Your Vinyl Transfer Workflow (sharing best needledrop practices)*

Discussion in 'Audio Hardware' started by Vocalpoint, May 11, 2011.

  1. ghost rider

    ghost rider Forum Resident

    Location:
    Bentonville AR
    Sounds fine personally I won't use declick on large segments. I have in the past and always thought it was fine recently I have watched closely on small segments and even level 1.0 can remove transients or reduced them. You should be able to test it for yourself any jazz with a lot of high hats.
     
  2. Stefan

    Stefan Senior Member

    Location:
    Montreal, Canada
    Just a couple of things.
    • If you're going to use the loudness control to adjust the average level to -12LKFS, there's no point in your 4th step of adjusting the RMS level to -18dB. The mention of -18dB in this thread and elsewhere has to do with what some folks consider the optimal level for recording through audio-to-digital converters. Since you're recording at a low volume, I assume it's below -18dB (don't go too low, or you'll be boosting the noise floor needlessly). If your step 4 is to reach a level you're comfortable with editing, then that's different, but it won't do anything to improve your recording once you've done it.
    • I would do any manual declicking before doing the full-file declick because if you have any big clicks you don't fix first, a light declicking will often result in their turning into thumps that are harder to remove.
    • As for the full-file declicking, I'd find a section of your file with a lot of treble transients such as repetitive high hats, strummed acoustic guitars, etc., and try your declicking Preview with the Output clicks only checkbox checked. This way you can hear if it'd remove more than just clicks. This is especially so if you hear repetitive clicks being removed that sound like they're actually part of the music. If so, back off on the declick strength until it sounds like you're only removing rfreal clicks.

      (Actually I find myself using the output noise only checkbox for denoising in the same way. I try to find a level that removes the most noise without removing music)
     
  3. ghost rider

    ghost rider Forum Resident

    Location:
    Bentonville AR
    @Stan94 This is a really good idea. Just test it on a clean section you know has no clicks. You will see you are removing music.
     
  4. Stefan

    Stefan Senior Member

    Location:
    Montreal, Canada
    Not always. Depends on the music. I've seen people here swear that Clickrepair ruins all music by simply running it at 1, but of course that's utter nonsense. I often run Clickrepair at 10 or 15 and on a fairly clean record, if you listen to the noise-only output, you hear almost nothing except for the occasional click. On heavily clicked material I often run something through at 30-40 and yes, in that case there's some softening of transients but the end result is much more listenable than trying to listen to it full of clicks and pops. The only reason I don't use RX for full declicking of heavily damaged stuff is because it's slower, and when I listen to what's being removed in RX, I hear a lot more phasing, warbling, etc. On the other hand, if the material has a lot of sawtooth synth sounds such as Van Halen's Jump or records by The Cars, 80's synthpop, etc., any declicking can reak havoc on the synth. So it's always a case by case thing.
     
    stetsonic and Grant like this.
  5. Stan94

    Stan94 Forum Resident

    Location:
    Paris, France
    I've found that those declick settings are pretty harmless for the records I've transferred so far, they even leave very subtle clicks in the files. As for step 4, it's true I could dispense with it, and on the other hand it's an easy way of adjusting the balance between L/R channels that works fine.
    I'm taking your points though, I will try again without auto declicking (or try with sensitivity 1 instead of 2).
     
  6. Stefan

    Stefan Senior Member

    Location:
    Montreal, Canada
    As I mentioned in my previous message, instead of just picking an arbitrary number, find a part of the record that has a lot of high-frequency transients then listen to the clicks only output in preview, then adjust the strength so that all you're removing is clicks. This will give you a fairly safe setting to use on the whole record.

    As for balancing the channels, I always do this myself, but never before denoising, because a surprising number of records are not evenly balanced and it has nothing to do with the playback system but is in fact the way they were cut. For instance, I have three different versions of Roxy Music's Avalon I recently needledropped. One is an original UK pressing that's very balanced. The second is an 80's Canadian pressing that's on average 2dB down on the right channel of side 1 and 2.5dB down on the right channel of side 2. The third is a recent reissue that's about 2dB down on the left channel of both sides! So if you balance such records before denoising, you're actually unbalancing the noise you want to remove.

    I feel it's better to do all your denoising and declicking, and then do the channel balancing when you do your track fades and silences between tracks ("topped and tailed" to use an old broadcasting term) and add markers.

    I used to just balance per album side (and do phase corrections) but lately I've found that for a lot of albums, balancing per track and phase correction per track is a better approach as even after balancing and phase correcting per side, there are still discrepencies per track. You can use the Azimuth module's suggest function, but I find it quicker to have the Wave Statistics window open along with the Gain and Azimuth modules. It's a good little mental exercise to quickly calculate the value required to balance per Total RMS and then enter the value or values in the Azimuth module to quickly adjust. Then I wait until Wave Statistics updates with the loudness values and pick a LUFS target that's appropriate for the dynamics of the recording (sometimes 18, sometimes 20, sometimes 14 or 16). I use the Gain module to quickly adjust the track level.

    I know some purists would cringe at reading this and insist that the artist's and/or mastering engineer's intent should be what determines track level, but hey, I'm doing this for my own listening pleasure and quite frankly, I don't always agree with some of the track level choices on LPs. I sometimes think they were chosen for position on the LP (lower level for inner track) or musical content constraints (more bass in the master requires lower level).

    I've found lately that matching LUFS levels works extremely well for making all the tracks sound like they came from the same performer in that the level of vocals from track to track seems exactly the same. In fact, what I've been reading lately seems to indicate that this kind of targeting loudness level with LUFS is actually a way out of the loudness war because since the major streaming outlets all have a target level they use, there's no point in mastering engineers trying to be louder than everyone else because their loud tracks will be turned down by the system anyway and that reduces the impact. The only other way to add impact is to preserve dynamics. Anyway, not relevant to our needledropping hobby perhaps but interesting nonetheless.
     
    gabbleratchet7 likes this.
  7. Stan94

    Stan94 Forum Resident

    Location:
    Paris, France
    What you're saying in the end amounts to normalizing on a track-by-track basis which is not what I would do in order to keep subtle volume differences between the tracks. Working with the whole album keeps the balance just fine, one side at a time is good also (that's the way master tapes were prepared actually). But if it works for you, that's fine by me.
     
    Grant likes this.
  8. Stefan

    Stefan Senior Member

    Location:
    Montreal, Canada
    Well yes and no. Most folks consider normalizing to be adjusting so that peaks reach a certain target level just below 0dBFS and in fact most software that has a normalizing function works this way, but there is also RMS normalization, which is closer to what I've been doing lately, which is normalization to an LUFS target. I find this to work even better than Replaygain in matching relative levels of similar tracks.

    As I mentioned above, vocals sound like it's the same person singing in the same space because LUFS levels are based on curves based from listening tests and a lot of science (see the docs at BS.1770 : Algorithms to measure audio programme loudness and true-peak audio level ). On most albums I've tried this on, there's actually not a huge variance in levels but I don't trust all mastering "engineers" to be as good as our host or some of the other well-known pros such as Bernie Grundman, Chris Bellman, Bog Ludwig, etc. Occasionally I find albums with unbalanced levels between channels, big differences between sides (often because they were cut by different people at different times), or some folks who wanted to get creative with levels. LUFS normalization actually preserves the subtle differences in level and intensity in performances while as I said making it sound like the same vocalists are performing on the same night through the same gear. I'm also surprised at how bass levels seem to be well matched since LUFS like its predecessors (loudness curves, etc.) tend to focus on midrange.

    Anyway, to each his or her own. My workflow is always evolving in this hobby or ours and that's where I'm at now. I used to process by whole sides but I'm liking the results better I'm getting now.
     
  9. psulioninks

    psulioninks Forum Resident

    Location:
    KC Chiefs Kingdom
    For me, I try to keep things "as was" on the record I am ripping from. This means normalizing based on the largest peak. So for instance if my original capture looks like this:

    DR Peak RMS Duration Track
    --------------------------------------------------------------------------------
    DR14 -6.33 dB -23.41 dB 3:10 ?-01 - Heart Attack
    DR14 -5.58 dB -23.70 dB 4:30 ?-02 - Magic
    DR12 -8.18 dB -22.58 dB 3:44 ?-03 - Physical
    DR12 -9.40 dB -25.25 dB 3:06 ?-04 - Hopelessly Devoted To You
    DR12 -8.49 dB -23.27 dB 3:19 ?-05 - Make A Move On Me
    DR14 -7.32 dB -23.63 dB 3:30 ?-06 - A Little More Love
    DR13 -6.74 dB -22.14 dB 2:50 ?-07 - You're The One That I Want
    DR13 -6.95 dB -24.11 dB 4:29 ?-08 - Tied Up
    DR14 -8.27 dB -25.58 dB 4:00 ?-09 - Suddenly
    DR12 -7.71 dB -22.20 dB 3:33 ?-10 - Xanadu
    --------------------------------------------------------------------------------

    When I finish and normalize, it looks like this:

    DR Peak RMS Duration Track
    --------------------------------------------------------------------------------
    DR14 -1.45 dB -18.53 dB 3:10 01 - Heart Attack
    DR14 -0.70 dB -18.83 dB 4:30 02 - Magic
    DR12 -3.30 dB -17.70 dB 3:44 03 - Physical
    DR12 -4.52 dB -20.38 dB 3:06 04 - Hopelessly Devoted To You
    DR12 -3.61 dB -18.39 dB 3:19 05 - Make A Move On Me
    DR14 -2.44 dB -18.75 dB 3:30 06 - A Little More Love
    DR13 -1.86 dB -17.26 dB 2:50 07 - You're The One That I Want
    DR13 -2.07 dB -19.23 dB 4:28 08 - Tied Up
    DR14 -3.39 dB -20.70 dB 4:00 09 - Suddenly
    DR12 -2.83 dB -17.32 dB 3:32 10 - Xanadu
    --------------------------------------------------------------------------------

    This keeps things "in perspective" with regard to how the album was mastered (right or wrong...good or bad).
     
  10. candyflip69

    candyflip69 What's good?!

    Location:
    Melb, AUSTRALIA
    This is what I need to do for my vinyl comparisons - I need to keep the mastering decisions and the perspectives, good and bad.
    So *how* are you doing that normalisation? I know you have the Tascam DA-3000 for recording, but what's your post-method?
    And apologies if you have already written that up in this thread: if so, please direct me to which page?
     
    psulioninks likes this.
  11. Stefan

    Stefan Senior Member

    Location:
    Montreal, Canada
    Yeah, I`ve always done it that way, but I'm actually quite happy with the results I'm getting now. As for the peaks, as I and others have mentioned, using phase correction will remove any false peaks without any compression.
     
    Grant likes this.
  12. Stefan

    Stefan Senior Member

    Location:
    Montreal, Canada
    The problem with using peak normalization for your comparisons on Youtube is that it won't necessarily result in perceived loudness levels being the same. Take for example the tracks in psulioninks list, Magic (RMS -18.83, peak -0.70) and A Little More Love (RMS-18.75 dB, peak -2.44 dB). If someone were to peak normalize these two tracks, to say -0.5dB then Magic would be raised by .2dB and sit at an RMS of -18.63 whereas A Little More Love would be raised almost 2dB to an RMS value of -16.81. Since humans perceive loudness by RMS and not peaks, they would perceive A Little More Love as much louder (since I don't have LUFS values for either, I have to use RMS, which is similar but doesn't take into account the effect of frequencies on perceived loudness).

    Now in your case, since you're comparing samples of the same songs on different pressings, chances are that there won't be that big of a difference but again peak normalization is absolutely the wrong way to go for your purposes. If one of the samples you post ended up being a bit more dynamic than another, it would have higher peaks. If both are peak normalized so that their peaks hit the same target level, then the more dynamic clip will have its average level fall lower and it will be perceived as less loud. If instead you target each clip to -14 LUFS (the target level Youtube uses), then your clips will be perceived as being the same loudness level and your viewers/listeners could make a fairer judgement of the characteristics of each clip. You may need to use some limiting on peaks that exceed 0dBFS, but that wull have little to no impact on the perceived sound.
     
  13. psulioninks

    psulioninks Forum Resident

    Location:
    KC Chiefs Kingdom
    Not a problem...

    1. After I finish recording, I have two files - one for Side A and one for Side B (four if it's a double LP).
    2. I transfer these from my SD card to my laptop and run each one thru ClickRepair (CR) at level 1 (DeClick only, no DeCrackle).
    3. I then bring Side A into Izotope and split the file into individual tracks, export these tracks into a new folder and delete the one larger CR file.
    4. I repeat this process for Side B.
    5. I will then run a dynamic range report in foobar like the first one you see above as my "reference" file.
    6. I then work manually in Izotope song-by-song to clean-up each individual track exporting the finished track to its final folder and deleting the "working" track.
    7. When I am done with all the tracks, I will run another dynamic range report in foobar and compare this to the original to see if my "cleaning" has affected the loudness for any of the tracks. Most of the time absolutely nothing changes. Now and then I will get a peak that will be reduced due to cleaning. If there are changes, I use this second file as my new reference and pitch the original.
    8. Then I do two things:
    .....I gather dead wax info so I know if the album being ripped was mastered by the same engineer/same facility
    .....I look at the dynamic range report to see if there appears to be similar overall loudness levels between all songs
    9. If all looks similar, I will add meta information to the tracks in xrecode II and then merge/cue ALL tracks into one file. If not, I will create a separate file for each side.
    10. I open this merged file in Izotope and then use Modules > Normalize to set my peak at -0.70 dB.
    11. I save this normalized file over the original.
    12. Then back into xrecode II to split the tracks again using the cue file.
    13. I run a final dynamic range report in foobar like the second one you see above.
    14. Finally, I grab artwork and create the final FLAC files.

    That's how I do it...
     
    gabbleratchet7 and candyflip69 like this.
  14. Stefan

    Stefan Senior Member

    Location:
    Montreal, Canada
    Here's an example of stats from a needledrop I did last weekend of the Canadian Roxy Music Avalon pressing I mentioned.

    DR12 -3.54 dB -17.82 dB 4:32 01-More Than This
    DR15 -1.09 dB -18.03 dB 4:32 02-The Space Between
    DR12 -2.80 dB -17.55 dB 4:18 03-Avalon
    DR12 -2.20 dB -18.01 dB 1:46 04-India
    DR14 -0.68 dB -17.26 dB 3:27 05-While My Heart Is Still Beating
    DR13 -2.04 dB -17.24 dB 3:55 06-The Main Thing
    DR13 -2.46 dB -17.72 dB 4:44 07-Take A Chance With Me
    DR12 -3.08 dB -17.60 dB 4:18 08-To Turn You On
    DR12 -3.43 dB -17.43 dB 4:28 09-True To Life
    DR10 -10.88 dB -23.54 dB 1:27 10-Tara

    (As I posted earlier, I agree with our host that DR should never be used for measuring dynamic range. I consider it a fatally flawed approach, but it does give me a quick and easy readout of the stats)

    This doesn't show LUFS values but all tracks except for the last one were adjusted to an LUFS value of -18. It's interesting to see how the actual RMS values are close to that but they're clearly not the same thing. The last track is a quiet track of mostly ocean wave sound effects and some instrumental music but no vocal and it made no sense to adjust it to -18 LUFS.

    So you can see that my approach of targeting -18 LUFS still allows for lots of subtle and not so subtle variations in level.

    Here's the waveform.
    [​IMG]
     
    Stan94 and Grant like this.
  15. candyflip69

    candyflip69 What's good?!

    Location:
    Melb, AUSTRALIA
    Yes, right - it is complex.

    OK, so using the DR report in foobar and your suggestions from last month, I did make my latest video with all -14LUFs and it certainly did seem much more similar loudness across the 30 samples I did (as you would expect).
    So no complaints there.
    I see where psulioninks is coming from, but you're right, my needs are slightly different.

    So will stick to the new protocol I have, that you helped me with (and thank you very much for that, BTW) :righton:
     
  16. psulioninks

    psulioninks Forum Resident

    Location:
    KC Chiefs Kingdom
    I agree with the math here, but why would anyone change these two songs to reach -0.5dB each? The normalize function in Izotope would only raise Magic to -0.5dB and A Little More Love would then be at -2.24dB keeping the difference between their relative loudness peaks the same.
     
  17. Grant

    Grant Life is a rock, but the radio rolled me!

    I do think the obsession with dynamic range is getting silly.

    I don't worry about dynamic range until just before dithering. Then I shoot for LUFS -18 or LUFS -14. If a couple of files get some limiting, I don't worry as long as it stays around 2 or three decibels.
     
    Stefan likes this.
  18. Stefan

    Stefan Senior Member

    Location:
    Montreal, Canada
    Sorry I should have been clearer perhaps, I didn't mean to say that's what you should be doing. It was an example I was using for candyflip69 since he's creating comparisons of different pressings for his Youtube videos. Actually what you're describing as part of your workflow (i.e. normalzing per album side) is exactly what I've always done. I just started experimenting lately with normalizing per track to an LUFS value and so I described it here as I like the results I'm getting.
     
  19. Stefan

    Stefan Senior Member

    Location:
    Montreal, Canada
    Yes, this is very much where I'm at these days as well. I listen to a lot of Spotify in my car as I spend 1-1.5hrs daily commuting and I like how their -14 LUFS target works really well at keeping the relative level of tracks the same while still allowing for lots of dynamics (if the material is properly mastered). In fact, for M4A files I create for the USB stick I keep in the car, I use dBPoweramp to adjust all tracks to -14 LUFS and use Ozone's limiter to prevent any overs, whether they're modern tracks that actually get turned down or needledrops over really dynamic stuff that gets turned up. It all ends up sounding good for car listening. Obviously in a home listening environment, more dynamics are desirable, so I usually keep my hires needledrop source files at -18 or -20 LUFs depending on the level of dynamics.

    And yes, the whole dynamic range thing is a bit out of hand in these parts. I get that its desirable to preserve dynamics, but I think some folks (I'm not alluding to anyone in this thread) are overreacting to the whole damaging trend of the loudness war. What gets me is people dismissing certain releases because they don't have astronomical DR ratings. DR is a completely flawed approach to dynamics measurement that is easily fooled. For instance, I recently took the original CD release of Rush's Vapor Trails album and ran a DR analysis on the album. It came out DR5 as it was a celebrated victim of the loudness war. I then ran an analog brickwall high-pass filter at 20Hz, saved the file, and re-ran the DR analysis. the result. DR11! Yet of course, it still sounded every bit as squashed as before. What really gets me is folks who compare vinyl pressings with DR since different cartridges, phono stages, tonearms, etc., all have an impact on the dynamics. There are far more accurate, professional ways to asses dynamics (crest factor, loudness units, etc. Some really good info available here: TC Electronic | Loudness Explained ).
     
    stetsonic and Grant like this.
  20. psulioninks

    psulioninks Forum Resident

    Location:
    KC Chiefs Kingdom
    OK, that makes sense...I was confused there.
     
  21. Quick change of direction. I was wondering how many needledrops folks have done. I saw online a guy who has done 10,000 or so. Just incredible. His work is extremely good as well. I've done around 1500 over the last 10+ years. I currently have 497 needledrop albums on my hard-drive.
     
    BrilliantBob likes this.
  22. Stefan

    Stefan Senior Member

    Location:
    Montreal, Canada
    I've probably done a good couple of thousand since starting the hobby 15 years ago. I find I keep doing them over and over as my gear and techniques improve. Right now I have about 400 albums on file all done in the past 2 years.
     
    arisinwind and Grant like this.
  23. vinyldoneright

    vinyldoneright pbthal

    Location:
    Ca
    Pretty sure I am the 10,000 rip guy and I too have a habit of reripping for a hardware change
     
    finsaah, Stan94, Stefan and 2 others like this.
  24. Stan94

    Stan94 Forum Resident

    Location:
    Paris, France
    Has the SugarCube made your ripping process easier/faster (I bet it did) ?
    Isn't using such a device akin to cleaning a file with a software, with the possible drawbacks of using the wrong settings and overdoing the declicking, or underdoing it ?
     
  25. ghost rider

    ghost rider Forum Resident

    Location:
    Bentonville AR
    Here's another example of Stefan's denoise process. It'd the little drum solo in whole lotta love with before and after denoise samples. I don't hear any damage. The only thing I can find that changed is the very faint echo intro of Heartbreaker was slightly reduced. Most people don't even know it's there.
    Dropbox - Led Zeppelin 2.flac - Simplify your life

    [​IMG]
     

Share This Page

molar-endocrine