PME Records Audio Restoration Comparison CD

RAP Audio Restoration CD

Price is only $10 US, including shipping anywhere in the world.

Go Directly to Track Listing

Recently on the internet newsgroup rec.audio.pro, Jon Noring postulated that perhaps some of the newsgroup's members could help him out with the restoration of an extremely rare recording. Here is an excerpt of Jon's original post. The complete thread can be found here, although the rest of the discussion about the actual entries can be found here.

"The recording I propose is "Bugle Call Rag" by the "Whoopee Makers" (a studio band recording featuring Benny Goodman and Jack Teagarden in their very early days.) This recording was cut 23 November 1928, mx. 3562 (108515-2), on a near mint condition Perfect 15126 I am very fortunate to own. My copy is, as far as I and other well-versed classic jazz collectors know, the best known pressing of this recording anywhere -- the original metal parts are probably long gone and no test pressings have surfaced, again as far as I know. This particular take is extraordinarily rare (there may only be a few known copies in existence) and features a good short speech by Jack Teagarden at the start, and instrumental solos by Jack (on trombone) and Benny Goodman (on clarinet).
It is a very worthy historical work of classic jazz to restore, and it hasn't been properly done yet. (It is of the "corn-ball" genre of classic jazz, a style both Benny Goodman and Jack Teagarden preferred not to play, but the recording studio told them (and paid them) to play it, and they do their best to elevate this "corn-ball" to a more serious level of jazz, yet keep the recording contractor happy -- a delicate balancing act.)"

--Jon Noring

Once Jon's post filtered around the world, interest began to catch on. While several different methods were discussed, individuality eventually came through in every entry. Indeed, this became the focal point of the competition: While most projects of this nature are tightly controlled in order to promote The Scientific Method, this project was deliberately loosely reigned. As an intended result of this looseness, what was received was a cross-section of the restoration community, with as many techniques and approaches as there were entrants.

This was not a project intended to define the "best" method of restoration of classic audio. It turns out that there is no "best" method. Rather, with so much debate about the subject over the course of recorded history, there are many valid, legitimate approaches that each have their strong points. In order to explore the validities and intricacies of many methods, many differing viewpoints needed to be represented. Hence, no entrant was informed of any specific rules regarding their restoration effort. They were encouraged to do what their ears told them was best, while trying to retain the listenability of the original musical recording. The restoration also had to be done in the software realm, but there were no other stipulations.

Generally speaking, the entries revolved around three categories of processing: Declicking and Decrackling, Noise Reduction, and Equalization. The reasoning behind the use (or lack thereof) of each method and the finagling involved with each software program are detailed in the notes submitted by each entrant.

The learned observer will note that there are actually two audio samples for every file submitted. This was a compromise decided upon by the entrants, and allows three comparisons to be made. First, the files as they were originally submitted, which may have level changes brought on by whatever processes were in use during the production of the file. This allows for direct comparison of the various techniques of restoration, as a representation of the engineer's vision of what the song should sound like. Generally, the overall RMS level was brought down by the various processes, so each file was left "as is" in order to preserve all the detail invoked (or lost) by the collective alteration of the music. The second comparison involves the matching of the average RMS level with the file of the originally recorded song. This allows the listener to compare the alterations directly with each other, since the apparent levels will be the same for every file. However, with most files this presents an adulteration of the engineer's intent, because alteration of the final product will result in artifacts that some listeners may not care to hear, or the elimination of some detail that was present in the engineer's original submission. The final comparison can be made from set of files to set of files, to allow the listener to determine what those artifacts may be, and whether their addition is objectionable in the long run.

There has always been debate within the professional audio community regarding the significance of normalization artifacts, so to be sure, these differences will require careful consideration and concentration. But the final decision regarding whether those artifacts are musically important must be left up to you, the listener.

Click on a title to see the notes for each selection. Using the Next and Back buttons on each page will render a handy slideshow comparison of the histogram and spectogram diagramming of each entry, from the normalized file.

Track Listing:

Track 1 - The Original Recording

Normalized Attempts:

Track 2 - Barry Blumenthal

Track 3 - Ben Bradley, DeClick Only

Track 4 - Ben Bradley, DeClick and NR

Track 5 - Carey Carlan

Track 6 - Chris Johnson

Track 7 - Craig Patterson

Track 8 - Peter Larsen, EQ, DeClicked, LPF only

Track 9 - Peter Larsen, EQ, DeClicked, LPF and NR

Track 10 - Robin Farrell