MUSINGS: On the Digital Music Collection, Metadata Tagging, and Hygiene...

Happy New Year everyone! I hope the holidays went well and you're ready to take on 2017.

Today, I thought I'd spend some time talking about something extremely important and one which I'm surprised I don't hear more about looking around forums and audiophile watering holes... It's the topic of how one creates a collection of music. The difference between a collection and just plain hoarding is of course the discipline of organization involved in the collector's hobby. The collector knows what he/she has. The collector has mastery over the collection. Though responses may be variable, I believe a friend or even complete stranger would be able to appreciate the time and dedication that a collector has put into achieving this mastery as opposed to a sense of revulsion when faced with the hoarder (this is honestly the feeling I get when looking at this "collection").

Over the years, I have seen a handful of articles like this one which also introduces one to Picard, the free MusicBrainz software that will do the job in an automated fashion. If I were to start putting a new collection fresh today, I'll probably do something like this and grow from there, adding customizations, and checking accuracy along the way. However, I have been collecting CD's since the 1980's and over the years, especially after 2004, I have migrated all the "physical" music over to my music server. Through the years, although I have gone through multiple hardware servers, the data from the music collection really has not. It has been essentially rip once into a lossless format, and the CDs packed up in storage thereafter. As the years go on, I suppose like every collector, one develops a unique way to archive the albums, manage the directory structure, and a way to tag the files in a fashion that "works" for oneself.

Let's start the new year with a look at one way to manage the music collection (my idiosyncratic way :-). It has served me well and maybe some of what I do will resonate with you as well...

I. Directory Structure

Whether it's stamps, wine, LP's, etc. a collection is only as good as how well we can get to the "object" of interest. Since these are files we're talking about, and since I want it to be universally accessible across generations of hardware and in a form that can be imported by all kinds of software over the years, I need it to be flexible and easily inspected - easily understood by humans (ie. some kind of numerical system or descriptively lacking names will not work for me). The music directory structure has evolved over the years and these days looks like this (I will use <xxxx> to denote directory names):

<DSD>
      <Aleš Bárta - Toccata and Fugue - Organ Surround Illusion (DSD64 DST, Exton, 2001)>
            01 - Ales Barta - Bach - Toccata and Fugue in D minor BWV.565.dff
            02 - Ales Barta - Bach - Chorale 'Jesus bleibt meine Freude' BWV.147.dff
               ...
      <Arne Domnérus Group - Jazz At The Pawnshop [30th Ann] (DSD64 DST, 2006 SACD, 1976)>
      <Billy Joel - An Innocent Man (DSD64 DST, 2013 MFSL UDSACD 2094, 1983)>
      <A Far Cry - Dreams And Prayers (DSD128, Samples, 2013)>
            01 - A Far Cry - Beethoven Dreams Stereo DSD128 (2013).dsf
            02 - A Far Cry - Beethoven Dreams Binaural DSD128 (2013).dsf

<FLAC MCh>
      <Yello - Toy (Upconv DTS to 5.1 16-44, FLAC, 2016)>
            01 - Yello - Frautonium Intro DTS.flac
            02 - Yello - Limbo DTS.flac
               ...
      <Yu Hong Mei (于红梅) - Erhu Chant (闲居吟) (Hi-Res SACD to 5.0 24-88, FLAC, 2006)>
      <Roger Waters - Ca Ira (2CD's, SACD to 5.1 16-44, FLAC, 2005)>

<Music>
      <Rock>
            <The Beatles>
                  <The Beatles - 1 (FLAC, 2000)>
                        01 - The Beatles - Love Me Do.flac
                        02 - The Beatles - From Me To You.flac
                        03 - The Beatles - She Loves You.flac
                            ...
                  <The Beatles - A Hard Day's Night (1987 Release [MONO], FLAC, 1964)>
                  <The Beatles - A Hard Day's Night (2009 MONO Remaster, FLAC, 1964)>
                  <The Beatles - A Hard Day's Night (2009 STEREO Remaster, FLAC, 1964)>
                  <The Beatles - Abbey Road (1987 Release, FLAC, 1969)>
                  <The Beatles - Abbey Road (1983 Black Triangle SOX De-Emph'ed, FLAC, 1969)>
                  <The Beatles - Abbey Road (Hi-Res 24-44, 2009 Stereo Remaster, FLAC, 1969)>
                                             ...
            <The Rolling Stones>
                  <The Rolling Stones - 12 X 5 (2002 DSD Remaster, FLAC, 1964)>
                  <The Rolling Stones - 12 X 5 (Hi-Res 2011 HDTracks 24-88, FLAC, 1964)>
                  <The Rolling Stones - A Bigger Bang (First Pressing, FLAC, 2005)>
                                             ...
      <Pop>
            <Pet Shop Boys>
                  <Pet Shop Boys - Actually (First Pressing, FLAC, 1987)>
                  <Pet Shop Boys - Actually [Further Listening 1987-1988] (2001 Remaster, FLAC, 2001)>
                  <Pet Shop Boys - Alternative (2CD's, FLAC, 1995)>
                        <CD1>
                             01 - Pet Shop Boys - In The Night.flac
                             02 - Pet Shop Boys - A Man Could Get Arrested.flac
                             03 - Pet Shop Boys - That's My Impression.flac
                                   ...
                        <CD2>
                             01 - Pet Shop Boys - It Must Be Obvious.flac
                             02 - Pet Shop Boys - We All Feel Better in the Dark.flac
                                   ...
                  <Pet Shop Boys - Behaviour (2001 Remaster, FLAC, 1990)>
                                                ...
      <Classical>
           <Eiji Oue & MO>
                 <Eiji Oue & MO - Exotic Dances From The Opera (Hi-Res 2009 SACD to 24-88, FLAC, 1996>
                 <Eiji Oue & MO - Rachmaninoff Symphonic Dances (Hi-Res 24-96, FLAC, 2001)>
                       01 - Symphonic Dances Non Allegro.flac
                       02 - Symphonic Dances Andante Con Moto.flac
                       03 - Symphonic Dances Lento Assai.flac
                       04 - Vocalise.flac
                          ...
      <Jazz & Blues>
            ... 
      <Soundtrack>
           ...
      <Rap>
           ...
      <Hard Rock & Metal>
           ...
     <World>
           ...
      <Vinyl Rips>
           ...
      <LOSSY>
           ...
     <Singles>
          Baltimora - Tarzan Boy.flac
          
What I've done above is lay out a bit of the directory structure used for years now and the way I've named the files. The "root" directory contains 3 main directories - <DSD>, <FLAC MCh>, and <Music>. The reason should be obvious! Not every software system plays DSD files (i.e. Logitech Music Server does not recognize .dff nor .dsf). Multichannel music likewise is being streamed/handled by another system (such as using Kodi as I mentioned in my previous post with a TV box). And the majority of the music will be archived within the <Music> directory, standardized using FLAC (I imagine if I used an Apple more, ALAC would perhaps be the choice here), with a few MP3 and AAC files under the <LOSSY> subdirectory.

For the albums in the <DSD> directory, I label them as DSD64 or DSD128 as appropriate. Remember that years ago as discussed, I advocated for lossless compression of DSD and felt this is an important foundation to build on. Alas, as expected, DSD has not taken off (I still feel the foundational characteristics are poor and this is an impediment to mainstream acceptance), but I do have a number of albums collected over the years I felt was worth keeping in this directory. Notice the suffix I used for one of these albums - "(DSD64 DST, Exton, 2001)". This tells me the album is a DSD64, was compressed using the DST algorithm, I thought the Exton music label was worth mentioning in the event I look for this later, and of course 2001 was the origination year. For the Jazz At The Pawnshop album, notice I mentioned "2006 SACD" as the source to clarify which mastering this came from. As you might recall, .dff files are the only ones that can be DST compressed but there's no tagging metadata whereas .dsf can be tagged. Compression wins out for me since these DSD files are so inefficient on space requirements so if possible I'll do .dff with DST compression, and leave .dsf for files that cannot be compressed like the DSD128 albums (I have not seen any DST compression software that can handle them). I do want to give a shout-out to WavPack 5.0.0 for including DSD! Maybe down the road I'll convert these DSD files into that format provided there's software support in playback... DST-compressed DSD is playable with foobar using the SACD Decoder plug-in and also in JRiver.

In the <FLAC MCh> directory, all my multichannel music (anything > 2 channels) gets archived here. Even if the multichannel originated in DSD format like an SACD rip, it gets converted to PCM since ultimately bass management I feel is important in typical playback systems. Honestly, I think multichannel SACD was added on to help differentiate it from standard CD since the stereo sonic difference wasn't all that obvious nor necessarily better. Multichannel should always be PCM IMO. Anyhow, notice that I try to describe the source of the mutichannel. For example I found someone's custom DTS "upmix" of Yello's Toy and decoded it from DTS to 5.1 16/44. And an album I thought was worth keeping in high resolution like Erhu Chant (闲居吟) will be labelled as "(Hi-Res SACD to 5.0 24-88, FLAC, 2006)" referring to the 5.0 24/88 PCM format. We'll talk a little more about this below.

Unlike the aforementioned DSD and multichannel directories, because there are so many more albums in the stereo music collection, under the <Music> directory, I've subdivided them into the various <genres>. Under that I'll have a directory for <artist/band>, then the directory for <album>.

Notice that in instances of multi-disk albums like Pet Shop Boys' Alternative, I like keeping the actual CD's separate as <CD1> and <CD2>. Also notice that I have various different masterings in the collection - for example different versions of Abbey Road. I label the directories differently and when I've had to specially process an album like the old Toshiba Black Triangle version with SoX to perform digital de-emphasis, I'll say so in the directory name. This goes for other processed albums like HDCD decoding with dBPowerAmp or albums with special characteristics like QSound. It's also nice to know if an album is "MONO" (like A Hard Day's Night), again, these attributes will be put into the directory structure for ease of finding the actual albums/files.

"High resolution" albums also are labelled as "Hi-Res" in the directory name so that I can easily just search out all the album titles with this designation and know exactly how many of these I have. I usually see "Hi-Res" as albums of 24-bit depth and/or 96kHz and above. Remember that I'm rather picky about what I believe is worthy of hi-res status... Also, over the years I've downsampled many supposed 176/192kHz albums for having no evidence of value as well as many 24/96 albums (such as converting the recent live 24/96 Jimi Hendrix Machine Gun: The Fillmore East First Show due to high noise floor and no meaningful frequency extension into 16/48).

Classical music presents a bit of a challenge. Part of it is the length of filenames. For example, the "Eiji Oue & MO" is actually "Eiji Oue & Minnesota Orchestra" but because this would lengthen the file name, I'll usually abbreviate the name of the orchestra so as to highlight the conductor. Remember, if you're using a mixed OS environment, be careful of the longest directory+filename length that can be used. I believe in an NTFS Samba network environment, all the various level directory names + filename together have to be 255 characters or less. The thing with classical music is that I will populate the tagging metadata with more details which eventually will get imported into the library software.

I do have a number of <Vinyl Rips> I've done of pristine LP's, but since I view them as a different, lower level of quality to the usual CD rips, hi-res downloads, or SACD/DVD-A rips, I will put them in a separate directory with the view that if I find a better mastering or a good quality hi-res download, I'll delete the vinyl rip. This is the same rationale for the <LOSSY> directory where I typically keep the iTunes downloads I've accumulated over the years but will look at replacing them if a better lossless rip is acquired. <Singles> is self explanatory... Those miscellaneous and one-hit-wonders like Baltimora's Tarzan Boy :-).

II. Metadata and Tagging

A well arranged directory structure provides an easy way to find and manage the files, but of course we're also interested in adding details and maybe tagging files with notes and reminders. Plus tags provide standardization for importing this information into music databases. This is why good tagging is essential.

For years I've been recommending the dBPowerAmp software... Essential software that handles all my needs from CD ripping, error correction and verification with AccurateRip, automatic tagging from the start, and of course very easy file format conversions. Ripping with dBPowerAmp will typically import metadata from an online source (like AMG, GD3, SonataDB, MusicBrainz, freedb). As you see below, I'll fill in tags in a way that I like - similar in many ways to the filename and directory convention as above (here's Radiohead's classic OK Computer):

Another essential piece of software for me is the freely available Mp3Tag "Universal Tag Editor" to handle the ID3 metadata tags. Other than batch editing of fields, this program will also allow conversion from filename to tag and vice versa - great because I already have all kinds of information in the directory structure already as noted above.

As you can see, I make sure that the genre field is filled in correctly. Notice that in the example above, I will tag a hi-res album genre as "Hi-Res / xxxx" so I can search all the high-res stuff together. Likewise, I add a little suffix like "Rock / Live" so I can easily find "live" recordings in the future as a genre descriptor.

Album titles follow a similar convention as my directory structure:
     <Album Title> (<special information like remaster / hi-res>, <file format>)

I find it useful to know the "special information" like say if it's a "2009 Remaster". If it's an album worthy of keeping in hi-res, I might say something like "Hi-Res 2012 HDTracks Studio Master 24/96" as above. If it's a rip from another source like an SACD or DVD-A, I will indicate "Hi-Res 2002 SACD to 24/88" to indicate that it was from the 2002 SACD source and converted to 24/88 PCM. Likewise "Hi-Res 2002 DVD-A to 24/96" or "Hi-Res 2009 Blu-Ray to 24/192" as the case may be. I personally find this an easy way to search for pressings especially for albums notoriously known for all kinds of audiophile remastering efforts! For example, the Dark Side Of The Moon title in the tag could be "Dark Side Of The Moon (Japanese Harvest Black Triangle SOX De-Emph'ed, FLAC)" (we can talk about de-emphasis another time if anyone's interested in these old CD's), "Dark Side Of The Moon (1989 MFSL UDCD-517, FLAC)", "Dark Side Of The Moon (Hi-Res 2003 30th Anniversary SACD to 24/88, FLAC)", "Dark Side Of The Moon (Hi-Res 2011 Immersion Blu-Ray to 24/96, FLAC)" releases...

With most rock/pop albums, I typically will not bother with the details like composer... This field would of course make sense and important with classical music. Other fields like album artist I will use for compilations or to identify a main artist for the album. For example, these days one of the artists on the album might be tagged as "Maroon 5 & Christina Aguilera" but we know that as a whole, it's a "Maroon 5" album which then goes in album artist. The album artist field would also be useful when it's a true compilation (like say 2016's top hits) and I want to label album artist as "Various", leaving the artist field to reflect the artist for each song.

Typically, I'll also fill in the comments field as well. One common piece of data I will include is the DR value so I have an idea of how dynamically compressed the mastering is (using foobar plugin or if I can get it off the DR Database). Not only does it give me a sense of how potentially clipped an album is, but this is also useful in the future... For example, say I ripped the recently released The Weeknd album Starboy and see that it has an awful DR of 5 (sadly, it is 5), I'll stick this in the comments field as "Album DR5." In a few years, I might run into a "hi-res" version or a remaster of this same album and see that the DR is now DR10! I can easily look back at my rip and deduce that in fact the remaster is significantly different and quite possibly a better mastering job... In this fashion, I can pick out remasters potentially worthy of a purchase and weed out useless ones that severely compress the sound, especially nonsensical "hi-res" material that are just as compressed and worthless IMO.

III. Multichannel Music Tagging

As noted above in the directory structure segment of this post, I do have a number of multichannel 5.1 FLAC rips from my DVD-A, Blu-Ray, or SACD disks. These albums are all placed on a different directory (I call it <FLAC MCh>) outside of the main stereo <Music> directory. Since I don't have that many multichannel recordings, there was no point in having genre or artist/band subdirectories. The good thing about keeping multichannel separate is that this allows me to use the inexpensive Amlogic S905X TV box to catalogue and stream these through HDMI to my receiver as I discussed previously.

Here's an example of tagging used with DTS-to-5.1 FLAC from the DVD of Adele's Live At Royal Albert Hall from a few years back (conversion done with foobar's DTS decoder mostly but you see this example used JRiver):

Notice that I will always tag the album title with something like "DTS to 5.1 16/48" to let me know how the album was transferred and at what resolution. Also, the comment section not only contains information about DR value, but also the workflow used... In the example above, I used JRiver 19 to convert the DTS file to 5.1 24/48 PCM, then because I didn't feel it demanded 24-bit resolution, used iZotope RX to dither it down with its excellent MBIT+ algorithm.

As for the genre tag, notice I've added "Surround / xxxx" in the prefix. Again, little things like this done on a consistent basis helps me locate and identify album types better in the future.

IV. Conclusions and General Thoughts...

I know that this topic might appear a bit mundane. This is probably why there are few articles out there on it; audiophiles and the press preferring to spend time on the "latest and greatest" hardware. Nonetheless, this is a topic I've wanted to write about for some time! It's one of those "infrastructural" topics which as music collectors, we at some point have to reckon with in the hopes of getting it done "right". Just like nobody likes to think about backing up their music files, I honestly believe the quality of a "computer audiophile" is not in their dedication to the hardware, but the time invested in making sure the "hygiene" of the digital music library is "healthy". The music is after all the life blood to the hobby and IMO it would be ridiculous to not spend time on arranging, sorting, and curating one's collection. Not only will maintaining the "hygiene" of one's collection result in rewards like improving one's ability to efficiently search for music, but it also gives one freedom to be able to migrate between hardware platforms easily when there's consistency and internal cross-correlation between the directory structure and tagging. For example, even if a rogue music database program messes up my metadata and I lose some of this, I can always derive much of it from the directory structure. Also, a "human readable" and logically constructed directory structure is much better IMO than one created by a machine; there is inherent "knowledge" and familiarity when one is the architect of the collection.

Perhaps there is also an innate sense of value when one experiences the sense of emotionally investing in the collection even though it's virtual. It's taking ownership, probably a core reason of why we collect in the first place. Parallel to the digital one, I do maintain a small collection of vinyl numbering about 400 LPs these days and I do feel a similar sense whether the collection is "physical" or virtual. Of course, in the virtual world, I have collected thousands of albums over the decades, yet not have to worry about the hassles of organizing too many boxes and shelves around the house. (My wife I'm sure is thankful for that!)

The directory structure and tagging I've laid out here works well with Logitech Music Server, works when importing with JRiver, and even Kodi does a great job when pointed to the multichannel directory over the NAS. In the next while, I'll also discuss Roon which generally works with what I've done.

One last thing... It's always nice to have album covers for the music.

Over the years I've used artwork from Album Art Exchange where you can almost find any CD album cover in excellent quality. Usually I'll target for an 800x800 image - up to 1000x1000 if available, and typically will go no lower than 600x600. Within each album directory, I've standardized on the filename "Cover.jpg" which is recognized by all music server programs I've come across (another common filename is "folder.jpg"). These days when browsing albums using something like a high-resolution phone or iPad, I would certainly want the image to look good and not too pixellated!

There you go, a reasonably detailed look at how I do things on my home music server. As you can see above, I'm still using Logitech Media Server primarily as most of my music listening is still streaming to my Squeezebox devices or the Raspberry Pi 3 running piCorePlayer. Other times, I'll use JRiver and DLNA streaming to Volumio or RuneAudio among other options like Archphile.

Remember of course that although this is how I've been doing things after years of "evolution" catering to my preferences, like personalities, we'll each have to decide what works best... Our own idiosyncrasies embedded in the process is how we each impart a bit of ourselves in the manifestation and joy of the music collection hobby.

Final reminder: do backups. Make sure you have at least one other back up in a different physical location in the event of theft or God forbid, something happening to one's home... For the record, I actually do not use any special software for backups. I have 3 copies of all my music. I will manually copy over the gigabit network to the backup machine every once awhile with just a simple Copy/Paste, transferring files absent on the backup copy. Every 3-6 months, I'll do a copy to an external hard drive which can be stored elsewhere. At worst, even if I lose the 2 main copies, there's a recent backup... No big deal since the "essential collection" will be there and I don't buy as much music as I used to when starting out years ago.

Again, happy New Year everyone. Cheers!

May 2017 represent another year of enjoyment in the music collection and audiophile hobbies :-).