysis showed that most don't even use over 20 dB dynamic range. There's one sample which uses ~31 dB dynamic range.
As for sample rate conversion I suggest SoX. Dithering shouldn't be audible with those dynamic range numbers unless you use some aggressive noise shaping that adds a lot of HF noise .. which is only audible if you don't have HF hearing loss though.
I wouldn't worry about using SoX. Izotope's sample rate conversion is fine (except maybe Advanced RX 2's "intermediate phase"). See http://src.infinitewave.ca/ (http://src.infinitewave.ca/)
Thanks for the samples. A simple analysis showed that most don't even use over 20 dB dynamic range. There's one sample which uses ~31 dB dynamic range.
As for sample rate conversion I suggest SoX. Dithering shouldn't be audible with those dynamic range numbers unless you use some aggressive noise shaping that adds a lot of HF noise .. which is only audible if you don't have HF hearing loss though.
Thanks Xnor, I measured these on AudioLeak and I get about 20dB between peak to RMS. Typically my CDs measure less than that. So is your 20dB the same as what AudioLeak is measuring? Would it help if I gave the snapshot from AudioLeak?
The last I tested my HF response it was very good. I will check SOX and I'm also going to look at the frequency spectrum difference in the samples using Logic Pro's EQ match.
I wouldn't worry about using SoX. Izotope's sample rate conversion is fine (except maybe Advanced RX 2's "intermediate phase"). See http://src.infinitewave.ca/ (http://src.infinitewave.ca/)
Thanks, this is helpful, I will check it out.
OK, I have listened to Gretschguy's vinyl clips and done a cursory ABX test with Foobar.
I was unable to hear any difference between the 24 and 16 bit versions. The playback was via an M-Audio AP2496 soundcard and modest Sennheiser HD435 headphones. I will acknowledge that the playback chain is perhaps not adequate, but I am not in a position to perform proper ABX tests on my main system (Slim Devices Transporter into ATC SCM100A speakers).
Notwithstanding my failure to distinguish 24 and 16 bit versions (which frankly does not surprise me), I do want to say that these are very nice vinyl rips. I have three of the tracks on CD and prepared the same 30s clips from those in order to compare. I also measured the RMS-to-peak ratio for the CD and vinyl rips:
1. Yes/Roundabout. My CD is an original (pre-remastered) British pressing. It has a bit more weight, whereas GG's vinyl rip has more sparkle and sounds more "airy". Perhaps the impression of "weight" on the CD is simply due to the lack of top end. RMS-peak ratios: CD=16.5dB, vinyl=17.1dB.
2. Prefab Sprout/Faron. My CD is an original (pre-remastered) US pressing (where the album is called "Two Wheels Good" instead of "Steve McQueen"). There is very little difference between this and GG's vinyl rip. The vinyl rip seems to have slightly better defined bass, but only on careful listening. RMS-peak ratios: CD=17.0dB, vinyl=16.7dB.
3. Dire Straits/Romeo and Juliet. My CD is a the first British remastered version. (I suspect there have probably been more remasters since then). The CD clearly has more dynamic range compression than GG's vinyl rip. Apart from the compression, the balance and detail seems very similar. RMS-peak ratios: CD=15.9dB, vinyl=19.0dB.
It interesting to note that there is precious little difference in RMS-peak ratios between the CD and vinyl for pre-remastered CDs, whereas there is a marked difference in the case of the remastered Dire Straits CD. Which pretty much confirms what a lot of us think about remastered CDs.
"Candyman" from the same remaster CD is measuring 13dB in Pure Music. The whole remastered CD is in this 13-14 range and the vinyl was measuring up around 23-24.
[again, I'm not claiming technical superiority of vinyl here, I'm claiming practical superiority in my opinion due to this lack of artificial compression that has been applied to recent remasters.]
Hey, any one want to buy some Siouxsie & The Banshees remastered CDs really cheap???
Sorry to reply a third time!
I've been enjoying your samples in the other thread (http://www.hydrogenaudio.org/forums/index.php?showtopic=93477&hl=). Here are some thoughts...
1) They're very good. I can come close, but not consistently.
2) They all have a slight smiley EQ (boosted bass + treble) compared with the digital copies I can find. It seems fairly consistent, which is remarkable given the various sources. The only conclusion I can draw (unless anyone has another one) is that either the vinyl mastering, or your playback, adds this smiley EQ - intentionally or not.
3) There are examples where every digital copy I can find includes audibly reduced dynamics, while the vinyl copy you posted sounds fine. I wish the Romeo and Juliet sample had lasted longer - I suspect the loudest drum beat a few seconds later would have revealed an even greater difference.
4) Even when apparently perfect, there's something about the sound of them that I associate with analogue. People say things like "more air" as a good thing, but my best guess is its down to small amounts of wow/flutter, distortion, noise, and reverb-like properties which add to the signal in a "nice" way. I may be imagining it (can ABX due to DRC+EQ differences, so that doesn't help!), but that's what I think I hear in your samples.
This last point is interesting to me. I think it's what some people prefer about vinyl. Many people claim that the vinyl lets you hear more of the original, but having heard pure digital recordings have this "added" to them on vinyl, I'm convinced it's (almost?) all "nice" distortion, and not revealing "more" of the original.
Take this as pure speculation if you wish. I'm as guilty of placebo as anyone. Can anyone think of a way us mortals can test it properly?
Cheers,
David.
P.S. I can't hear a difference between the 16/44.1 and 24/48 versions.
@Gretschguy,
I also have several different copies of certain favourite albums - sometimes because finding a decent version is so elusive. One is enough if it sounds great. Five isn't enough if they're all flawed. Strange though that the most incompetent / couldn't-care-less record companies get to make the most money out of this racket.
As for 24/96 vs 16/44.1, you're not doing any harm using the former (assuming storage is inconsequently cheap) - with a ADC and/or DAC that doesn't support 44.1kHz, you're probably doing some good.
Only problem is, you can't come on HA and claim vinyl as a medium sounds better than CD, or 24/96 sounds better than 16/44.1, without double blind test results to back up those claims. Such results are rare, because the actual advantages are vanishingly small on well designed equipment, and inaudible.
Cheers,
David.
Cheers David, thanks -- can I say that the Decca release of "Who's Next" is the best (in my opinion!!!) Noisy as all heck but sounds like (in my opinion) that there is little noise reduction or compression on that version. Check it out sometime if you like The Who. All the best...
Sorry to reply a third time!
I've been enjoying your samples in the other thread (http://www.hydrogenaudio.org/forums/index.php?showtopic=93477&hl=). Here are some thoughts...
1) They're very good. I can come close, but not consistently.
2) They all have a slight smiley EQ (boosted bass + treble) compared with the digital copies I can find. It seems fairly consistent, which is remarkable given the various sources. The only conclusion I can draw (unless anyone has another one) is that either the vinyl mastering, or your playback, adds this smiley EQ - intentionally or not.
3) There are examples where every digital copy I can find includes audibly reduced dynamics, while the vinyl copy you posted sounds fine. I wish the Romeo and Juliet sample had lasted longer - I suspect the loudest drum beat a few seconds later would have revealed an even greater difference.
4) Even when apparently perfect, there's something about the sound of them that I associate with analogue. People say things like "more air" as a good thing, but my best guess is its down to small amounts of wow/flutter, distortion, noise, and reverb-like properties which add to the signal in a "nice" way. I may be imagining it (can ABX due to DRC+EQ differences, so that doesn't help!), but that's what I think I hear in your samples.
This last point is interesting to me. I think it's what some people prefer about vinyl. Many people claim that the vinyl lets you hear more of the original, but having heard pure digital recordings have this "added" to them on vinyl, I'm convinced it's (almost?) all "nice" distortion, and not revealing "more" of the original.
Take this as pure speculation if you wish. I'm as guilty of placebo as anyone. Can anyone think of a way us mortals can test it properly?
Cheers,
David.
P.S. I can't hear a difference between the 16/44.1 and 24/48 versions.
David, thanks for the notes -- this is very interesting. Take a look also for more presence around 1k in modern CDs -- anecdotally that's what I'm seeing. I hypothesize (no proof) that vinyl sounds clearer to some people because those frequencies are not built up so much thus providing a little less subjective harshness and the vocal perhaps sounds clearer as the upper midrange is often (from what I see) a little boosted on vinyl. This is PURE speculation but worth checking out.
Also, when I use the term "air" I'm usually just referencing the upper frequencies (> 12k). It is short hand for saying "the upper high frequencies beyond about 12k". In mixing terms I see the term often used, when you add "air" you usually boost the highest frequencies available to you. I don't think of it as an pseudo science term at all -- now the word "engaging" is perhaps over the line, but I feel I really finally understood what that term meant when I actually started tapping my foot and bobbing my head to "Oh Yoko" off a vinyl rip because it sounded so fantastic (to me).
Cheers for taking the time to have a listen.
GG