From: Colin J. Bray
Date: Mon, 5 Jun 1995

I have a few points to make concerning my experience of Stylii and equipment for playing 78s. Over the last few years I have got to know John R.T. Davies, who in collecting circles is highly respected for transferring 78s to LP and now CD. Collectors who like jazz and blues recordings from the 20s will be very familiar with his work. He did the transfers for the Fountain/Retrieval LP label and has done most of the work for Timeless, JSP, Halcyon, Frog and many other labels. He did the transfers for the the "Complete Louis Armstrong with Fletcher Henderson" 3 CD set (which I co-produced) and is doing the transfers for my new label - "Jazz Oracle". The first issue is by "The Halfway House Orch 1925 - 28". John uses a Lenco deck (as do I) and until recently the original arm. He now uses a SME arm. Generally all his transfers are made using a Shure M44 cartridge, occassionally he uses a Stanton 500. He categorically told me the much more expensive Shure V15 was inferior for 78 playback.

The most important point is the size and shape of stylii used. Stylii made for 78s by the big manufacturers are just not good enough. About three or four different sizes are needed for the majority of records although John R.T. Davies uses about thirty different ones! Generally elliptical stylii are better than spherical, but THE MOST IMPORTANT point is to use a truncated stylus, i.e. flatened off at the bottom. The stylii will then pick up the relatively unworn part of the groove. The bottom of the groove is usually badly damaged by the steel needles originally used on gramophones. Problem is the big manufacturers like Stanton, Shure etc. do not, and have never made, truncated stylii. These type of stylii can be bought from a company in England called "Expert Sylii" in Ashtead, Surrey (They claim to have manufactured the first magnetic stylii assemblies). They make these for the Stanton 500 and Shure M44 assemblies but can fit ANY size stylii to ANY cartridge. They can be a bit slow as they have loads of orders but they are an excellent company. The stylii cost around 30 British Pounds - about US$45. They will make any size stylii you want. There was a company in California advertising truncated stylii a couple of years back, perhaps there is a company in the US doing this service, but I suspect they sub contract to "Expert Stylii". I have been using these stylii for some years and am very pleased with the results. They are not, however, going to make a worn recording sound like new.

I note David Hoehl's comments on 78-L about playing 78s mono. Absolutely. Amazing how many collectors don't. Another useful point is to play one side of the groove, often one side of the groove is more damaged than the other, so this can make a big difference. I have built a unit that enables me to switch backwards and forwards between each side of the groove as well as being able to play vertical recordings. Why no one manufactures this I can't understand as it can make a big difference. This feature is available on the "Owl" unit along with a lot of other features. I have a copy of King Oliver's Creole Jazz Band playing Krooked Blues/Alligator Hop on Gennett which has quite bad groove damage. Playing one side of the groove makes an incredible difference. This is the rarest of the Oliver's except for "Zulus Ball" so it's very worthwhile. I only use an equaliser (Radio Shack 2 X 12 band) to roll off the high end slightly. This could be better done by a variable roll off filter - if such a thing was commercially available. The Lenco deck mentioned above was made in Switzerland and is of the continuosly variable speed type. It is ideal for 78s. They were marketed in the UK as Golding Lenco.

What I've mentioned above is not going to produce a product good enough for transfer to CD but will produce results good enough for pleasant listening.


From: Gabe M. Wiener
Date: Sun, 2 Jan 1994

Here is how 78's are generally remastered these days.

First, the record is carefully cleaned. There are special machines for doing this well, such as the Monks. DO NOT under any circumstances use an LP vacuum machine on 78's! Next, the groove width is determined. This usually involves the engineer's simply knowing, for a given record company, how wide their grooves were. But if there's any doubt, several sizes can be tried, or the record can be put under a stereo microscope with a reticle that indicates millimeters. So anyway, to make a long story short, the proper stylus is mounted on the arm and the tracking force is set up for the particular type of record. Generally tracking force runs between 3-6 grams. I use a Technics SP-15 with an SME-3012 arm for restoration and a stanton transcription cartridge. From there, it goes to a restoration preamp. You *cannot* use a regular phonographic preamp for 78 work because 78's did not conform to the RIAA equalization curve. The standard preamp used for 78 work is the OWL-1 which has adjustable EQ for low and high, as well as flat modes, and Hill & Dale mode. It is advisable to tweak the low and high ends by ear in the analog domain before proceeding. Sometimes the EQ curves for a specific comapny are known, but usually a good mastering engineer can tweak the knobs by ear and get it sounding good.

Now, there are numerous other issues that we could go into now, such as figuring out the speed of the recording. What, you thought that 78's were recorded at 78 RPM? Dream on, gentle reader. 78's were recorded between 66 and 90, generally. And many weren't punched on center, and many are warped, and...and...So let's say you have copied the 78 direct to a DAT. You now have a fairly decent rendition of what's on the 78. But the sound is generally not acceptable to modern listeners. First you have to remove the impulsive clicks that occur on the recording. These are clicks called by scratches, pops, ticks, and gouges in the material. Although there have been many analog units over the years to do this (such as the Packburn ), most studios are doing this sort of thing using one of two processes....the CEDAR de-clicker, or the Sonic Solutions NoNOISE production de-clicker. You can buy the CEDAR de-clicker as a stand-alone audio component for $16,500. It is a remarkably simple device to use. Only a few parameters to tweak and it can be operated by any competent audio engineer or a knowledgeable audiophile who knows something about interfacing. The Sonic Solutions system is much more complicated and requires a trained engineer to operate, but in certain situations it can do a lot of things that the CEDAR can't do in denoising. It is a complete workstation system, and a system fitted out to do de-noise can easily cost $50,000 plus months to train someone in using it, plus time to practice. If you want a complete system for mastering CD's, including the optical recorder, call it $80,000. The basic idea behind de-clicking is that a click is a break in the waveform....you can see it as a spike if you display it on a workstation. The concept, greatly oversimplified here, is to examine the area surrounding the click. Algorithms look at its context information, its periodicity, its various rates of change, etc. and using some fast arithmetic, can recompute what should have been there but wasn't. Okay, so you've de-clicked the recording. You still have other noise, namely crackle....crackle refers not to the egregious pops and clicks, but to that subtle bacon-frying sound that underlies many 78's. Once again, you can run this on either the Sonic Solutions or CEDAR digital workstations, or use a stand-alone CEDAR box for about $20,000 to do this. These are high-speed floating point processors that do a tremendous amount of high-order math to reconstruct a signal sans distortion.

Last of all, you have the broad-spectrum surface noise from the playback process. This is the one very tricky area where an engineer can ruin a recording if he does it wrong. The problem is that since the noise is broad spectrum, an engineer can slice off ambient noise from the program material as well as the noise. One can use either Sonic or CEDAR workstations for this, or there are a few other products to do this sort of thing, such as the EMT 248 digital audioprocessor. The basic idea is this....a noise curve is taken as a "thumbprint." Next, the program material is played and divided into a certain number of bands. Those bands are analyzed in real time to see if they contain program material or noise. Program material is left alone, while noise is attenuated. Of course, there are parameters to tweak....threshhold, amount of attenuation, cohesiveness of the bands, sharpness of the attenuation fall, minimum and maximum operating frequencies, etc. After all this vivisection, many 78's could do with a little EQ to improve clarity. Finally, the tape is ready to be mastered to CD. The tape is loaded to a CD mastering system where it is edited, assigned track start codes, etc., and sent for pressing.

Anyone in NYC want to see how it's done? :-)

Gabe Wiener
Sound engineering, recording, and digital mastering for classical music

"I am terrified at the thought that so much hideous and bad music will be put on records forever." --Sir Arthur Sullivan


From Dan Hladik:

Here is a *brief* description of the CEDAR system.

CEDAR (Computer Enhanced Digital Audio Restoration) is a group of processes used for audio restoration. Some are performed either using stand-alone "black-boxes" *or* through a PC workstation. Other processes are PC-only.

I used various parts of the CEDAR system (at a sound studio) for my work for the Met opera starting in 1993, so I can only tell you what the parts are and what they seem to do.

Azimuth-correction - This process (black-box or PC) is for phase correction of tapes, especially those recorded with mis-aligned tape heads. For 78s, this process is useful in choosing the quieter groove wall.

De-scratch - This process (black box or PC) is the 'original' CEDAR for removing 'clicks' and 'pops', i. e., transient or impulse noises. The software has been refined to be able to 'recognize' sharp musical attacks (such as soprano staccati acuti, precussion, brass) which other devices (like Packburn) chew up.

De-hiss - removes continuous higher-frequency noise (PC only). Requires a noise 'reference sample' (i. e., hiss but no music) which it then removes from the source.

De-crackle - removes lower-frequency, less continuous noise (black box or PC) and also requires a 'reference sample'.

Digital EQ - (PC only) - allows multiple equalization curves to be applied to the source. Cannot currently replicate the 'hard-wired' curves supposedly used for recording electrical 78s.

The CEDAR black boxes cost between $12,000 and $18,000, the PC workstation about $25,000 plus extra for software.

Studio rates for CEDAR run $10-$15 per minute of (finished) music, depending, of course, and how many processes and passes are needed.

As with any process, choosing among not enough / just right / too much is where the skill and taste of the producer/engineer/techician comes into play. My own personal philsophy is 'less is better'.


October, 96 - New Affordable Software for 78 Restoration!

Diamond Cut Productions
P.O. Box 305
Hibernia, NJ 07842-0305

REVIEW:

Date: Sat, 19 Oct 96 19:27:00 -0500
From: Graham Newton

It is probably the biggest bang for the buck presently available. I've compared it with other programs like Cool Edit, Sound Forge, DART and DART Pro, as well as the pre-eminent CEDAR real time stand-alone modular processing system.

The real time preview function "stutters" on a 486DX4-100, so the recommended optimum system requirements would be "as fast as you can afford". I'll be upgrading my system to a Pentium 166 which should allow the "preview" to work without that problem. Still, though, expect processing of a 4 minute file to take anywhere from 30 minutes down to 10 or so, depending on what degree of processing you want the program to do. The more elaborate the process, the longer time needed to process the .WAV file, likewise, the slower the computer.

The features are impressive, BUT, remember that you will likely need to run more than one pass to accomplish what you want... One pass to de-click, another to take a fingerprint for de-noise etc., etc.

This work is time consuming, and there is no fast AND cheap answer... which is why CEDAR can get on the order of $20,000 per box for their system... it does this in real time. Put dirty audio in, get clean audio out, in the time it takes to play the tune.

What do I know about this stuff?

I'm a former RCA Victor recording engineer, and I cut thousands of acetate or lacquer masters for phonograph record production using Scully and Neumann lathes. I now operate a modest digital audio restoration suite specializing in vertical or lateral 78's, broadcast transcriptions, acetate transfer and restoration using CEDAR and a variety of other specialized equipment to do the transfers, de-clicking and noise reduction. The results are often spectacular, provided the original material is in playable condition and the grooves have not been destroyed by repeated playings with worn styli. I do the restoration work for a few producers that issue CD compilations of old phonograph recordings.


ANOTHER REVIEW:

Date: Sun, 20 Oct 1996 01:38:38 +0000
From: Robert J. Lang

I've been using Diamond Cut Productions software for a while to restore Brunswick acoustic recordings from the 2000 series (mostly Gene Rodemich's Orchestra). I haven't used all the features that the software comes with, but I'll comment on the things I use most.

- Impulse noise filter for removing ticks, clicks & pops.
This filter is really great. It does an excellent job of removing what I consider to be the most disturbing noise types: the ticks, clicks and pops. I always check to see if I've affected the music after using a filter by subtracting the filtered wave file from the original source wave file, and this filter removes very little, if any, of the music.

- Continuously variable Zoom-In & Zoom-Out functions for precise editing.
- Adjustable markers for sample accurate editing.
- Cut, copy & paste sections or entire wave files.
I use these three features in what is probably an unusual way. I load in more than one recording of a song and, after using the impulse noise filter on both, open one as the "source" file and the other as the "destination" file. After determining which file is best, I use the above three features to cut and paste sections from the secondary file to fix any major defects in the best file. The markers and zoom functions, coupled with the fact that the "source" and "destination" files lock together when zooming, make this an easy job.

- On-Line Audio Restoration Tutorial with Help file.
The Help and Tutorial screens are excellent, probably because the people who wrote this software developed it for their own professional use in restoring the Edison Records. The idea of selling the software to others came later. So when they wrote these files, they really understood how to use the software in a way that gives first class results.