Editorial - Diary of a remastering session

From 78s to CD - a step by step guide to a Pristine release 

This week's release of American piano music has been, happily, one of the more straightforward 78rpm restorations I've carried out recently, so as I neared the end of the final restoration process on Tuesday of this week I decided to start writing descriptive notes on the process for this column,  while everything was still fresh in my mind:


My remastering work this week got off to a less than brilliant start. On Monday I was suffering from a stinking cold - never a good idea when you need your ears at peak condition. I spent the day cleaning and transferring 78s on which I began a couple of restorations, but which I ultimately decided to leave to one side for a future week and a clearer head.

Fortunately Tuesday sees me with nothing more than a sore throat and a bit of a cough (it's just that time of year - kids back to school means loads of germs floating around). I'd spent a chunk of Monday unpacking more boxes of Al Schlachtmeyer's 78s (the fact I've still not finished may give some impression of his incredible generosity), and a couple of sets had caught my eye.

We've been putting out a good bit of American music over the last few months, the vast majority of which is entirely new to me - as often are the composers concerned. So when I found an album entitled "Piano Music by American Composers" my only regret was that there were only eight 12-inch sides of it. Then I stumbled across MacDowell's Marionettes Suite, as performed by Rudolph Ganz (and apparently still the only recorded performance of the work), on a pair of 10-inch US Decca 78s, and decided there was enough to create a release, albeit a shorter one than I'd like.

I always begin with the cleaning and transfer of the discs. Each side is wet-cleaned using Disc Doctor fluid and rinsed with medical-grade purified water, which is sucked dry by a VPI 16.5 record cleaning vacuum machine.

I use a Rega Planar 78 record deck for 78rpm transfers. I've had it a number of years and it's a good, reliable belt-driven single-speed deck. I have a set of around a dozen different stylii to cover most eventualities; often one particular stylus will allow you a quieter or less swishy replay than another that's slightly wider or narrower - we're talking two or three 10000ths of an inch difference between them, and it can make or break a transfer.

This time I began with the two short Decca discs. Since developing XR I've taken to carrying out flat transfers and equalising them later. I admit I was also being a little lazy - 99 times out of 100 the best transfers - quietest, least swishy, cleanest - are done with my trusty 3.1 stylus (thousandths of an inch), and as these sides sounded fine going in I didn't check for other stylus sizes. This later turned out to be a mistake.

Anyway, having recorded the output using my new RME sound card to digitise the analogue output of my Graham Slee Jazz Club pre-amp, I saved the file and got to work on it. The first two processes were a standard declick and decrackle - the declicking routine takes about a fifth of the overall audio duration, the decrackle is a little quicker. The file at this stage is still "stereo" - meaning you've got the individual clicks from each side of the groove as a separate channel, which generally gives better results when declicking. The declicked file is then mixed into mono, which has the effect of boosting the signal-to-noise ratio by about 6dB - a useful amount - and phase cancellation can also sometimes reduce or remove remaining clicks and crackles.

Then it's over to Capstan, my new pitch stabilisation software. This loads the audio file, analyses it, and offers various options to smooth out any detected wow and flutter, as well as repitching, if required, to a specific frequency. In this case it told me that the average frequency here was A=440Hz, and I'm happy to stick with that!

The next step is XR equalisation. Normally this involves matching up an average frequency response of the recording in question to a reference recording of the same work. This is, however, a big problem when you don't have a modern recording of the same piece, and I tried several ideas to try and get around this. The closest to satisfactory was to create a very big piano music recording, made up of lots of different piece of music (all pitched to A440) to try and make a generic "piano" profile. But it didn't work for me. Something just didn't sound right. I decided to put the MacDowell to one side and get on with cleaning and transferring the other records while I pondered how to proceed.

As is often the case, the next set of discs suggested the best way forward. The first side suffered a distinct swish with my 3.1 stylus. But when I swapped it for a 3.5 (yes, just 4/10,000ths of an inch wider) the swish virtually disappeared (I checked with a 2.8 and it got worse, as did the surface hiss). Clearly the bigger stylus was needed here. Then I got thinking about that equalisation problem, and I starting playing with the various settings on my pre-amp.

The Jazz Club pre-amp is one of a small handful available today which allow you to select different equalisation curves for treble and bass. Before RIAA standardisation in the mid 1950s, all sorts of different settings were used by different record companies when mastering records. The bass was cut and treble boosted artificially in order to work in the confined space of a record groove - these alterations then need to be reversed on replay to regenerate a 'flat' frequency response. The Jazz Club allows you to alter the "turnover" frequency at the bass end (the point at which a standard roll-off begins) and adjust the degree of treble cut - UK pressings generally used a flat treble response, but some US pressings require a massive 16dB-per-octave reduction to sound right - and there are a number of other settings in between that have been used by different companies over the years. These days the standard is 13.7dB.

After some experimentation and careful listening I felt that a 500Hz bass turnover with an 11dB treble roll-off sounded about right for these recordings, and so I got on with my transfers. I went through the same process outlined above for the four twelve inch Victor discs - and when it came to equalisation I found that when I looked at the overall frequency profile of these recordings against my generic 'piano' they looked very similar indeed. I'd got it about right - and no further XR-style equalisation seemed to improve the sound. Again I had no "proper" reference for these rare recordings.

At this stage I decided to return to the two Decca discs. This time there could be no short cuts. I made a series of test recordings and ran through initial processing on each of them before running A/B/C comparisons of each. Once again the slightly wider stylus won out, rather than the 3.1 I'd initially used. Again I worked with the EQ settings on the pre-amp until I was happy that I was getting the best possible sound. I then repeated the entire process of transferring, declicking and pitch stabilisation before checking the results against my piano reference. As with the Victor discs I decided no further XR-style re-equalisation made any improvement, at least not with the rather improvised and less than perfect reference I had to hand. So now I had all the building blocks for the final assembly ready.

The next stage in the process for both transfers was pretty much the same. Using Izotope's RX2 audio restoration software I looked for remnants of mains hum and any other regular interference noise. It was no great surprise to find some US-standard 60Hz floating around in both recordings, and this was easily zapped by a very fine digital notch filter. I also rolled off ultra low frequency turntable rumble. Then I checked the overall frequency response to see how much top end was present - a piano is often surprisingly lacking in very high frequencies unless played loudly, and these were of course frequency-limited 78rpm recordings as well. I was able thus to gently roll off some very high frequency hiss without eating into any musical content using a simple equalisation curve - often a 'nicer' way of achieving lower noise levels than simply applying digital noise reduction, thought it's an entirely indiscriminate method, hence the frequency checks first.

Then comes the digital noise reduction, often carried out in two or three passes for each recording, where a general pass is followed by frequency-range-specific passes to tackle particular problems. Several gentle passes usually do a much better job than a single, cure-all noise reduction. Very high resolution noise reduction on a mono file typically takes a little over half the duration of that file - double that if you're working on a stereo file.

Finally we reach what I call the 'mixing' stage. Often there isn't actually any mixing to do, but with many longer 78rpm recordings this is where side joins take place, and my procedure is always the same, whatever the material. I assemble what will be the finished release, in playback order, in a multitrack editor in Adobe Audition.

I first began experimenting with side joins of the MacDowell suite, but soon abandoned the approach - thanks to a huge drop off of treble hiss by the end of the sides the transitions from one side to the next sounded pretty awful. I elected therefore to "fade to black" between movements rather than spend ages trying to contrive something fake in the joins that would only add noise to the recording.

At this point I also convert all the music into Ambient Stereo (unless it's already true stereo) and finalise any last-minute changes to equalisation and the like, mainly using Izotope's Ozone 3 mastering suite. If any convolution reverberation is to be added (as I've discussed in previous columns) this is when it would be used. I want to be able to hear pretty much the finished master sound at this stage, and in the case of these piano recordings, a very small amount of convolution reverb was indeed implemented. (The Ambient Stereo processing, by the way, is entirely self-cancelling if the two resultant audio channels are mixed back together - this is how we make our final mono issues, which are identical in every respect to a recording where the processing had never been brought to bear.)

Now comes the real work! At this stage I may have a finished sound, but I can still be hours or days away from the final release master, depending on the source material. This is when I get down to the nitty-gritty of going through every bar of music, note by note, looking for stray clicks, swishes, clunks, bumps, clangs, whistles, motor vehicles, sirens, aeroplanes... You name it, I've probably removed it from a recording at some point. Each stray noise has to be selected on screen using the mouse and then either zapped or attenuated using one of an armoury of restoration tools.

Sometimes it'll take several goes to successfully remove a noise - a history file in RX2 allows me to go back several steps and hear the before and after effect of my efforts to make sure I've not nipped out something I'd rather keep. It's painstaking work, but this set of discs is providing few major difficulties - as I type this I'm at the halfway stage and I've (only) made 1054 individual manual interventions since beginning this final restoration phase - and I feel I'm sailing through at several times the rate I'd expect from a Schnabel Beethoven Piano Sonatas restoration!

By the end of today I hope to have a completed remastering that's pretty much ready to go. I'm adding in and out fades and silent gaps as I go - the durations of the latter will be finely-tweaked later. Tomorrow I'll listen through and check this, then get to work preparing Mark Obert-Thorn's Zimbalist release, also for this week - he sent me the finished files electronically last week but wants a minor level adjustment to some of the tracks, and has highlighted a couple of issues inherent in his source material that I need to be across - occasionally I might have something hidden away in my restoration armoury that can fix a problem he's struggled with (and vice versa, of course).

Thursday will see artwork, notes, download and CD files and so forth prepared and the download files transferred to our servers, then on Friday comes the website update and the newsletter.

And of course during all this I'll also be working on next week's releases, opening the remaining crates of 78s, answering e-mails, offering technical support, updating PADA, assembling an up-to-date Digital Music Collection drive, researching background information, contributing to discussion forums, testing new software, and generally keeping the show on the road whilst wondering what I might write about in next week's column!


(For the record, the final count of those manual interventions - not including those tried and abandoned, only those what made it to the final 'cut' - was 1743. That's an average of one for every 1.61 seconds of audio. It's one reason why I rarely "hear" the performances, rather than the recording flaws, when I'm working - each intervention requires me to stop, make a selection on screen, select a remedy and apply it, listen to the result, correct or improve it if necessary, before I can move on. As I said, the American piano music CD was indeed pretty straightforward and easy to complete by comparison to the likes of Schnabel and others.)


Footnotes on EU Copyright changes

Further to last week's lengthy editorial, I can report that not only has the European Commission thus far failed to correct the headline "facts" on its website concerning the effect of the new laws when applied to recordings such as radio broadcasts, and I've had no response to my contact on this specific matter from either the commissioner or his spokesperson.

I did read several newspaper articles and their follow-up posts, which suggested that lobbying by parts of the music industry had been successful in implanting the idea into gullible minds that these changes were all about protecting the "poor musician" rather than lining major record company pockets.

This all got me thinking. And I felt that a much better change of copyright law if you really wanted to help the "poor musician" could have been proposed. Such as this:

1. Keep the 50 year rule, but split it in half, so:
2. The record company gets the first 25 years
3. The performers get the second 25 years

 

Say the average age of pop music performers is somewhere around 25 - this means when they're 50 and still of an age to do something about it they could disinter their long-deleted albums from the record label vaults and release them for themselves (this would account for the vast majority of 25+ year old recordings, which I would strongly expect aren't currently available).

 

Of course if the musicians think their record company could do a better job of sales and marketing for the next 25 years they could license the recordings back to them - give them first bite at the cherry, and an incentive to promote and sell them. 

 

And assuming they were 25 when the recording was first issued, they'll have until they're 75 to get the most from it, and prepare pension funds and the like - including a good 25 years where their hands haven't been tied by a dodgy contract they signed, wide-eyed, back in the 70s.

 

It strikes me that I've met so many pop musicians over the last few years who'd been screwed over by record contracts, or had albums they couldn't reissue and their record companies wouldn't put out, or were wrestling with contracts written in any time over the last 50 years where the labels wouldn't even send a reply to simple letters of enquiry, and so on.

 

Some of these are people active, gigging musicians with often significant followings, who could easily make a little extra at a concert if they had their own "classic" CDs to sell at the back of the hall - but a combination of copyright law and record company indifference or intransigence means they can't gain any access their own work.

 

By the time the real new law "allows" them this option (and it's most definitely stacked against their being able to claim it) some of them - those that are still with us - will be well into their 70s and 80s, their gigging days almost certainly long behind them. These performers and creators ought to be able to use their own hard work and talents to create a future pension right now, but the law balance remains firmly tipped against them, and decidedly in favour of the major record companies.

 

I do know of at least one minor 70s star who's taken illegally to selling his old albums online on CDs he's had made himself. They sound infinitely better than the terrible job his then-major-ish record label did when they were issued on vinyl, and he's had a great response from fans around the world who thought he'd disappeared in 1976. How do I know all this? I remastered them from his own master tapes for him (and with him) - just don't tell his old record company. Fortunately he can now afford to be in it for the love rather than the money - but that's no thanks to royalties from his musical career, alas. All he wants is for those who've stuck with him to be able to hear his music as he'd intended it and created it some 35 and more years ago. 

 

 

 

 

Andrew Rose
23 September 2011