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
    I discovered a weird anomaly using M/S and denoise. Not sure what to make of this. I did Pink Floyd the wall and everything looked good as it consistently has been. This is a huge defect now I'm questioning whether or not I need to abandon the process. I have been denoising in M/S and when I convert back out of M/S after denoising the keyboard bass line at the very end of side 2 converted back into both channels. If I switch to M/S and back without denoise no issue but denoise under these conditions the bass line bleeds into both channels.
    [​IMG]
    [​IMG]
     
  2. Grant

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


    I've been seeing that too. I also wonder if there's a bug somewhere. I'll denoise in M/S and convert back, and the denoise will not have taken affect.
     
    ghost rider likes this.
  3. BrilliantBob

    BrilliantBob Select, process, CTRL+c, CTRL+z, ALT+v

    Location:
    Romania
    The best de-noise method I discovered by experience is to reduce from the beginning anything under -75 dBFS. Why this?
    - 1. An LP made out of perfect vinyl would have a theoretical dynamic range of 70 dB.
    - 2. The measured aggregated noise added into the signal, eg. the quantization noise, aliasing, the crosstalk, the recording chain noise (TT-preamp-digital inteface-PC with proper ground), etc. reachs up to -85 dBFS total RMS with peaks up to -75 dBFS.

    I create a 30 sec learning sample of -75 dBFS gaussian noise before the recorded audio signal, and then I denoise the audio file back and forth in more passes with this mask until the 30 sec gaussian noise completely dissappear.

    I remove the RUMBLE together with the tonearm vibrations and I keep the bass signal and harmonics with a Butterworth high-pass filter order 6 at 17 Hz (-36dB/octave). I gave up the M/S BS or steep signal cuts. In the sub-bass area there are harmonics of the higher freq signal which give power, consistency, soundstage and presence to the audio signal. That's why is not recommended to cut everything under some freq threshold.
     
    ghost rider and arisinwind like this.
  4. ghost rider

    ghost rider Forum Resident

    Location:
    Bentonville AR
    I'm going to try it in 7 and 6 if necessary. To see if it is a bug with 8.
     
    Grant likes this.
  5. ghost rider

    ghost rider Forum Resident

    Location:
    Bentonville AR
    I am curious if anyone else can duplicate my results using the Wall. You can highlight just the end of side two to see. I don't think the settings of denoise matter. It's after denoising switching out of M/S. Perhaps similar to how declicking after denoising is a different animal. Denoising corrupts the process and M/S now fails to convert it back perfectly.
     
  6. ghost rider

    ghost rider Forum Resident

    Location:
    Bentonville AR
    Glad I have all my 32/96 wav archives.
     
    paulybauls and arisinwind like this.
  7. Stefan

    Stefan Senior Member

    Location:
    Montreal, Canada
    (I removed your images from the quote of your message because of the forum rule of no more than 5 images per message). Did you forget to do a switch at some point? On my copy of The Wall, the stereo L/R version looks like your top image because the bass line is panned mostly to the left.
    [​IMG]

    When I switch to M/S, it looks like your bottom image.
    [​IMG]

    Here's the same section with denoising done on the Side channel:
    [​IMG]

    And here it is back in L/R with the second pass of denoising done and the end silenced.
    [​IMG]
    Note that there's some crosstalk between channels which is common as there was likely an elliptical filter applied to sum the bass and since it was analog in those days, the filter goes up to the lower midrange. My denoising really exposes the lower level crosstalk on the notes into the right channel. Also, you can see the limitation of RX's denois in that the noise below each note is not removed. I could have done that by cranking up the Reduction value, but that would have removed some of the music. It's always best to "tune" the denoising value by previewing with "Output noise only" checked.

    So double check your steps because it seems a bit strange for you to feel you need to completely scrap a method that's been working for you really well (i.e. denoising in M/S) just because one result didn't turn out! M/S doesn't cure everything; it always depends on the source but it does work about 95% of the time in my experience. If you do it properly, there should be no bleed.
     
    ghost rider likes this.
  8. Stefan

    Stefan Senior Member

    Location:
    Montreal, Canada
    The problem with this is that just because there's noise below -75 dBFS, that doesn't mean there's no musical (or audio) information below that level. Reverb tails, note tails, low-level percussives, etc., they all fade to a much lower level than -75dB. They may be masked by the noise but they don't suddenly disappear at exactly -76dB! As I`ve mentioned before, a lot of the background noise from vinyl occurs primarily in the Side channel, which is why normally denoise it separately in Mid/Side mode because then I can do a much milder denoise in L/R so that lower bass is preserved (since most older vinyl especially was summed to mono in the lower frequencies anyway). If one just denoises in L/R, a lot of ambience and note fades, reverb tails, etc., are removed.
     
  9. ghost rider

    ghost rider Forum Resident

    Location:
    Bentonville AR
    I get what you are saying about scraping the whole process. I'm going to try it again to see what I did wrong. Is it also possible some times weird things happen?
     
  10. ghost rider

    ghost rider Forum Resident

    Location:
    Bentonville AR
    This morning I did a test to just that part at the end of the record and it seemed to work. I reran the whole denoise process over. Adding -85db of noise, getting the noise scan, convert to M/S and 1st pass only to side, 2nd pass to both channels new scan much lower level, convert back from M/S. Again the bass line is in both channels.

    Stefan from your post above it looked like it also worked for you? Were you only denoising the end of the file?
     
    Last edited: Apr 5, 2021
  11. ghost rider

    ghost rider Forum Resident

    Location:
    Bentonville AR
    It's the unknown that concerns me. I'm not sure what I'm doing different than you but additional tests are leaving the bass line in both channels.

    Also I think it is a good thing the this level of denoise does not attack the noise under the music, kind of fades it beneath it. The music masks the noise and it's need primarily in the quiet parts.

    This was looking promising a simple process I could run mostly on every file. The hard work is the recording and declicking. I always use very clean records and all clicks are removed manually. This can take half the day. I can almost live without denoise if it can have unintended consequences that I discover months later.

    I think in this case because all the bass was in the left channel this may have created the perfect storm for this anomaly.
     
    Last edited: Apr 5, 2021
  12. Stefan

    Stefan Senior Member

    Location:
    Montreal, Canada
    Retrace your steps.
    1. For this section of the song, you should start with the bass primarily in the left channel.
    2. Then switch to M/S using the preset I've mentioned in the past or even the one that comes with RX (if the latter then you can't use the same preset to switch back to L/R; mine is designed for toggling, but the one with RX means you need to use the other one that comes with RX for M/S to L/R).
    3. In M/S, you should have the bass in both channels.
    4. Never mind denoising for now, just switch back to L/R. The bass should again be mostly in the left channel. If it's not, you're doing something wrong.
    5. Now switch back to M/S again. Denoise just the Side channel.
    6. Switch back to L/R and check that the bass is again mostly in the left channel (or even just preview with the Mixing module for whichever preset you use to switch back to L/R.
     
  13. ghost rider

    ghost rider Forum Resident

    Location:
    Bentonville AR
    If I switch back and forth in M/S no issues. I'll go through it again specifically following these steps. I did do a 2 pass denoise without M/S with lower 1st pass level and bass is removed so.

    Just redid only the last part pasting it in it's own file. Again all the bass is summed into both channels.
     
  14. ghost rider

    ghost rider Forum Resident

    Location:
    Bentonville AR
    Okay if I use +/- 100 in M/S it works better. It still leaves a small amount of bass in the right channel.
     
  15. Stefan

    Stefan Senior Member

    Location:
    Montreal, Canada
    You can't toggle using the same preset with +/- 100 because that will result in an imbalance. it really needs to be +/- 71 (actually it's exactly 70.71 but RX rounds it off).
     
  16. BrilliantBob

    BrilliantBob Select, process, CTRL+c, CTRL+z, ALT+v

    Location:
    Romania
    This is an example of crosstalk. I recorded this sample with right channel disabled. The red and green wires are not connected to the TT cartridge. Crosstalk is pure noise, not part of the music on vinyl. That's why I reduce anything under -75 dBFS. To make the needledrop as transparent as possible. But maybe is just me. Anyway the denoised music sounds very good. The music acquires a three-dimensional relief as if the performers were singing live in front of you.

    [​IMG]
     
  17. ghost rider

    ghost rider Forum Resident

    Location:
    Bentonville AR
    I'm doing my own learning. This example makes me question if using M/S to get a strong level of reduction will work in all instances. After spending several hours using the noise only output and reading the help several times I came up with my own settings that did not remove any bass. That part of the the record was very good for this. I used a 2 pass denoise for less aggressive but still removing the noise floor in the quiet areas.
     
  18. Stefan

    Stefan Senior Member

    Location:
    Montreal, Canada
    Normally when we talk about crosstalk as it pertains to vinyl, it's produced by the cartridge. The cartridges most of us use are only rated with between 20-30dB of separation and it's very dependent on frequency. There might be a rating of 25dB at 1kHz but only 15dB at 15kHz. In addition, as I mentioned in a previous post, a lot of vinyl, especially non-audiophile releases from the pre-digital days had their bass summed with an analog elliptical filter, which not only affected frequencies below 100Hz but sometimes up to 500Hz or even 1kHz (you can try some of the free TDR plugins as their bass sum feature does a really good emulation of this).

    However, in this case, if you disconnected the right channel, then something else in your chain is transmitting signal to the other channel; it could be your phono stage or your analog-to-digital converter, but as you can see, the noise in your right channel increases in amplitude along with the music in the left channel. So it's not just noise, I'd say it's coming from your gear. I see you have a Cambridge Alva, which has a crosstalk rating of -85dB for MM and -75dB for MC. The MOTU M2 has a rating of -102dB so it could be coming from the Alva.

    Bob, you can certainly do whatever works for you, just as we all do. However, I stand by what I posted this morning, that the audio information on an LP does not suddenly disappear below -75dB. It's still there. In fact, one could make the case that the noise floor works like dither for digital audio. In the case of a 16-bit CD, by definition, there should be no perceptible audio below 96.32 dB, but by adding dither noise, there's identifiable signal to down to about -115 dB!
     
    Last edited: Apr 5, 2021
    BrilliantBob likes this.
  19. Stefan

    Stefan Senior Member

    Location:
    Montreal, Canada
    But did you sort out why the bass was appearing in both channels equally? There was definitely something wrong with your workflow in that case because there's no way it should have happened. Could you post one of your dual raw and denoised versions so I can process your file?
     
  20. ghost rider

    ghost rider Forum Resident

    Location:
    Bentonville AR
    I never figured out why that was happening. I guess it is possible the 1st two thing I have been doing to every file is cut 0-5 hz to fix dc offset possible issues and and highlight 0-20 hz and eq highpass brickwall eq at 20 hz. Has been reducing a fair bit of grunge.

    I'm restarting my computer because of other weird issues maybe my entire morning was just a big windows malfunction.

    Here is my latest malfunction. After denoise and using this loudness I get these peaks??? I exported the file and the stats say I have 4 and 6 possible clips. There is not one clip the shotgun blasts at the end are -1.6 and -1.0 something is whacked.
    [​IMG]
     
    Last edited: Apr 5, 2021
  21. ghost rider

    ghost rider Forum Resident

    Location:
    Bentonville AR
    I see nothing wrong with this file. How can these stats be right?
    [​IMG]
     
  22. Stefan

    Stefan Senior Member

    Location:
    Montreal, Canada
    Yikes!
     
  23. john morris

    john morris Everybody's Favorite Quadron

    Location:
    Toronto, Ontario
    No argument there. My sister, Rosie, does the same thing. She wants to give people needle drops that sound like compact disks. With $65 000 worth of vinyl playback equipment it should be real easy for her.

    She takes any vinyl record, cassette, or RTR that was never released on CD or digital file and transfer them to Pro Tools at 24/176.4 and/or DSD. As well as CD. No ticks and pops are allowed. But she won't remove noise if it hurts the music. Mostly.. He services aren't cheap. Did you know they are $8500 Jamaican music albums / singles that never made it to CD? For South Africa the number is higher.
     
    Anton888 and ghost rider like this.
  24. john morris

    john morris Everybody's Favorite Quadron

    Location:
    Toronto, Ontario
    You and me buddy. No argument there.!
     
    ghost rider likes this.
  25. john morris

    john morris Everybody's Favorite Quadron

    Location:
    Toronto, Ontario
    Hey! In 1982 they told us the CD was perfect sound forever. I want my money back!
     
    BrilliantBob and ghost rider like this.

Share This Page

molar-endocrine