Pressing Matters

It is my sincere hope that this blog provides entries of interest to music lovers, musicians, and audio enthusiasts, as well as folks who make records.  A few previous entries, such as Can you hear what you’re doing? (Part 1) and Can you hear what you’re doing (Part 2), were aimed specifically at those setting up studios in order to make records, among whom there are a great many musicians.  Of course, it has been my hope that others would find these of interest as well.  So it is with the current entry.  While it is intended primarily for those who make records, if I’m lucky, those who purchase and listen to records will also find something of value herein.

With audio mastering completed for the new Work of Art album entitled Winds of Change (first mentioned in the August 22, 2014 entry of this blog, also called Winds of Change), and with the album artwork approved, it was time to contact the CD replicator in order to get the “pressed” versions manufactured.  Actually, CDs are not pressed like vinyl records.  They are made using an injection molding process, but the term pressing seems to have endured in common use.

Those familiar with my label, Soundkeeper Recordings, know that we release each album in several different formats.  In addition to the regular CD, we offer six custom burned formats, including CD-R and five formats with higher resolution than a CD can provide:
–   Music-only DVD-R with 24-bit, 96 kHz audio, playable in most regular DVD players
–   24-bit, 96 kHz .aif files-on-disc
–   24-bit, 96 kHz .wav files-on-disc
–   24-bit, 192 kHz .aif files-on-disc
–   24-bit, 192 kHz .wav files-on-disc

For more about the different resolutions, see the May 22, 2014 entry in this blog, Is “too much” not enough?

As far as standard, 16-bit, 44.1 kHz CD resolution, the reason Soundkeeper Recordings offers our releases in CD-R format, and the true subject of this entry, is something I’ve said since I heard the finished product for the very first CD I mastered, back in January of 1983—CDs made at different plants all sound different from each other and none sounds indistinguishable from the master used to make it.  This may sound strange in view of the marketing that has accompanied the CD format from the beginning, primarily in the form of the slogan “Perfect Sound Forever” and the widely accepted idea that nothing can change once the signal is in a digital format.

Imagine my surprise then, when I first started mastering CDs and found that the same digital U-Matic tape (the format used at the time to send CD masters to replication facilities) sounded different depending on which side of the Sony DAE-1100 editor I used to play it.  The DAE-1100 was commonly used at the time to assemble CD masters.  The unit controlled tape machines for the ¾” tape cartridges that comprised the U-Matic format.  One or two machines could be used on the Playback side and another machine was used on the Record side.  The CD master was assembled on a U-Matic tape in the machine connected to the Record side of the editor.

Early on in my experience with this system, I wanted to compare a tape that was copied from another, just to hear for myself that a digital copy was indistinguishable from the original, as we’d all been told.  Unfortunately, the test never got that far.  What I found was that the original tape, played from the Playback side of the editor sounded better than the copy.  Something was getting lost on the copy, as it seemed coarser and less well defined than the original.  I don’t recall what made me try it but I decided to swap the tapes, listening to the copy from the machine attached to Playback side of the editor and the original from the machine attached to the Record side.  To my surprise, now the copy sounded better (i.e., more like the analog source tape I was using) than the original digital conversion.  When heard from the Record side of the editor, the original digital tape now sounded coarser and less well defined than the copy!  Clearly, there was something else going on.

Perhaps it was this experience that diminished the surprise when the finished CDs for that first CD mastering project came in and I compared them with the CD master used to make those discs.  Here the coarseness was even greater than what I’d encountered on the different sides of the DAE-1100 editor.  The finished CDs almost sounded “out of focus” compared with the CD master, such was the extent of the loss of clarity and fine detail.

Things got more interesting when I created CD masters for albums where large sales were expected.  In those days, there were fewer CD plants than there are today and they were all working at capacity.  In order to accommodate expected demand for the big sellers, the CD master would be cloned and those clones were sent to multiple replicators in order to get back sufficient numbers of finished discs to meet the demand.  This was an education in that I discovered that CDs from different plants all sounded different from each other.  Sometimes CDs from different lines within the same plant sounded different from each other.

So much for “Perfect Sound Forever”.  The format has been claimed to deliver perfect copies of the master.  Logic would demand that if this is the case, all those perfect copies would sound indistinguishable from each other and they’d all sound indistinguishable from the masters from which they were made.  But they weren’t then and they still aren’t today.  (There is an exception that I’ll get to shortly.)

Having sent CD masters to plants all over the world and all over the USA, I’ve had the opportunity to compare a lot of finished product to the masters from which said product was made.  Happily, the days of U-Matic tapes are long gone and the advent (long ago) of computer workstations made for many improvements.  Still, even with the most sophisticated CD mastering tools, the reality from the replication facilities remains—the finished discs don’t sound like the masters.

In my experience, a slow-burned CD-R made directly from the computer-based CD master, sounds more like that master than any pressed CD, even the best in my experience.  This is why Soundkeeper Recordings offers our releases in CD-R format as well as replicated CDs.  But how then, to select a CD replicator?  If they all produce discs that sound different from the CD master, how does one find the most faithfully made discs?  This is the question that was on my mind when I started the label.  Knowing that a lot of folks just prefer a factory-made disc to a burned version, even if the latter is more faithful to the master, I needed to find a replicator for Soundkeeper CDs.  My whole reason for starting this label was to avoid the compromises I feel are too often part of the record making process.  I wanted a no-compromise replicator — if such a thing existed.

I reached out to contacts at most of the plants I’d sent masters to over the years.  I told them about my concept for the label and that I needed the most faithful to the master, highest quality discs.  All but one of them told me essentially the same thing.  They said their CDs were perfect replicas of the CD master.  Since my own experience consistently told me something quite different, I could only conclude they were not hearing it the same way I was.  Or they just weren’t listening and were simply repeating the received mantra.  I thanked each in turn and moved on to the next person on my list.

Out of all the replication facilities, only one person at one facility told me, with no prompting from me whatsoever, “Oh the finished CDs will never sound like the CD master.”  I wanted to hear more but knew by then that I’d found my CD replicator.  Here at last, was someone who appeared to actually be listening.  It turned out, this replicator took an unusual approach to making their finished CDs too.  Where many plants increased their throughput – and hence, their income – by speeding up the process, this plant kept with the slower methods.

The first step in manufacturing a CD involves cutting what is called the glass master.  The CD master from the mastering facility is fed into a Laser Beam Recorder (LBR), where a laser is used to create the pits in a photoresistant coating on a glass disc.  This disc is used in the subsequent steps of CD manufacturing.  Most plants cut the glass master at high speed.  Some will cut the glass master in real time, at additional cost.  Many folks have found real-time glass cutting to result in finished discs that sound closer to the original CD master.  The person at this plant told me they cut all their glass in real time, at no additional cost.  It is just how they do it.

In addition, most CD replicators have moved to shorter injection molding cycles.  The faster the cycle, the more finished discs that can be produced in a given day.  Typical injection molding cycles for CDs are now about 4 seconds long.  The person at this plant told me they use a slower cycle, closer to 9 seconds long.  This makes for better formed pits on the finished discs, making it easier for the laser in the CD player to read the discs and minimizing the incidence of playback errors.

Whether the real-time glass cutting and slower injection molding cycle are the reasons or whether some other factors might be involved, I don’t know.  What I do know is that when I master an album, I listen to it so many times that I become intimately familiar with all the details of its sound.  Often, when I hear the finished CD that comes back from the replicator, it takes only a few seconds to hear the typical loss of focus and fine detail.  Something like a chord strummed on an acoustic guitar becomes a loose mélange rather than the six discrete, individual string sounds heard on the CD master.  With CDs from this replicator, the sound is so much closer to the CD master, I need to synchronize playback of the finished disc with the CD master in order to discern the remaining differences.  (Still not as close to the master as the CD-R but closer by far than I’ve heard from other CD plants.)

Now earlier on in this entry, I mentioned an exception.  In fact, I wrote about this in the February 23, 2014 entry in this blog, entitled Listening to Tomorrow.  Basically, what I’ve found is that what I’ve written about in the current entry comes into play when the CD is played in a CD player or via a CD transport.  This has been my experience regardless of the player or transport, or its price.  However, when the CD is properly extracted to a computer, the audible differences do go away.  To date, after 31 years of the CD format, it is only via computer that I’ve heard the audio from a CD disc sound indistinguishable from the master used to create that CD.  Still, those listening to computer music servers with CD or better resolution (as opposed to mp3 or other reduced formats) are in the minority.

Most of the music lovers I know of who purchase CDs will listen to them in CD players or via a separate CD transport feeding an external digital-to-analog converter (DAC).  In order to provide these folks with a CD that truly represents the CD master approved by the artist and producer, selection of the replicator is critical.  To this end, I feel very lucky to have found Bryan Kelley and the folks at GrooveHouse Records, who I have been recommending to mastering clients since my first conversation with Bryan, and who, as far as I’m concerned, are the official CD replicators for Soundkeeper Recordings.

Advertisements

Perfect Sound Forever? (Part 2)

There I was in 1984, Atlantic Records’ “CD mastering department”, responsible for creating a good portion of the masters used to replicate the monthly CD releases for the label and associated divisions (Atco, Elektra, etc.).  Demand for CD was on the increase and it was clear this was where recorded music was going.  The small CD section at the local Tower Records store was a bit larger every time I visited, slowly but surely encroaching upon the real estate that was, for the moment, dominated by vinyl LPs.  I saw customers so eager for new CDs, I got the impression even a disc of dog barks would be a hot sales item.

The manufacturers behind the format proclaimed “Perfect Sound Forever”, distortion-free music on a medium that would not wear out.  It sounded too good to be true.  Like most things that sound too good to be true, it wasn’t true.  I remember the expectation with which I first listened to digital masters and to the earliest CDs.  Despite the raves of my colleagues and those in the press, what I heard every time I listened sounded to me not like an evolutionary step forward for audio but like an electronic equivalent of fingernails on a blackboard, an irritating harshness that felt like a good deal of the music had been replaced by something unnatural, something mechanical, something cold.

A number of colleagues I spoke with did not seem to have the same experience.  In fact, they looked at me askance when I expressed great disappointment in what I’d heard, as if I was missing something so obvious, they couldn’t believe it.  They would point out how flat the frequency response measurements were, that the wow-and-flutter (a measure of speed inaccuracy) was virtually unmeasurable.  They would say “Just listen to the noise!”, amazed to have a medium that did not add any hiss.  I would respond “Just listen to the music!”

Yes, piano recordings did display a steadiness of pitch devoid of the indeterminacy sometimes engendered by analog media (played on less than great tape machines or turntables, or when either the tape was stretched or the vinyl pressing suffered a slightly off-center hole).  If any hiss was audible at all, it was the hiss from the original analog recording.  The digital medium wasn’t adding any that I could detect.  Yet, what good were rock steady speed and dead silent backgrounds when the piano sounded like it was made of aluminum?  And the cello sounded like a cousin of the kazoo?  Instrumental harmonics were bleached into thin, pale ghosts of themselves and the very air around the players (on recordings that had such) seemed to have been sucked from the room.  A great rock record invites the listener to turn up the volume.  Doing so with a rock CD just brought on the headache that much sooner.  What was wrong?

I had done everything I knew to ensure the highest possible quality.  I set up the CD mastering room with the audiophile sensibilities I sought to bring to my work.  I created CD masters bypassing most of the electronics in the room, keeping the signal path as short as possible, introducing only what was absolutely necessary and avoiding extra switches, wires, patch bays, consoles, etc.  I even took to carrying my own cables to work every day, replacing the generic studio cables connecting the output of the tape machine to the analog-to-digital converters with one of the best audiophile designs of the day, one that had repeatedly shown me it was capable of passing more of the musical information, with less degradation than the regular studio cabling.  Still, even with the CD masters created this way, a comparison with their vinyl counterparts, made using a far less purist approach, showed just how much more of the musical information on the master tape made it to the finished LP than ever made it to the CD.  There were no exceptions.  This was the case every single time.  Digital acolytes in the press attributed any favor shown the LP to euphonic (i.e., pleasant sounding) colorations in the medium, where CD was supposedly truer.  But as is often the case, the audible evidence said otherwise.  A well set up $100 turntable/cartridge combination would, in terms of bringing back the sound of the master recording, sonically wipe the floor with a $1000 CD player.

A fellow mastering engineer, one whose work I had admired for years, called one day and invited me to sit on a panel of mastering engineers to discuss CD at a meeting of the Audio Engineering Society in New York City.  I gratefully accepted and not long afterward, found myself sitting at a long table on stage in an auditorium, next to four other colleagues, all of us involved in CD mastering.  When I spoke, I felt quite alone in that my colleagues all sang the praises of the new medium while I (quite shyly at the time) said “I just don’t feel it sounds as good as my vinyl yet.”  (Yet?!?)  I explained how I felt vinyl was revealing much more of the musical information contained in the master tapes.  Despite any technical flaws or issues in manufacturing and playback, things that did not at the time seem to plague CD (at least not when one just looked at the surface of things), vinyl was providing more music and to my ears, that was more important.  When I left that evening, I thought folks were looking at me as though I had two heads.

What we came to learn as time passed and more audiophile companies got involved with digital and CD, was that a major part of that bad sound in the early days was due to the digital recording and playback gear itself, perhaps most specifically in the filtering that is an essential part of these mechanisms but also in the converter chips at their core.  I found it interesting that when folks like Bob Stuart started writing articles about jitter (timing irregularities between samples in the stream of digital data), a number of folks who had previously raved about CD (seemingly because of the “good” specifications they’d read) now found issues with the format.

With the advent of new knowledge came new filter designs and new converter chips.  The players were starting to get better.  Even the Sony 1630 converters I used in the studio got new retrofit filters that made for noticeable sonic improvements.  The CD format was growing in popularity every day and clearly was going to be around for a while.  The thought occurred that vinyl mastering engineers were routinely credited for their work on albums but no one as yet (at least to my knowledge) had been credited with CD mastering.  I spoke about this with management and after a conversation with the art department, saw the first CD booklet with my name in it.  As the format continued to grow and demand for more releases increased, outside facilities were contracted to create masters in addition to the ones that were keeping me busy full-time.  The only problem was the art department was not always informed when a master was going to be done by a third party.  As a result, some CDs I mastered did not have a credit and some CDs mastered by others have my name on them.  (In a way, I came to know whence the phrase “Be careful what you wish for” comes.)

I made some other observations regarding the digital audio of the day.  First, the playback and record sides of the Sony DAE-1100 digital audio editor did not sound the same.  The official word was that a digital tape could be cloned (“clone” being the term used to describe a digital copy) to create an identical copy.  Yet, when I cloned a digital tape and played it back to compare it with the original, the original always sounded cleaner.  Was there some degradation in the copy?  I found it interesting that when I took the tapes out of their respective machines and swapped them, putting the copy in the “playback” machine and the original in the “record” machine, the original now sounded degraded.  It turned out (for reasons I’m still not sure of) that playing back a tape from playback side of the editor just sounded better than playing the same tape from the record side.

As CD grew, we started using more and more replication facilities.  When sales for a particular release were expected to be large, often a single replicator could not produce a sufficient quantity of discs, so I’d create a CD master and then send clones of that master to different replicators.  When the discs came back, I made another discovery.  The discs from all the replicators sounded different from each other, sometimes subtly so and other times not so subtly.  And none of the discs sounded indistinguishable from the master used to make it.

It was plain to see there was much more to be learned about this digital juggernaut.  My thinking was that we’d had vinyl for about a hundred years.  In another hundred years, I expected CD would be pretty good.  Happily, it hasn’t taken nearly as long as that.  Today, CD can be “pretty good” if not exactly competitive with fine vinyl, despite what is said in some quarters.  Perfect sound forever?  Not to my ears.  It is more like “Decent sound, once in a while” but I can see how that is a bit less catchy as a marketing phrase.

Sonically, there was lots of room for digital to grow.  As futuristic as the equipment seemed at the time, it too, along with many of the very techniques involved in recording and editing, would soon undergo a revolution, as recording and mastering began to take advantage of the nascent world of desktop computing.

Perfect Sound Forever? (Part 1)

In early 1983, I created my first master for Compact Disc.  I first heard of the format nearly a decade earlier, while still in college.  I remember a promotional mock-up, looking very much like a miniature LP jacket.  Inside, was a cardboard disc printed with the distinctive rainbow reflections of the real thing.

Atlantic’s west coast affiliate, Warner Brothers, was already creating CD masters when it was decided that Atlantic would open its own CD mastering room.  I was to be the CD mastering “department” and was sent to Los Angeles to spend a few days with my counterpart, learning the procedures Warner Brothers had in place for creating CD masters.

At this point, the only CD mastering rooms I knew to exist were at Sony in Japan, Polygram in Germany, Warner in California, perhaps DADC in Terre Haute and now, Atlantic.  To my knowledge, I was one of the first engineers to do CD mastering.  Technically, the process of creating a master for CD replication is referred to as “premastering”.  To the replication facilities, the term “mastering” refers to the first stage of manufacturing, when the glass master is “cut”.  Glass mastering is the creation of a glass disc, etched by a laser beam recorder.  This disc is electroplated and used as the first part in the process that yields the injection molded finished CD.  Still, in terms of the creative process, which occurs prior to manufacturing, creating a CD master is still referred to as “mastering”.  Mastering, for any format, not just CD, has always been the last step in the creative process and also, the first in the manufacturing process.  It is the last chance to make any adjustments to the sound and it is where the “part” used to initiate manufacturing is created.

In those days, the CD master sent to the replication facility was recorded on a U-Matic video tape cartridge, housing ¾” (~19 mm) wide tape.  It was recorded using the video capacity to store the digital audio signal.  A parallel track stored the usual time code used by both video and digital audio.  The system was built around two U-Matic machines (one to play, one to record), the 1610 (later 1630) analog to digital (and digital to analog) converters, and the DAE-1100 digital audio editor.  Ancillary gear included another Sony device, the DTA-2000, to analyze finished tapes and provide a printout of error occurrences per minute.  This, along with a written “table of contents” indicating start and end time code locations for every track and other incidental details was sent to the replication facility with the CD master.  A pair of U-Matic machines, the 1630, the analyzer and the electronics associated with the editor filled an equipment rack several feet tall.

The editor itself was a small console, a few feet wide.  It contained controls for up to three tape machines (two for playback, one to record), readouts of the time code indicating the location of the tape in each machine, controls to perform editing, and a fader used for gain (i.e., level) adjustments.  Editing in the digital domain no longer involved using a razor blade to physically alter the original tape, as we had always done with analog tape.  (There were some short-lived exceptions in the form of the digital multitrack reel-to-reel recorders that were to come later.)  Digital editing was now effected by playing the original tape while recording the edits onto a new tape.  The finished result needed to be created sequentially.  If, upon listening to the results of an editing session, the producer decided to add to or remove anything from the middle of the program, a new tape was created, requiring the entirety of the program prior to the new edit to be copied first.

As the music was playing and the engineer heard the section where the desired edit point was located, the press of a button on the editor would store a 6-second sample of the music — the three seconds before the button press and three seconds after.  The playback and record machines would stop.  A small wheel in the middle of the editor was used to manually move forward and backward in the captured sample of audio, so the engineer could precisely locate the edit point on the newly recording tape.  Turning this wheel accomplished what used to be done with analog tape by having one hand on each reel and manually “rocking” the tape past the analog machine’s playback head in order to locate the desired edit point.  Where the edit point used to be marked with a grease pencil, all the engineer needed to do now was press another button on the editor.  Now that the “out” edit point was selected on the record machine, a similar process of location would be done on the playback machine to find the “in” point from which the new tape was to continue.  Once the edit point on each tape was selected, a preview button started a process where both tape machines would shuttle backward a predetermined amount of time, still synchronized with each other and then both started to play.  The audio would be from the record machine (i.e., what had already been recorded prior to the edit point) until the edit point was reached, when audio would switch to the playback machine, in effect, allowing the engineer to hear the edit before committing to it.

If a recording or mixing studio console was reminiscent in some way of an airplane or Space Shuttle cockpit, my first look at the DAE-1100 editor reminded me of Star Trek.  It felt like the future, with its smooth, uninterrupted surface of subtle grey, with darker gray, red, orange and blue “buttons”.  Being able to test an edit without committing to it meant all sorts of edits could be attempted without fear of having to splice together a missed edit.  I used to describe the precision of the edits as allowing me to “get in and out within the width” of the razor blade cuts we used to make.  In comparison, I described the thought of editing with a razor blade as now feeling much like editing with a hammer.

Having long experienced what seemed to me to be the inadequate monitoring in the studios I’d worked in, visited and read about in the professionally oriented magazines, I sought to do something different in the new CD mastering room at Atlantic.  Rather than loudness optimized speakers, placed against the wall, near the corners, over the engineer’s head, or small, dynamically challenged speakers placed where they would create a midrange dip at the listening position – both commonly seen in every studio in my experience, I wanted to bring some audiophile sensibilities into the room.  At my request, studio management agreed to install a pair of Dahlquist DQ-10 speakers (my favorites at the time).  These I placed a few feet off the wall behind them, in free space, with nothing else near the speakers.

Once the room was set up properly and known master tapes played back to my satisfaction, it was time to get my first really good listen to digital audio.  The advance word from the hobbyist and professional magazines, as well as from colleagues who’d already gotten to listen to a bunch of the earliest CD samples, was very positive.  Everyone was enthusiastic.  I was going to hear what had widely been touted as “Perfect Sound Forever”.  With great enthusiasm and anticipation, I listened to my first sample.  Then I listened to another one.  And another one.  I listened to all the samples we had.  I went back and listened to some analog master tapes and vinyl LPs to make sure the monitoring was what I expected it to be.  With the analog sources, it was.  With the digital sources, I wondered just what everyone had been raving about.