| Polyphony Suite Documentaiton | ||
|---|---|---|
| <<< Previous | Archiving | Next >>> |
A file name can only have 239 characters in VFAT and NTFS, so we will use that as our limitation.
Bear in mind that .mp3 steals 4 characters, and artist name, album name, and track number can eat up a good chunk of the 239 as well. It's not uncommon to max out the 239 character limit with classical music, so be careful there.
A directory name can only have 227 character in it (in NTFS, and I'm assuming in VFAT as well). It is unlikely the name of an artist or an album + year of release will exceed that. If it does, the track listing will be screwed (since we'll only have 239-227 characters to name tracks!), so we may want to use an abbreviation instead.
VFAT is the most restrictive here, so we'll ban the use of these characters:
/ \ : | = ? " ; [ ] , ^
An interesting headache I encountered: I ripped and encoded a bunch of music on a Windows 2000 machine. I then moved all of this music to a Linux server via Samba (fairly stock install). That worked fine, and if I looked around at the files in Linux, thinks were ok.
After creating a Samba for the music on the Linux box, Windows decided to freak out if the name of a file or folder contained a high-ASCII character like "é". Long story short: Samba uses a default character set of UTF-8, and Windows uses code page 1252. Windows can't seem to read certain UTF-8 characters in folder/file names, so it goes into some unknown state when you try this.
The fix is to add the following command to smb.conf in the [globals] section BEFORE you copy your files over: unix charset = ISO8859-1. This instructs Samba to use Latin-1 encoding to write file and directory names, which in turn keeps Windows happy when it goes to read the filesystem.
If you did what I did and copy the music over before fixing Samba's config, you'll need to manually search for and replace any high-ASCII characters that may have been converted to UTF-8, and change them back to ISO8859-1 characters. Yes this sucks, the Unix find command is useful to do this, as is File::Find in Perl (you could automate the whole thing this way), and even Windows' search program works.
I don't know enough about Unicode in filesystems to be coherent here, so if anyone has any comments, let me know. Obviously I'd love to support it to maximize internationalization of the software, but at present I'm just not that interested in researching it since I won't actually use it.
I've somewhat arbitrarily chosen ## to be the delimiter in directory and file names. This just makes it easier to parse :). I figure ## is pretty rare in album, artist, or song titles, so it's probably a sensible choice.
The year of the original album release, then the name of the album. i.e.:
Cymande/
1973##Cymande/
Blackalicious/
1999##Nia
|
Polyphony naming conventions follow one simple guideline: if you sort MP3s in the filesystem by name, they should appear in a sensible order. With this in mind, you just need to keep the most significant elements of a track or album near the beginning of its name. As you get closer to the end of the name, the parts should be more likely to vary.
i.e. For Phish's Hampton Comes Alive, there are six discs, but should be organized as two sets of three, since they're songs from two different concerts. In my system, the tracks from all six discs live in the same directory. It makes sense to have the date of the show as one of the first things in the name of a track, then the disc number within that date when sorted, tracks from the same disc will sort together, which is the goal. Track number and track title are usually the last two things in the name of an MP3, since they vary the most.
Observe: 11-21-98 show, Guyute is the 5th track on disc 2. The name of the MP3 in this case would be:
Phish##Hampton Comes Alive 11-21-1998 (Disc 2)##05##Guyute.mp3
There are some subtleties in this particular example (i.e. including the disc number in what would normally just be the album title), but it should give the idea.
In each mp3 for the album, append "(Disc n)" to the end of the album name.
Pink Floyd##Is there anybody out there - The Wall Live (Disc 2)##09##In The Flesh.mp3
Pink Floyd##Is there anybody out there - The Wall Live (Disc 1)##10##What Shall We Do Now .mp3
MM-DD-YYYY
We convert the dates to a standard format for a few reasons:
Many file systems do not support the slash as a valid character in a filename.
We can automatically parse dates out of album and track names for use in meta data.
No confusion on which numbers represent the month, which is the day.
General consistency.
EAC format string: %A\%Y##%C\%A##%C##%N##%T
e.g. In the filesystem, the following type of directory structure would be seen:
Radiohead/ 2001##Amnesiac/ ... Radiohead##Amnesiac##06##Knives Out.mp3 ... |
EAC format string: Various\%C\%Y##%C\%C##%N##%A##%T
e.g. In the filesystem, the following type of directory structure would be seen:
Various/ Live From Bonnaroo/ 2002##Live From Bonnaroo/ ... Live From Bonnaroo (Disc 2)##07##Ween##Bananas & Blow.mp3 ... |
Notice that we put all various artists' albums under the same "meta-artist" called "Various". Soundtracks would also go in this directory.
Given an arbitrary MP3, we can determine that it is from a "various artists" album by checking that the second field (when split on the "##" delimiter) is the track number.
Lord knows there's no shortage of info out there already on how to name live shows. There's equally as much info for recording shows. I like etree's naming standards, and hope to support it down the road, but for now we'll stick to a scheme similar to what is being used for other album types.
[BUGBUG: this whole idea is only at the proposal stage!]
The naming format will be very similar to a normal album, the main difference being that album titles will always start with a MM-DD-YYYY date followed by a venue name. i.e.
Medeski Martin & Wood/ 1993##It's a Jungle In Here/ 2000##Tonic/ 1995##07-12-1995 - Bluebird Theatre, Denver, CO/ Medeski Martin & Wood##07-12-1995 - Bluebird Theatre, Denver, CO (Disc 1)##01##Dracula.mp3 Medeski Martin & Wood##07-12-1995 - Bluebird Theatre, Denver, CO (Disc 1)##02##Sequel.mp3 Medeski Martin & Wood##07-12-1995 - Bluebird Theatre, Denver, CO (Disc 1)##03##Chinoiserie.mp3 ... |
It is a given that sometimes you just won't know exact info for a live show. Sometimes the tracks melt together, or it's a weird jam, or you got a CD with no info from some guy, or whatever. I've decided to standardize on how this missing data should be represented.
If the track/song title is unknown, call it [Unknown].
If the venue is unknown, call it [Unknown Venue].
If any part of the show date is unknown, fill the missing characters with lower-case x. i.e. Stevie Wonder##03-xx-1972 - [Unknown Venue]##01##[Unknown].mp3
We organize classical music in the filesystem by composer. Our database of meta info can provide cross-references to conductors, featured soloists, etc. For classical compilation albums we follow the various artists naming scheme -- with track composer as the artist.
BUGBUG: I'm very open to ideas on how to do this better.
e.g. I have the Deutsche Grammophon edition of The Planets by Holst. It's performed by the Berliner Philharmoniker, conducted by Karajan. That's just too much info to encode into the filesystem, so we strip it down to composer and albumb title:
Gustav Holst/ 1981##The Planets/ Gustav Holst##The Planets##Mars, The Bringer of War.mp3 |
Franz Liszt/ 1997##The Complete Etudes/ Franz Liszt##The Complete Etudes##S.139 1. Prélude. Pesto.mp3 |
| <<< Previous | Home | Next >>> |
| Archiving | Up | Ripping |