Digital Music Management – Taking Full Advantage of Your Data

You don’t even want to begin to re-catalog THIS.

One of the challenges of a 77,000+ song library is the hundreds of artists a user may never hear. I decided to tackle this dilemma this week and found a solution.

The first task was to shed my database systems and Explorer navigation and find a way to manage my library from within a single application and really take command of the music metadata.  I decided that, moving forward, I would reserve my database strictly for physical media, and put Media Monkey Gold to the test of managing my digital library.

To facilitate the process, it seemed appropriate to lose some dead weight. I spent the first evening methodically rounding up my adolescent-era artists and relocating them to a folder outside the primary directory, aptly titled “Plebian Audio.”

Let’s get to work.

The process of elimination was simple.

If %GENRE% includes “rock” but does not include “progressive” or “avant-garde” then [Plebian.]

If mean album track %length% < 210sec then [Plebian.]

If demographic age < 20 then [Plebian.]

Else [Primary Audio Folder]

The resulting folder was a pool of p2p-era albums. Many had not been case-corrected. The file naming conventions were anything but uniform having sourced the tracks from multiple users, and not a single artist’s folder was a complete discographic archive. These were the millennial mix-tapes of the early days of file-sharing.

The next step was to clean-up the remaining digital files. For beginners considering this task, I recommend this article to get you started.

This was followed by an evening of fine-tuning – tailoring the %GENRE% values to be consistent throughout the library.

Trip Hop / Trip-Hop
Neo-Classical / Neoclassical / Modern Classical
etc.

It was at this juncture that I made the most valuable discovery of the project – that semicolons can be employed to apply multiple values to the %GENRE% field (or to any other field) in an ID3 tag.

I used Florian Heidenreich’s lightweight but powerful MP3tag utility to batch-replace all comma separators in my library with semicolons. The process took 36 minutes, and I confess that I did not take my eyes off that progress bar for the entire duration of the task. I just kept thinking of the incredible control that multi-genre categorization would offer. For the very first time I was able to navigate my entire catalog by genre without having to tie albums to one specific tag.

The Mp3tag interface

It was a difficult lesson to learn, but digital music inherently stores all its data within its ID3 tags, making an external music database redundant. I’d dedicated considerable time to indexing my vinyl, CD and digital music into a single database, but the reality is that it stagnates your data. The realization was reinforced by the fact that private trackers no longer have a need for massive discographic torrent archives, and have instead replaced them with decentralized manageable and crowd-sourced collages. If for example a newly remastered edition of a recording is released, it can easily be introduced to the library and is seamlessly integrated into the collage with a single click – a vast improvement to the previous method, which would require reconstructing and redistributing the entire static archive.

The same principle applies to local digital music libraries. Changes or updates to tag information, whether for a single track or for 77,000 tracks is a simple task, and is instantly reflected within your digital music application.

Perhaps my epiphany is common knowledge for the general public. I’m curious, because mine is the only library I’ve ever managed. I’d be interested in seeing statistical data with regard to what database solutions the average music consumer uses to manage their own personal libraries.

Discogs.com appears to be the most popular tool for record library management at present, with the particular benefit of real-time value assessment based upon up-to-the-minute sales data from the Discogs marketplace. And recently, Libib.com has premiered as a web-based library application to manage users’ book, movie, music and video game collections. I contacted the site’s support team, but was disappointed to learn that they are entirely UPC and ISBN-based, with no plan to implement record catalog numbers or digital music data in the foreseeable future (alienating perhaps the two fastest-growing collector demographics.)

But whatever software you use for your digital library, be it foobar2000 or MediaMonkey, the applications of audio metadata are empowering. For example, I could use this new multi-genre data to host multiple web radio stations (providing I can set up a server with sufficient dedicated bandwidth.)

Which brought me to my next discovery – the concept of autoplaylists. By entering criteria such as BPM range, genre keywords, or targeted folders, MediaMonkey’s autoplaylists will automatically populate qualifying tracks, relieving the user of the burden of manually-adding new music to existing playlists.

Unfortunately, my current personal media server does not support autoplaylists, so I am considering moving from Subsonic to a MonkeyServer system. This reinforces my realization that the MediaMonkey Gold may be a single-solution for all my music management needs.

Over the next two days I created 85 primary playlists to showcase my catalog, based on the library’s most prominent genres and for artists between 30 and 381 albums to their name. With an average of a 258 hour running time per list, these playlists totaled more than 2.5 years of continuous 24hr playback and will automatically update as new qualifying recordings are added to the library.

The immediate advantage is that these playlists will surely expose me to hundreds of artists I might otherwise never have heard.

In the evenings ahead I’ll be experimenting with MM community plug-ins to really get some analytical power out of the application.

So there you have it. With a weekend project-duration of approximately 20 hours I’ve re-evaluated my cataloging system, standardized the metadata of 103,633 files, retagged every recording, constructed 85 autoplaylists and relocated my database from a static system to a dynamic environment with an incredible number of potential applications.

Below are the first pieces of data I’ve exported from the new library.  These charts were constructed based on %GENRE% data clustered by autoplaylists with run times ranging from 10 – 8,501 hours.

Happy listening everyone.