Charlie Parker and Miles Davis perform at the Three Deuces jazz club in New York City, 1947.
In the long and vast history of jazz, Charlie Parker stands as one of its giants. The saxophonist known as “Bird”, alongside trumpeter Dizzy Gillepsie, invented bebop in the middle forties, forever expanding jazz’s musical possibilities. A genre characterized by fast and complex playing by small groups, bebop saw the use of extended chord structures, which opened the door for more harmonic and melodic choices. Surrounded chronologically by the prevailing swing era, bebop seemed to be its antithesis. The recorded legacy of this new genre was truncated by the American Federation of Musicians ban, waged by its infamous president, James C. Petrillo. Petrillo wanted the record companies to pay royalties, and when they didn’t comply, he enforced a ban on recording for all union musicians starting July 31, 1942. For this, Petrillo was lambasted from every conceivable corner of the entertainment business. The halt on recording dragged on – in September 1943, Decca and the then–new Capitol Records agreed to pay royalties. Eventually, in November 1944, Victor and Columbia likewise capitulated. In hindsight, observing record sales shows the majors weren’t that affected by the ban. In fact, due to the need for product to push, the labels scoured their vaults and ended up inventing the compilation album in the process. But a certain musical damage entailed – the earliest bebop was simply never recorded.
While this new music was flourishing at Minton’s Playhouse, its development evaded the recording studio. Herman Lubinsky, a middle-aged no-nonsense radio repairman, founded Savoy Records in 1942. Lubinsky managed to get a few records released before the ban, and once it was enacted, he recorded musicians under pseudonyms in attempts to sidestep it. One of his A&R men, Teddy Reig, was a regular at many New York City jazz clubs. Hired in early 1945, right after the ban was lifted, Reig befriended up-and-coming artists and brought them into various studios to make recordings. Parker and Gillespie, along with the genre’s other great movers and shakers – luminaries such as Dexter Gordon, Erroll Garner, and Max Roach – all started cutting records for Savoy.
As exhilarating as early bebop performances are, in listening, the modern ear is quick to notice sonic shortcomings. Savoy didn’t feature the best recording fidelity at the time. This isn’t to say they’re bad recordings – only that the engineering know-how and analog equipment in use at a smaller label couldn’t compare to the majors. The fact that Charlie Parker never experienced high-fidelity recording is regrettable, and remains a large disservice to his talent. But another problem compounds the Savoy sides’ sonics – what labels now use as the source material. Music historian Frank Kofsky claimed that Lubinsky, a famous cheapskate, had “a scarcely disguised disdain for Black art” and an unwillingness to understand much of the music his own record label recorded. Like some record companies’ masters, unfortunately Savoy’s no longer exist. It’s not known what happened to the metal parts*, only that: they were once transferred to tape, and their owner was a hard-line businessman who was shortsighted, indifferent or perhaps even animus to the revolutionary music that surrounded him.
Because of those factors, today, the sound of these recordings exist in limbo. When the master source is missing, the true fidelity – however good or bad in the first place – is obscured to some degree. The Savoy catalog was purchased by Concord Music Group in 2019. The Savoy disc masters were transferred to tape in the early fifties for the production of five 10” compilation LPs covering the label’s early history – and CMG’s Craft Recordings have recently reissued that very set of LPs. A re-release in the truest sense, Craft maintained their original title (“The Birth of Bop”) and they even feature jackets with the exact same cover art. As the disc masters can’t be located, those disc-to-tape transfers, now performed over 70 years ago, have been the sources used for almost every reissue, boxset or repackaging of Savoy recordings since, including this release. It appears Mosaic Records’ Classic Savoy Be-Bop Sessions 1945-49 boxset was sourced from commercial pressings, although per “limited edition–style” licensing agreements, this mastering is absent in the realms of hi-res digital and streaming services, not to mention out of print on CD.
Disc transfer and the instrument that made it feasible, the modern phono cartridge, were in their infancy at the start of the fifties. Early in this decade, magnetic tape – with some hesitancy at first – replaced disc recording as the primary format in professional recording studios. In the Victor Recording Manual from 1940, the most comprehensive surviving technical document from the 78 rpm era (credit to the late Ray Rayburn for saving it), Victor’s engineers term Disc transfer “Re-recording From Records”, depicting it as “[an] operation [that] represents such a diversity of problems and a wide variance of operating conditions…” a statement that painfully rings true a staggering eight decades later. Groove size, disc speed, playback emphasis/de-emphasis curves, off-centeredness, and pressing material vary widely between the discs known as “78s”, the dominant format for recorded music from the late 1890s to the mid-1950s.
In order to make a great disc transfer, the engineer must keep track of all of these variables and have the tools, both on the analog side (different styli shapes and sizes, various record cleaning fluids) and digital as well (repair-oriented DAWs) to deal with those “diversity of problems”. The later convergence of tape recording, improved lathe motors, the RIAA curve, and the scale production of vinyl alleviated most of these differences.
The Savoy recordings present a strange dichotomy. What sounds better: A 70-year old tape transfer of a master disc source, or a modern transfer of the original commercial pressing? Another caveat lurks: because World War II cut off access to India (where high-quality shellac could be imported) records made especially during and sometimes after the war suffer from poor, gritty and noisy surfaces. Sadly, Savoy’s commercial pressings were made from that era’s poor stock. For the curious, this is the same problem that exists with the Paramount blues recordings: Missing metal parts, poor record surfaces, a recording fidelity that is decent but unfortunately not as great as it could’ve been.
Can modern, reference-grade transfer and restoration of a grainy shellac 78 really surpass the transparency of the master source – in spite of another analog step in between and the sheer age of the original transfers?
To find out, I sourced a commercial pressing of one of the headlining tunes of the whole box set – Tiny Grimes’ “Romance Without Finance”, featuring Bird on sax – and got to work. As this track was recorded earlier than the 1945 start date for the Mosaic release, no official modern mastering from a disc source exists. After a scrub with distilled Disc Doctor fluid, vacuum cleaning, and the same process with lab-grade water for rinsing, I placed the disc on the rubber mat, elevated above the spindle of my turntable, a Technics direct drive. For this transfer, I first centered the disc by knocking the edge with a porcelain chopstick until no lateral motion was observed when the tonearm was lowered onto the record surface. I used a 2.8 mil elliptical stylus, which sounded best, and transferred with a flat frequency response.
As I had a stereo transfer, and needed to get to mono, I created a sum-and-difference mix that canceled out some of the noise and damage. For those of you who make Mid-Side recordings, this is the inverse function. To do this, the engineer must use a DAW to split the interleaved file into two mono channels (which in this case are the left and right groove walls), mute the stereo, and pan the two mono tracks center. Name them to avoid confusion. Duplicate the worst sounding channel and mute it. Invert the polarity of the original of the duplicated track and alter the level until you hear the least amount of music. At this point, you are canceling out the most vertical modulation and are listening to what disc transfer engineers call the “null signal”. When you unmute the duplicated channel, the signal you just heard will cancel out, leaving you not only with less noise and damage but also a theoretical 3 dB signal-to-noise ratio improvement. This uncomplicated practice can virtually eliminate damage present on only one groove wall by replacing it with information on the other.
The sum-and-difference mix helped this disc out a lot. I then applied a 500-cycle Infinite Impulse Response (IIR) bass emphasis curve, which sounded appropriate, by using a Nyquist prompt with biquadratic equations required to closely replicate analog. In this case, the curve is trying to electronically simulate the mechanics of the restoring spring in the cutterhead. Thanks to my friend Dustin Wittmann for collecting a document of these equations which can easily be copied and pasted.
The disc had quite the grainy surface, so I knew I had to use CEDAR for initial declicking. In my opinion, CEDAR is the best restoration hardware/software available, although most others can work well as long as the engineer is careful and doesn’t overuse it. The CEDAR Studio declicking (which compared to the flagship Cambridge is based on an older algorithm) wasn’t getting the results I desired, so I called in a friend. Seth Winner, whose studios bearing his name have done exceptional work (especially for recordings by the great maestro Arturo Toscanini) was able to help me process this recording. To summarize: Seth used CEDAR’s declick module in SADiE to get the disc sounding a lot cleaner. To my ears, it seemed to put the air back in the studio that was obscured by the disc’s poor surface. I then manually removed around 50 more clicks using iZotope RX’s Interpolate tool. Seth used a light pass of CEDAR’s NR-5 noise reduction to cut back on noise. I then applied a high end de-emphasis curve, took out some rumble using iZotope’s Spectral Denoise module, and soon after did a bit of EQ at the low frequencies. From there it was done.
The result is of course more noisy than the tape transfer from metal parts. It’s also complete, which strangely can’t be said for the tape: the ending was cut off. In my opinion, we were able to achieve something that’s more musically transparent, closer to, and does more justice to the original performance than any previous mastering.
Most importantly, what do you think?
To be clear: This isn’t a drag on Savoy, CMG/Craft, or the many talented engineers who were occasionally tasked with returning to the tapes as digital restoration improved. Some of them I’ve gotten acquainted with and I admire their perseverance in dealing with sonic challenges. What I’m writing is plainly a cautionary tale: When a master source is missing, it should be imperative to correctly identify the best source possible and begin working from it only. All these years later, with powerful tools such as CEDAR and iZotope at the engineer’s side – don’t count anything out.
*Before tape, records were made by cutting direct to disc – wax or lacquer – and then plating that disc with nickel to make a father, or master record. From the master, mothers (groove) and stampers (ridge) could be produced, which facilitated the mass production of pressings (groove) for commercial sale. Fathers, mothers, and stampers today are known collectively as “metal parts”.