Vinyl Restoration  -  Digitising Your Vinyl Record Collection

Planning

"If a job's worth doing, it's worth doing it well" applies to digitising vinyls.  However, doing it well is an extremely time-consuming process.  Obviously, you have to record the tracks in real time, but the time spent digitally editing the wave files to remove scratches is usually far more significant.  I had about 150 singles, 400 - 500 LPs, inheriting another 100 - 200 when my mother died, and after some years I have only just finished.  If you have any quantity, you need to think about reducing the workload.

Popular music, in particular, dates very quickly, and a lot of it is very similar and formulaic.  Generally, only a small percentage from any era is worth keeping.  Even allowing for the fact that my era  -  late 60s/early 70s  -  was a particularly creative time, the number of pop albums that I want to keep in their entirety is low.  Realising this more than halved my problem.  Accordingly …

1. Listen critically through the entire collection, and classify them as to whether:
   • Like most or all of the album  -  as you go along, gather these into an album list and trawl through shops and retail web sites for reduced "3 for £18" offers, and the like, also for compilations that would make a significant enough dent in the track list to be worthwhile, for example 'Greatest Hits' may contain all or nearly all of a of a particular artists' tracks that you actually want, leaving few to obtain elsewhere.  I've replaced much of my collection cheaply this way. 
   • Want perhaps one or two tracks  -  gather these into a track list; you might be able to obtain individual tracks from a range of preferably legal or other sources such as legal downloads, local libraries, etc.
   • Outgrown them, can live without them now  -  give to a charity shop, etc.
2. Hopefully this will reduce those to be digitised to relatively small numbers mainly of individual tracks plus the odd deleted entire LP  Concentrate your digitising efforts on these that you can't obtain any other way.

What You Need

• Record Cleaner

  If you have a serious quantity, then a cleaning machine can remove much of the dirt from the grooves before recording, which is much easier to do and gives better results than trying to remove the resulting noise using software.  Here are two samples (each a *.wav file of about 8MB, 45s ) before and after  -  note how many 'scratches' actually turned out to be dirt in the grooves that was removed by cleaning, and that remaining scratches are mostly quieter and less intrusive, and therefore will also be more easily fixed by software.

  For those who haven't seen one before, I will describe my washer.  It consists of a motorised turntable the size of a vinyl's centre label, the central spindle of which is screw-threaded, and a radial, velvet covered, hollow arm, the top surface of which has a slot level with the turntable, and which is connected to a tank and a vacuum pump.  The motor and the pump are controlled by seperate switches on the front.  The procedure for washing a vinyl is:

  1. Attach it with Side 1 uppermost clamping it with a plastic nut which screws onto the spindle.
  2. Switch on the motor.
  3. Using a fine brush provided, which in width is a little over half the playing radius of an LP, wet the entire upper playing surface (not the label) thoroughly with a cleaning agent provided, a solution of 1 part in 4 IPA.  Once completely wetted, starting at the outer edge, angle the brush into the grooves against the normal direction of travel for a few revolutions, then move to the inner edge and repeat.
  4. Stop the motor.
  5. Turn the vinyl over.
  6. Start both the pump and the motor.  The wet Side 1 is now underneath, rubbing over the surface of the arm, which sucks off the IPA and hopefully all the dirt from the grooves with it, leaving it clean and dry.
  7. During 6 for Side 1, do 3 for Side 2, now on top.
  8. Repeat 4 through 6 for Side 2.

  Either the record the vinyl straightaway, taking it straight from the washer to the deck without putting it back in its sleeve, or else buy new plastic inner sleeves for storing cleaned vinyls.

  IPA is quite strong, and may affect some plastics such as soft brushes, and possibly even some poor quality vinyls, and also the washing procedure is another potential source of accident such as dropping.  Therefore follow the failsafe procedure outlined below of recording twice, before and after washing.

  IMPORTANT

  • IPA is highly flammable, and must be stored and used appropriately.  Couriers may refuse to carry it or charge extra for so doing.
  • The washer's tank is open topped, and must be periodically drained, especially before moving the washer.  Before transportation, the washer needs to be stood, in a plastic washing up bowl or similar, on the end nearest the drain tap and the bottom removed so that the dregs of the fluid can be cleaned out and the machine allowed to dry out completely.
• Record Deck or Player

  There are various important considerations wrt vinyl equipment:

  1. Setting Up  -  Ensure that:
     • The pick-up arm swings, both laterally and vertically, freely but without appreciable play in either sets of bearings.
     • The turntable bearings do not have play and are lubricated as prescribed by the manufacturer.
     • The turntable is level, use a spirit level in two directions at right-angles to each other.
     • The turntable rotates at the correct speed; some decks have a stroboscopic display and a method of adjustment to set the speed accurately from the mains frequency.
     • The cartridge is aligned to be tangential to the vinyl groove in the middle of the playing radius.
     • The cartridge head is level, that is, its top is parallel to a radius of the turntable.
     • The counterweight is adjusted so that the arm sits level without stylus pressure.
     • The stylus pressure is correct for the type of cartridge and stylus.
     • The anti-skating bias compensation is correct for the type of cartridge and stylus, and the chosen stylus pressure (circular motion and friction of the stylus in the groove combine to give the head a tendency to skate inwards, bias compensation corrects this).  Having said that, whether the 'correct' setting is actually the best has been disputed historically.  Feel free to experiment to improve equalization of the two channels (a mono vinyl might be best for this), stereo separation, and perhaps reduce distortion.
     • The stylus is clean and not worn  -  examine it under a 20x hand lens or equivalent; dry deposits such as fluff can be removed with a conventional stylus brush, sticky deposits by dipping the brush in IPA.  The sides of the stylus should be conical and not stepped by wear; if necessary, replace it or get it retipped.
  2. RIAA Equalisation  -  To optimise reproduction, frequency response is biased before cutting a master, requiring vinyl playback output to be corrected for this.  Further, a cartridge can not drive an audio Line-In such as that on a soundcard.  Therefore, cartridge output should pass through a special phono preamp which both corrects the bias and raises the signal to line level.  Naming varies, but generally a player contains an integral phono preamp and thus its output can be connected to the Line-In of the soundcard, while a deck has no preamp and outputs the cartridge signal directly, which therefore must be fed through an external phono preamp.  This might be a separate dedicated unit, or the Phono inputs of a hifi amp, or those of a mixing desk.  Connect this unit's Line-Out to the Line-In of the soundcard.
  3. Electrostatic Charge  -  Friction of the stylus in the groove can build up enough charge to give a nasty kick to the operator, and for this reason record decks and players should be earthed.  Traditionally, in the UK, with players this was via a three-core mains lead, but nowadays with decks is more often via a dedicated connection to an grounding point on the amp.  However, the amp itself is unlikely to be earthed via its mains lead, so I wonder whether this is really adequate.  Earthing may also create a mains hum loop, see the next point.  No advice can be guaranteed, so experiment and choose what works best.

  4. Mains Hum  -  The curse of decks, even players, the world over.  Worse still, there is no guaranteed, painless cure!  From theoretical electronic considerations, the best way to cable a deck is probably something like as follows, however this is not commonly done, presumably because the smallest commonly available co-axial cable is stiffer than the delicate cabling commonly used for an arm, and would therefore give problems with stylus pressure and tracking, although I suspect that, if they would but choose, manufacturers could almost completely eliminate these at the design stage:

     • The signal leads from the cartridge should be coaxial with the inner being the signal and the outer being the ground.
     • Preferably the entire arm assembly, but at least the head, should be electrically insulated from the metal of the deck.
     • For decks whose motors are powered by mains AC, the metal of the deck should be earthed via a three-core mains lead in the manner formerly common in the UK.
     • There should be seperate ground lead from the metal of the head to the ground point on the (pre)amp.
     • Underneath the deck, the signal cabling must be kept as far away as possible from the metal of the deck, the motor, and motor electrical supply cabling.

     Based on this, here are some things to try to get rid of hum:

     • It's usually, but not always, best that both or neither deck and amp, rather than just one of them, are earthed to the mains via three-core mains leads.
     • If both deck and amp have twin-core mains leads with figure of eight connectors, as is now common, try all four combinations of polarity, by pulling leads out of the back, turning them round, and re-inserting them.
     • Relay the cabling under the deck to move the signal cabling further from the metal of the deck, the motor, and the motor's electrical supply.
     • If there is a direct connection between the cartridge signal ground and the metal of either the head or the deck, replace this arrangement by one taking the cartridge connections straight out to the amp input, while grounding such metal separately to the ground point on the amp.
     • If the motor is mains AC, earth the deck's metal, if necessary replacing a two-core by a three-core mains lead.  Note, if previous points are not attended to first, this might increase hum rather than reduce it.
     • Rewire the whole arm from the cartridge back, replacing existing delicate cabling with co-axial cabling, but, as this will involve dismantling the entire assembly, soldering probably awkward connections, reassembly, and readjustment, it's not for the technically inept or faint-hearted!  Whether someone should attempt this will depend on technical competence, how bad the hum, and whether the arm assembly's design will allow thicker cabling without significantly affecting the arm's horizontal and vertical swing  -  ideally:
       • The point where the cable turns from horizontal to vertical in leaving the arm should allow both for the thicker cable and for it to flex sufficiently not to affect stylus pressure; preferably the cable should simply hang off the arm at or very near the point of balance, so that any extra weight of it can be accommodated by balancing the arm in the usual way;
       • The arm should be mounted on ring bearings allowing the cable to hang vertically through the centre of them;
       • There should be sufficient clearance under the deck to allow the cable to be brought back up to the fixed point of attachment in a gentle loop so that it will not affect tracking.
       I have done this twice successfully, in the sense that all hum was removed and so overall the results were better, but while the first deck, a Garrard, fitted the above conditions quite well so that the problem of the new cabling's extra stiffness was easily solved, the second, a Dual 601, had a badly designed arm assembly with such a tortuous route to get the cable below the deck that the problems were only solved adequately with great difficulty after much trial and error.

• Soundcard
  Must have a stereo Line-In socket, and the ability to sample in stereo at 16 bits and 44.1KHz.
• Digital Audio Recording And Editing Software
  I have been using DCart v5 and Sound Forge v7.  Both are commercial software with a significant price tag.  I've tried various cheap or free pieces of software the names of which I cannot now remember, but their scratch and noise removal or other features were inadequate.
• CD Burning Software
  Either this or the recording/editing software must be able to set track breaks at sector boundaries so that continuous albums will not have breaks at track boundaries.
• CD (Re)Writer.

Procedure

1. Ensure that the deck is well set up  -  appropriately connected, appropriately earthed with no mains hum, correct and accurate turntable speed(s), good quality stylus, balanced arm, correct stylus pressure, correct bias compensation, etc.
2. Clean the stylus.
3. Clean the record using a cleaning cloth or other such method.  Do NOT use a machine yet.
4. Record both sides of the vinyl as wave files (*.wav), usually about 420MB.  Set levels such that the loudest signal excluding scratches almost fills the available volume range without clipping.  Afterwards, check recording peaks for clipping, and if necessary re-record with lower levels.  Avoiding writing with the vinyl surface as background, note on a corner of the inner sleeve, or on a sticky note attached to it, the final levels and where the loudest non-scratch peaks are on the album.
5. If you have a cleaning machine, repeat 2 - 4, using it in 3.  Choose the best recording to work with, but do not delete the other one until work on the album is finished.
6. Especially for the first few album sides you record, ensure that you listen critically to them in their entirety before moving on and recording others.  You need to ensure that there are no breaks caused by scheduled computer tasks such as screen-savers blanking the screen, automatic operating-system and program updates, etc  -  if necessary for the duration, set the screen saver timeout to 40 minutes or more, and set updates to be performed manually (and don't forget to do them).  Hopefully, if all has gone well, you now have a recording of each side looking something like (for a stereo vinyl):
7. If the album consists of tracks separated by periods of silence, it's probably easiest to split the wave file into individual tracks now, and deal with each track separately, but if the tracks run into one another, you will have to keep the whole side together as a unit for the next stage.
8. What happens now is software dependent.  Some can mark or interpolate scratches (aka impulses) automatically, but usually many more normal transients than genuine flaws are marked or interpolated, so false marks overwhelm one, or automatic interpolation ruins the sound.  I only trust my ears to find scratches, and then use the Interpolate command in DCart to remove them by replacing the current selection with interpolated waveform, though care is needed in choosing start and end points for the selection, as small changes can make large differences to the replacement.  For some types of music, for example pipe music, interpolation will reduce the severity of a scratch, but still leave an appreciable knock, and for these it is necessary to replace the damaged section of waveform by copying from an identical section close to the damage.  Notwithstanding the caution above, I sometimes apply Sound Forge's Click and Crackle Removal for auto removal of vinyl 'wear'.  My method in more detail is:
   1. If you have turntable rumble, sample it from the side or track lead in or lead out, and save the sample.  For obvious reasons, this has to be done now, but is easily overlooked.
   2. Mute silent passages such as the lead in and lead out of each track or side.  Fade in and out the second immediately before and after the genuine sound.
   3. Manually interpolate all the big scratches, the ones that are easy to find both aurally because they disrupt the sound unacceptably, and/or visually because their peaks stand out and perhaps cause clipping.  Here's an example from a classical guitar recording (click images to play samples):

      Before
      After

   4. Remove the smaller scratches.  This will be much more time consuming because they are too small to stand out visually in the waveform, the only method of finding them being aurally.  A typical procedure might be to use what programmers call a binary chop method by dropping markers to bracket the location of the scratch ever more closely:
      1. Playing back in the digital waveform editor, drop a marker the moment a blemish is heard.
      2. Drop another marker before the blemish, and play between them to check that they bracket it.
      3. Select from before the leading marker to half way between the markers and play this back.
      4. If you hear the blemish, then it lies in the first half of the marked area, so move the trailing marker onto the trailing edge of the selection, otherwise it should be in the second half of the marked area.  To check, drag the leading edge of the selection to a little beyond the trailing marker, so it becomes the trailing edge, and play this new selection.  You should hear the blemish, proving it lay in the second half of the original marked area, so drag the leading marker to lie on the leading edge of the new selection.
      Thus you have halved the size of the marked area.  Repeat this halving procedure until you've located the scratch visually, and fix it by whatever method of interpolation is supported by the software.  Here's an example containing a number of them close together, before and after they've been interpolated (click images to play samples):

      Before, some of the blemishes are just becoming visible at this zoom level
      Detail showing one of many small blemishes selected for interpolation, and others yet to do
      The same showing it interpolated (the audio demonstrates the result of interpolating them all)

   5. If 'needle in the groove' noise significantly mars quieter passages, then a mild application of automatic removal might be applied.  Note that the intent here is not to completely remove all trace of such noise, as that is not likely to be possible without tearing the guts out of the sound, rather just to reduce it to an acceptable, less obtrusive, level.
   6. If turntable rumble significantly mars quieter passages, then most software can analyse the frequencies from the lead in and lead out samples that hopefully were saved earlier, and use this as a basis for a filter, which again is applied just sufficiently to reduce such noise to an acceptable, less obtrusive, level.
9. If you haven't already, split the CD into tracks.  With recordings where individual tracks run into each other seamlessly when playing from start to finish, this will only happen on CD if track boundaries coincide with sector boundaries, and either your wave editing or your burning software must be able to arrange this for you.  In DCart, you select where the track boundaries should ideally go using markers, then choose CD-Prep, Quantize for CD Audio, whereupon the software adjusts the markers to lie on the nearest boundary, and then you split the file into tracks.  Other software will doubtless be different in its behaviour.
10. Burn a CD-RW.  Check that besides playing from start to finish correctly, the tracks start and end at the expected locations in the music, and that random choice of tracks works.
11. Burn a CD-R.  Check that besides playing from start to finish correctly, random choice of tracks works, the latter being a test of a good burn.  Failure of this test may be a sign of poor quality CD-Rs