Cedar: DCX Declicker, CRX Decrackler, DHX Dehisser

My dad raised me on swing-era big bands and crooners. In the heat of the summer, we'd be down in the cool basement with a stack of 78 RPM records. Many years later, he showed up at my New York apartment with those same records. A well-equipped son, I have all the necessary gear to play everything but an Edison cylinder, doing my very best to squeeze as much music out of those gnarly old beasts as possible. But I didn't have any noise reduction technology... CEDAR created the Series X and Series X+ products based upon their already popular and successful noise reduction algorithms. The three units I tested are each 1 unit rack-mount devices, with the most simple user-interface, operating in real-time and doing such a great job as to justify their $6k per unit price tag. Compared to their predecessors, that's three for the price of one. For half the price, software only versions are available for the Soundscape and SADiE workstations.

BASIC HARDWARE: All Series X and Series X+ modules offer 24-bit AES/EBU plus IEC958 (SPDIF) I/O. Inside you'll see the same PCB, customized with front panel controls specific to the requirements of each respective process. All units have 40-bit floating-point processing. What I didn't know, but discovered via their website, was that CEDAR's award-winning algorithms were originally developed for the Windows environment - non-real time - sort of an odd twist when you consider that now, "everything" is a plug-in. I didn't test the two Series X+ modules: the BRX+ debuzzer, and the AZX+ Azimuth Corrector, but I can tell you that the units I did test successfully removed the buzz from some telephone recordings. Like modern recording, it is important to start out with the best possible source material. It's hard to find a 78 that has never been played so it helps to have several styli to get a good transfer. With rare exception, a modern stereo stylus is not an option. I have a handful of styli that cover the 3 mil to 1 mil range required by 78s, transcriptions, original mono LPs and 45s. Until this review, I formerly leaned toward using smaller styli to play deeper into the groove away from the major ticks and pops on the surface. I "settled" for a more consistent wash of record material noise because, unlike vinyl, shellac is a gritty material that is only made more obvious by a smaller stylus. These three CEDAR products are very specialized tools, equally suited for cleaning up records and optical film soundtracks. The process of removing various noises is much like the care taken at an archaeological dig, removing layer by layer of dirt without disturbing the artifact. In fact, the boxes must be used in order - Declick, Decrackle and Dehiss - to maximize efficiency. Layer one, major ticks and pops, are transients so far above the recorded signal as to be easily skimmed off the top. Just turn the single Sensitivity knob clockwise until they disappear. No metering or additional controls are provided or required. Whenever in doubt, two transfers were made, a fortuitous decision that perhaps my dad didn't appreciate as I cycled through a handful of styli for each record. At that time I couldn't imagine the good fortune of doing a CEDAR review. It turns out that using a larger stylus is quite okay because the Declicker does such an absolutely wonderful job of removing surface pops and ticks. You could stop here and be happy. The Decrackler is a bit more "sophisticated" in that it includes not only Sensitivity and Level controls, but also a Detect switch that allows the user to more precisely locate the processing Threshold. Decrackler has a more difficult job, because "record grunge" is not so far above the music. If the disk condition and transfer were really good, however, it took less than a minute to optimize. I never really found the Detect switch to be very helpful for this material although the artifacts are fun to listen to, kinda bubbly like a clarinet section. Sometimes the filter artifacts made me wish that Cedar made the inverse of a de-esser - a re- esser - because recordings made before the Neumann U- 47 was "invented" lacked vocal presence. Instead of using the Detect switch, it was easier to momentarily turn the Declicker process off, find the Decrackler's threshold and then turn the Declicker process back on. I generally ran the Sensitivity control at or very near full up, varying only the Level knob from the 11 o'clock to 4 o'clock positions. Layer three, removing the remaining noise, was most difficult. While the high frequency capabilities of WWII- era electronics were quite good, the disc cutting equipment did not have the extended high frequency response and later, the pre-emphasis, that would cut through the surface noise of a new pressing, let alone one that had been played a few zillion times. Sessions up to and including the forties used ribbon mics that have a smooth, understated top-end compared with any condenser mic. When used on a vocalist, any directional mic will be quite warm - the proximity effect in action - so that some vocalists had to work extra hard to achieve intelligibility when competing with a blaring trumpet section. For 78 RPM recordings, the end result above 5 kHz is "que sera sera." The archivist is thankful for any gifts and surprises above this frequency where a healthy amount of remaining surface noise is either tolerated or filtered. On fifties-era hi-fi recordings, the lead vocal stands miles above the music - all of the recording equipment was more capable - assisted by the aforementioned U-47, the AKG C12 and its cousin the Telefunken 251.

The Dehisser has three knobs: Level, Attenuation and Variance. Rotated clockwise, Variance optimizes the algorithm for material that has a lot of variation in the noise content. Most of the time, I set Variance full CCW and Level at full CW, which I am told is not the intended way of using the Level Control. Perhaps it was my source material, but I noticed artifacts when used as intended. CEDAR claims my approach over- identified the noise content leading to a much greater risk of artifacts, as well as less than perfect dehissing. I felt the Attenuation control operated as the equivalent of a dynamic low-pass filter to which the manufacturer noted... "The Attenuation control sets a limit on the amount of noise reduction at each frequency. So, for example, if the Attenuation is set to 6 dB then, if the process identifies 4 dB of noise at a given frequency at a given moment, removing the full 4 dB. However, if the Attenuation is set to 6 dB and the process identifies that there is 10 dB of noise at a given frequency at a given moment, it will only remove 6 dB. This drastically reduces the risk of unpleasant artifacts and also leads to a more musical result." For my purposes, increased clockwise rotation reduced the upper-limit of high frequencies that were allowed through. Incorrect as I may have been, I merely chose a less aggressive setting - most of the time, the Attenuation control never made it past the 10 o'clock position - willing to settle for some background surface noise. Besides, after several late nights, I had to ship the loaner units off to the next lucky evaluator, so I wasn't about to over do it. The beauty of dedicated hardware was not lost me. Quite often the surface noise was more apparent at the beginning and end of each record. With a little rehearsal I rode the controls to more aggressive positions, backing off when the trumpets stopped and the clarinets or lead vocal came in. (Remember that the algorithms had a bubbly, clarinet-like effect that was hard to distinguish from the real thing forcing me to be less aggressive during those passages.) With each record I started "with scratch," redialing each time only to repeatedly land the controls within the same useable window. In that sense the control parameters were well thought out. The end result sounded much like an old tape without noise reduction - revealing more music and detail than I had ever heard from these records - much of the time forgetting I was listening to records. (I could even hear pages turn!) While the Dehisser was effective at removing some noise, I did not overuse it. My ears quickly tuned in to the artifacts - always pushing the envelope - then returning to a safe place where at least some of the noise was eliminated. The post-disk transfer chain started with a Tascam DA-45 24-bit DAT deck feeding the three CEDAR processors, a TC Electronic dBMAX (a five-band broadcast device that is like a Finalizer on steroids) and a Sony PCM-R500 DAT recorder. A Studio Technologies "Studio Comm" interfaced the Sony DAT with Dynaudio Acoustics BM15A powered monitors. In most cases BASF/EMTEC tape stock was used. The dBMAX became an important tool. At first I used it as a static filter to reduce low-frequency rumble and noise above 15 kHz. In between, I boosted the region around 5 kHz using a 2.5 to 4-octave bell curve. This not only added some excitement to the recordings but also improved the ability to hear the area being processed, allowing me to better judge the level of acceptable aggressiveness. I also took advantage of the five-band compressor in dBMax. In those days, the lead vocal was mixed way out front. Since it was the only "instrument" that was truly close-mic'd, the threshold was weighted to "see" more midrange - the vocal and trumpet region - using ratios between 1.4:1 and 1.65:1. The attack and release times were optimized for each frequency range. No more than 3 dB of compression was used, followed by a peak limiter that only occasionally flashed. The vocalist, Art Lund, recorded with the Benny Goodman Orchestra on OKEH (Columbia), which I believe, was just after WWII. "Blue Skies" was his most popular recording with Goodman. In a very odd way, Art Lund and Morrissey (from The Smiths) occasionally have similar vocal styles. The bulk of the recordings were made by MGM from 1947 to 1951. Most of the orchestral arrangements were by Johnny Thompson, who I must say was particularly tasteful. Both the vocalist and the arranger were under-appreciated, especially according to my dad! The records started out as shellac and made the transition to Vinylite. With the right stylus, shellac can sound pretty good if not too worn. Vinylite is a much better compound that corresponds with the introduction of the Microgroove, the name for the 33-1/3 RPM "LP" and 45 RPM records (both played with a 1-mil stylus). Of course, CEDAR really made the difference! I processed over sixty songs. The easy way to show one's love for a parental unit would have been to buy a CD boxed set, but I knew that wasn't possible. A few years before Seagrams purchased Polygram, I was lucky enough to do some maintenance at their Edison NJ tape library. It turns out there are less than a handful of Art Lund recordings - some transferred from disc - the lack of catalog extending into many classic jazz recordings that were mysteriously lost over time, well before the consolidation of the record labels. There was no easy out for me. Even after all this work I still had to sequence, master and burn two CDs. The reward was seeing my dad's face light up as he was transported through time back to his "courting" period. After you read this, check out www.tangible- technology.com for "before and after" samples. Thanks to CEDAR, I may have another career option. (CEDAR Audio USA, 43 Deerfield Road, Portland, ME 04101, info@independentaudio.com)