Slashdot Mirror


5000 Cylinder Recordings Placed Online

Jon Noring writes "The Department of Special Collections at the University of California Santa Barbara (UCSB) Davidson Library recently placed online, with free access, over 5000 sound recordings as part of its Cylinder Preservation and Digitization Project. These recordings date from the 1890's to the 1920's, all transfered from Edison cylinders using state-of-the-art equipment. The restorations are first-class, using CEDAR tools. Besides MP3 and streaming audio, the raw transfers are also available for diy'ers to try their own hand at audio restoration. For those who like their music 'hot', there's not much there since most of the cylinders predate the start of the Jazz Era (ca. 1917), but there is some early 'mouldy fygge' dance-type jazz, like 1920's 'Peggy' by Lopez and Hamilton's Kings of Harmony Orchestra."

11 of 156 comments (clear)

  1. Cylinder recordings are actually quite good by Animats · · Score: 4, Informative
    Many of the early Edison recordings are of surprisingly good quality. It turns out that the mechanical recording process wasn't too bad. The tinny quality of early cylinder recordings came from the mechanical playback process, which was terrible. When those cylinders are played with modern equipment, they sound much better.

    The Amberoll cylinders were tough, too. They had to be, to survive repeated mechanical playback, with a stylus pressure of about a pound. So they're much tougher than vinyl records.

    There's now optical equipment for reading damaged or fragile cylinders and records. UCSB isn't using it, but it's available for the tough cases.

    Some of these recordings are a century old. The original media are still playable. It's sad that we don't have something to transcribe them to that will be playable a century from now. All we can do is hope that someone will recopy the files periodically.

  2. Re:Lossless compression? by great+throwdini · · Score: 3, Informative
    mp3? Would lossless compression have been a better choice for archiving all these ancient songs?

    From the project site:

    "Surrogate files for online distribution were created with Sound Forge 6.0's batch converter (mp3 files) and Cleaner XL (mov files)."

    The mp3s/webstreams are for the unwashed masses. The assumption is that the original captures have been retained in a more suitable archival format.

  3. The raw transfers (lossless) are available by Jon+Noring · · Score: 3, Informative

    Note that the original raw transfers (lossless wav) are also downloadable for each song. It would not surprise me (but I have not checked), that the restored version (using CEDAR) in lossless format is also available in the directory of the archive (but there's no public link to it from the discography page.) It's the high-quality transfers that are the most critical to do right, and UCSB did do them right. Save those in lossless format, and they'll always be around for anyone to restore. Algorithms and applications to restore old recording will continue to improve, and these raw transfers can be re-restored at a future time to improve the sound even more.

    1. Re:The raw transfers (lossless) are available by babaluma · · Score: 2, Informative

      This is correct. We have six files for each cylinder. The original archive file is a 24-bit, 44.1KHz wav file and can be downloaded. We captured at 24 bits not because cylinders need the 144 db of possible dynamic range (if only!) but because noise reduction supposedly works better on files of greater bit depths.

      The second file is also a 24 bit file that was run through CEDAR in real time to reduce the amount of noise. It's very mild denoising compared to what is often done for CD reissues, but we just wanted to make it a bit more palatable to the ears of the uninitiated and remove the worst of the pops without clamping down the sound too much. That's why these cylinders sound relatively bright compared to a lot of the CD reissues you'll find of early recordings. There are a lot of cylinders in the archive that are still very noisy and are very difficult to listen to, but this site is not just about entertainment, it's a tool for research.

      The third file is a CD standard file that was dithered down to 16 bits and there are three compressed files for online access. The sixth file was for dialup connections, but we didn't put them up. I don't know anybody with dialup anymore that could test them.

      Our goal in allowing downloads of the raw file is that if somebody wants to do a CD reissue or reuse the material we should provide for that as they are in the public domain. But if they want a cleaned up version they should do their own restoration, so we don't provide access to the cleaned up wav file (unless somebody asks).

      And if anybody cares, these are two pop songs I've been listening to a lot:

      Theodore http://www.library.ucsb.edu/OBJID/Cylinder3429

      Any Rags http://www.library.ucsb.edu/OBJID/Cylinder4374

  4. Re:A good example of why we need to limit copyrigh by Jon+Noring · · Score: 5, Informative

    Actually, in this case, had there been Federal Copyright law applied to sound recordings as it was to other types of creative works (like books), the pre-1923 sound recordings would all be public domain.

    But they are not the kind of example you are talking about. It is worse than that! Why? Because pre-1972 American sound recordings are NOT covered by Federal Copyright Law. Yes, you heard me right, Federal copyright law does NOT apply to pre-1972 sound recordings, and according to Title 17 of the U.S. code won't apply until 2067. In the meanwhile, then, sound recordings are covered by a patchwork of state copyright laws (both statutory and common law), plus other mechanisms. The Capitol vs. Naxos case was filed in the State of New York under New York copyright law, for example (Google that for more information).

    What does this mean? State copyright laws, by and large, have no limits. So, for example, Columbia cylinders recorded back in 1890 (technically owned today by Sony-BMG) are still copyright protected (at the state level), and won't revert to Federal protection until 2067 (if the copyright terms remain the same as today, all pre-1972 sound recordings will then revert to the Public Domain in 2067). This means that these earliest cylinders will, unless Congress acts, have at least 177 years of copyright protection.

    Most of the UCSB collection is from Edison cylinders. Edison is a unique case in that the ownership of the Edison recordings is the U.S. Federal Government (via the National Park Service), and I believe they are not claiming any state copyright protection of them (but they might be able to). So UCSB felt free to go ahead at least with the Edisons. There are a few other early labels whose ownership is totally unknown and likely abandoned, such as the Grey Gull "group" of labels of the 1920's. These are very interesting to transfer as well. There are some really oddball stuff from before World War I, too, that are probably abandoned.

  5. Re:One reason it's better by Anonymous Coward · · Score: 1, Informative

    Since the recording head and the playback head on a disc move at the same rate at the same radius, there is no frequency drift, there's just less music in the middle than on the outside ;)

  6. Re:I wonder if it could be cleaned up more by Jon+Noring · · Score: 2, Informative

    Yes, they could have "stomped" down harder on the CEDAR processor they were using. But the problem they faced is having so many recordings to restore, where the optimum noise-reduction settings for each recording will vary. If you set things wrong, you will remove a lot of the hiss (which usually is more like pink noise rather than white noise, thus a little harder to deal with), but then introduce a lot of annoying artifacts in the sound. Most afficionados of the early sound prefer to hear some hiss than to distort the sound any more than it already is. So long as the hiss is clean and free from pumping), it is acceptable.

    The important thing is that the project has placed the original raw transfers online in lossless format, so anyone may restore the recordings themselves. This is a major shift from other early recorded sound archives where the raw transfers are not preserved or made available. As I've noted elsewhere, UCSB has raised the bar in what early sound archives should do.

  7. Bang! Hunt is Over. by TheStonepedo · · Score: 3, Informative

    http://flac.sourceforge.net/

    Now you can save your time and bullets.

    --
    I'll be your candy shop of infinite deliciousity if you'll be my discotheque of endless rump-shaking.
  8. Re:Lossless compression? by ZipperLips · · Score: 2, Informative

    I believe the music curator and his staff have transfered their cylinder collection to a lossless format prior to cleaning. I say this not as anyone who knows what the hell I am talking about with regards to the audio processing aspect of this project, but as one of the system admins associated with the Davidson Library at UCSB, where the project is hosted. By the way, the front end is just an old gentoo box, please don't beat it up to bad. I don't want to work on the weekend!

  9. Have a Read of the Copyright Message by femto · · Score: 2, Informative

    It makes an interesting read/rant.

  10. Re:An unfortunate license choice by Ray+Radlein · · Score: 2, Informative
    On the other hand, there's this notice on the web site:
    The raw transfers created by the University of California are in the public domain. Users of this website are free to use these raw transfers as they see fit, not limited to redistribution to others, including distribution over peer-to-peer file-sharing networks; reissue, mashups, mixes for commercial or non-commercial purposes; or other uses that could be imagined.

    Restored versions of the audio files, including the downloadable MP3 files are © 2005 by the Regents of the University of California. They are licensed for non-commercial public use under a Creative Commons Attribution-NonCommercial 2.5 License.
    So it seems to me that all you need to do is download the original raw recordings yourself (rather than the MP3s, which are, as you mentioned, CC-attrib-noncommercial) and crunch them into MP3s yourself. More work for you, of course, but at least it should be doable.