Lots of people re-discovering or discovering Vinyl
nowadays want to know how to convert Vinyl to CD?
So how do we convert your Vinyl LP, 45, Cassette Tape and Shellac 78s to CD? By the use of Audio Restoration is the simple
answer.
Anything less than full Audio Restoration will not realise the full potential of
your precious media.
The actual Audio Restoration process involves at least 12 separate processes
each individually optimised for the project in hand. A typical LP restoration
takes 6-8 hours' using professional audio restoration tools and is the only
way to provide the level of service offered by OLDIES
made GOODIES.
To obtain the best initial transfer from analog to digital is crucial as it will
ultimately affect the finished audio restoration.
To restore Vinyl or Cassette Tapes to a 70% standard (or less) is fairly easy to
do yourself using automated techniques and some of the reasonably priced or free
Software available. The last 30% requires much more time, effort and experience
- and this is the basis of the service offered.
The audio restoration process inevitably leads to losses in audio so a certain
amount of re-mastering is required to bring the audio back to the original
sound.
Re-mastering can also be benefical in improving the audio of Cassette Tapes,
which by their nature are lossy at high frequencies and tend to "smear" the
lower frequencies.
To hear what to expect from a typical restoration, have a listen to this sample
of a 1944 Jazz recording, released in the UK on a mid-50's mono LP.
It probably sums it up better than any words.
The track changes from the initial transfer to the restoration about every 36
seconds.
On older records the equalisation curve used can vary significantly from the
"standard" RIAA curve so an additional pre-amp is used to supplement the
standard RIAA Phono Pre-amplifier. This is essential as many 78s' and quite a
few early vinyl LPs, EPS' and 45rpm didn't follow the RIAA curve.
This arrangement gives a choice of 14 bass turnover curves and 11 treble
roll-off curves, in any combination.
It may seem a little extreme for a bit of black vinyl or shellac but it's the
only way to do your vinyl and shellac the full justice it deserves. A piece of
overpriced resonating plastic with a poor pre-amp and a USB lead just doesn't
hack it I'm afraid.
Regardless of the quality of the transfer it becomes meaningless unless the
soundcard is up to the job, particularly the ADC converters within the
soundcard. Unfortunately, many PC soundcards are configured for gaming and audio
quality seems to come a poor second.
So, the final link in the analogue to digital transfer chain is a professional
audio quality soundcard capable of sampling at up to 192kHz - if required.
It's easy to discount the quality of Vinyl recordings as being inferior - use
mediocre equipment and you won't hear what it has to offer. I suppose the plus
side is that, if the detail is missing it's easier to restore?
If you have any doubts about the level of quality that can be achieved with your
project why not take advantage of the "TRY BEFORE
BUY" policy offered by our service? - for the cost of return
postage - see Cost of Transfer page for details.
Audio Restoration.
To get the best from your media means that all the processing has to be tailored
individually to that particular source. Running the media through a couple of
presets in an hour won't do it justice. It will normally give a result that is
over processed or under processed - lots of examples dotted round the net of
"warbling" or "birdy" effects when overcooked, some quite legitimate claims that
"To reduce the noise further would remove some of the audio" - some of it's tosh
and probably a case of working down to a price point.
It's important to remember that - the unwanted artefacts are buried in the
wanted audio content, so the approach in optimising the final result varies
continually.
So, the level of the artefacts compared to the level of the audio will vary
considerably, possibly from track to track. An LP, for example, may be composed
of similar levels of audio but may have been played more on one side than the
other, or perhaps favourite tracks played and others not? All will have an
effect on general wear and tear and could result in a different approach to
restoration. Maybe the kids have kicked the LP round the floor for a while, so
that one side is badly damaged and the other side not? - happens believe me.
To illustrate this I 've deliberately chosen a solo piano piece that has a very
low noise floor in which the audio drops down to the level of the noise in
places. The piano itself is one of the more difficult instruments to restore due
to the long sustained decay on many of the notes and is pretty unforgiving in
the "wow and flutter" department. The comments on the restoration are by the
nature of the beast, relevant to all restorations - regardless of genre.
The track is taken from a Decca LXT ffrr LP made in 1954 Liszt Recital #2 played
by Peter Katin. I've used the full track to illustrate all the problems within
it. To give a better idea of whats going on I 've also upped the bit rate to
160kbps - so it may take a few seconds to download.
Step 1 - Decisions, decisions.
My standard transfers are at 48kHz 24-bit, unless requested otherwise. All the
software operates at either 24-bit or 32-bit float so it's kept at that until
the restoration is complete to utilise fully the detail present. There is plenty
of discussion on the net regarding the benefits of higher sample rates but my
own personal view is that 48kHz is about the optimum for this type of work.
These old ears and eyeballs can hear and see the advantages of 48kHz 24-bit over
standard 44.1kHz 16-bit but not at 96kHz or 192kHz. I see and hear the same sort
of differences between high quality MP3 and FLAC for example.
The LP is certainly pre-RIAA standard curve, so I've used the Decca ffrr curve.
I transfer all mono recordings in "stereo". This has several advantages
when deciding on how best to approach the restoration. Wherever possible I use
the "stereo" transfer as this maintains the imagery of the original - precisely.
1 - One channel may be severely damaged so the least damaged file is used as
dual mono.
2 - Mixing L and R channels will cancel out much of the crackle and phase
differences but will tend to take the edge off of some of the wanted audio - not
the end of the world if you correct it at a later stage by adding a few
harmonics with an exciter, for example. Worth mentioning here that excessive
sibilance on a stereo LP can be reduced/removed by reducing the stereo image
slightly, usually in conjunction with a de-esser.
3 - If things are really bad on L and R channels there is the opportunity to
"mix 'n match" sections of both channels, by using sections of both - long and
time-consuming = more expensive.
So what are we starting with?
Relatively high surface noise and crackle, compared to the audio - particularly
at the end of the track. The "hiss" component of the noise varies with the
playing - just to make it more interesting?
Some "scrape" noise due to sidewall damage - possibly worn stylus or incorrectly
set up? It may also have been caused by "non-fill" during pressing?
Various degrees of clicks.
Have a listen - the text isn't enough. The track is around 5mins. so may take a
few seconds to load.
(May take a few seconds to load)
Step 2 - The right track?
To be more precise - the left track. After lookig and listening carefully to the
initial transfer I decided to disregard the right track due to the complexity of
some of the clicks - so I'm working from a dual mono left track, initally. Much
later I will create some imagery back into it.
The next task, with this particular track, was to start and remove some of the
clicks. There are no perfect de-clickers out there (regardless of the hype) -
and all are influenced to a lesser or greater degree by the audio content. I
have accumulated four different de-clickers over a period of time - all good,
none perfect. I also assess whatever's new on the market and have not found any
that offers an improvement.
Forget presets, it's very important to get it right first time or it will cause
irreversible problems later on!!
It takes time to optimise the de-clicking at different sections of the track.
Once done, have a listen to the complete track, Chances are that you've
introduced other undesireables. Listen particularly to the low to low-mid
frequencies for distortion, graininess and coarseness. If it happens back off
the de-clicker, to recover the audio, and accept that you won't remove all the
clicks automatically. Deal with the residue by manually de-clicking, more time,
more expense, but essential if you want the best possible restoration.
All clicks aren't born equal. If you zoom in on any click the leading and
trailing edges are different. It may be that a different result can be achieved
by reversing the track?
Once the de-clicker was set I found I got a slightly better result from
reversing the track, so this was done for the initial de-clicking and then the
track reversed again to bring it back to normal. The next job was to deal with
the clicks left over by removing them manually. This involves zooming in on each
individual click and either repairing it or in some cases removing it
completely. Repairing the click can be done by using some form of waveform
restoring tool or by a freehand wave drawing tool - the first option being the
quickest. Sometimes if the click is large in amplitude but small in time e.g. a
few milliseconds, it may be better to delete the click and apply a repair to the
resulting "joint"?
I also used a de-crackling tool set very conservatively, to reduce the light
crackly bits.
Long way to go but making a bit of progress.
Still some crackle and the odd click and all of the noise plus a few other
artefacts that weren't obvious under the clicks. Case of the deeper you go and
the more you hear - good and bad!!
Have a listen
(May take a few seconds to load)
Step 3 - Listen to de noise (really should stay off the lager).
The next stage is to tackle the surface noise and vinyl "roar". Due to the
nature of the piece it's not possible to remove the noise in one pass. The aim
is to take out the majority of it and come back to the rest later. Of the three
noise reduction tools I have available the one of choice is pretty good but like
all, needs time spent on setting it up.
It's made more difficult, in this case, by the low level audio and sustained
decay on the piano. To help me, I'm able to control not only the amount of noise
reduction but the way it's applied (attack and release) and the bands of
frequencies it's applied to. To get me in the right ballpark a "noiseprint" is
taken, in this case, from the end of the track.
Why not the beginning?
Usually the lead-in gets most abuse and damage so doesn't really represent the
rest of the record - always try and leave a couple of seconds "silence" at the
end of the track for this purpose.
As ever, revealing more of the audio highlights problems that were hiding in the
"mire" before. These can be many things, faults in the original pressing,
degradation of the vinyl, wear etc.
Most will have to be dealt with manually, using a variety of techniques - long
and laborious, but no other way I'm afraid.
At this stage I'm settling for removing the majority of the noise. The downside
is that in doing so it will effect the audio to some extent. In this example
I've rolled off the frequencies up to around 40Hz and to a lesser degree up to
around 100Hz. Getting the hiss component down has resulted in taking the edge
off the high frequencies and a bit of a trade-off with the decay down towards
the noise flloor. To rectify this I've added a couple of dB of equalisation at
147Hz using a reasonably wide parametric equaliser at the other end I've added a
few harmonics from around 2kHz using an "exciter".
The "scrape" noise is still there but reduced and will be dealt with in the next
stage.
The tracks now beginning to take shape - Have a listen.
(May take a few seconds to load)
Step 4 - Listen to the Music.
The final stage is pretty time-consuming as it finalises the restoration.
First I need to reduce the noise even further which is exactly the same routine
as before but at a less invasive level.
Next job is that annoying "scrape". Big problem, as although it all sounds the
same it is fairly wideband and does in fact vary quite a bit in frequency for
each instance, as does the width of the band. Obviously, the wanted audio
frequencies surrounding it vary so this too will have an effect on how it is
reduced or removed - not easy.
The basic tools used for this were a de-esser in conjunction with a parametric
equaliser. All the settings on each tool had to be tweaked for each section of
offending "scrape" to minimise the effect on the surrounding audio at that point
- pretty time-consuming.
More time was taken to carefully scrutinise the track, a few seconds at a time,
and deal with any artefacts left over - wide variety of artefacts and a wide
variety of techniques - bit outside the scope of this illustration.
Once de-noised again, there was a slight residual noise which coincided with a
slight "ringing" on the high notes. This is not uncommon and tends to be in the
7 to 7.3kHz range, usually nearer the bottom of the range. No idea what causes
it, there on some records not on others - whether it relates to the quality of
the Vinyl - dunno, as yet?
The parametric was utilised again, set to the spot frequency and at a very tight
"Q", the gain reduced to get rid of the problem.
Almost there!!
I've worked on a dual mono track until know, but no mono record is truly mono,
all have some sort of imagery - not to be confused with stereo, but there is
some movement around the null point between the channels.
To maintain the integrity of the track I've widened the image to bring it back
in line with the original.
Losing all the noise etc. has effectively increased the overall dynamic range
and vinyl was generally cut with a lower overall "loudness" compared to CD. So
without going silly and getting into the "compression wars" a small amount is
added to the track(s) to compensate and then the L and R channels normalised
independently to correct the balance between the two channels.
At the end of the track I've used a non-linear fade out which gradually
increases towards the end - helps maintain the existing fade for as long as
possible.
At the start of the track I've zoomed in onto the leading edge of the first note
and moved back a few tenths of a second then deleted everything before it.
Next is to convert the sample rate down to 44.1kHz and apply the dithering
weapon of your choice to the file to bring it down to 16-bit.
Dithering introduces noise at a low level so I've zoomed back in on the first
note, made a selection back from that point and applied silence to the few
tenths of a second preceding the first note.
Job done - Several hours' on I can finally sit back and enjoy a fine performance
by a fine Pianist without listening to the faults - yes!! it does take that
amount of time to work on this type of project at this level.
Here it is - ENJOY!!
(May take a few seconds to load)
Please Note: - It's not possible to restore this type of Project to this
standard at the standard rates - too much time involved and this would have to
be taken into account.
Distortion is probably one of the most difficult artefacts to deal
with. Many of the used Vinyl Records available nowadays seem to have some degree
of distortion. EPs/45 rpm seem to be more prone to it than LPs'?
There are a variety of techniques that can be used to either alleviate or remove
it - all are time-consuming, which has to be reflected in the pricing of the
restoration.
The main culprits are dust and wear, usually a combination of both. Cleaning
will take care of the dust but can do little for the wear. The range of
techniques used are outside the scope of this article but the following sample
taken from a UK 1957 Teresa Brewer track illustrates the effect quite nicely.
I've deliberately cut about 36secs. out of it, so if the lyrics get a bit
confusing - that's why. The content left in is more than sufficient to hear the
differences. (Before Restoration) (After Restoration)
Surface Noise is on all Records to a greater or lesser degree. In
addition there can be audible mains hum and "vinyl roar" on top. When removing
all this hash allowance has to be made to the type and content of the wanted
audio. For example the audio may be spoken word with many pauses? The audio may
be at very low level, as in many Classical pieces? There may be a long fade at
the start or end? All will have varying levels of attack and decay on every
note.
Particular attention has to be given to sounds such as brushstrokes on a snare
drum, for example. Heavy handed noise reduction = no brushstrokes!!
Some favourite tracks may have considerably more wear than others - it's not
uncommon to find this on an LP.
All these factors have to be considered when removing/reducing the overall
background noise - to get the best from the audio restoration it is impossible
to just run the tracks through a preset noise filter - many times totally
different settings have to be used within the same record!!
The sample below, from a 1956 Decca 45rpm record illustrates part of this.
After cleaning the track was in pretty good nick as regards surface noise and
clicks but there are, in parts, some very low level brushstrokes to retain. The
record was cleaned three times, as I was initially disappointed with the level
of clicks remaining, but eventually had to recognise that the remaining clicks
and crackle were embedded in the groove forever. I've shortened the clip
by 26 seconds. (Before Restoration) (After Restoration)
Shellac 78s' can have a high level of surface noise but even with
this all is not lost - it is possible to restore many of them to a satisfactory
standard, using the right Software and lots of patience and time. Some of the
later 78s' do have a wide frequency response and are comparable with LP quality,
so well worth restoring and transferring to CD.
Have a listen to these clips from a Robert Earl early 50s' recording - high
surface noise, brushed snare, powerful vocals, string backing - from an audio
restoration point of view, not a lot going for it, or just maybe?
As with all 78s and some of the earlier 45s and LPs, care must be taken in
establishing the equalisation curve used. I well remember many years' ago
listening to some of my Elvis LPs and thinking that they sounded rubbish only to
realise years' later that they didn't conform to the standard RIAA curve - sound
great now!!
- Click on play and have a listen (clip goes from original to restored
version).
Many 78s' have suffered badly over the years' and can look quite good visually.
The truth is that until you play them it's difficult to know what can be
achieved?
Some on the other hand look poor and don't disappoint once played. This doesn't
necessarily mean they aren't restorable it means that it will require much time
and effort.
This Shellac 78 is a good example of a record bought at a Car Boot sale for one
pence and my first impression was that I'd been robbed?
Very little gloss left on it, lots of surface scratches with a few heavier
"dinks" and no sleeve.
The good news was that I bought it as a "practice" piece and wanted an example
in very poor condition to test the limits of Audio Restoration?
Have a listen!! (Before Restoration) (After Restoration)
If you've struggled this far down the page and are still with me, I'd like to
thank you for your time.
I hope it has given at least an insight into what's involved during restoration.
If you have any queries, mail me.
All restorations are done using a professional Audio Editor which allows changes
to be made in real-time which in turn makes it ideal to observe and hear the
effects of changes whilst the track is being played - A & B comparisons can be
made on-the-fly.
If you've searched this page looking for a cheap, quick and easy way of
transferring your own Vinyl to CD quality then I'm sorry - it doesn't exist!!!!
A typical LP Audio Restoration, to this standard and quality, will take 6-8
hours'. so is it expensive at the prices quoted? Work out the figures?
How do others charge less? You would have to ask them - or better still compare
the results?
Thanks go out to Joan Harper
for the use of the Dave Lee Sound clip on this page - a true Rockin' Pianoman -
and sadly missed!!!!
