Saving music wisely: Why save 'Queen - Bohemian Rhapsody.mp3' millions of times? [closed]

Question

As far as I'm concerned, Queen's song 'bohemian rhapsody' is one of the most popular songs all time. But for the purpose of this message you may replace this with another track.

At the same time I think 60% of the digital-music listeners have this track. Sometimes we have multiple copies: different versions of the track, different devices, unwanted duplicates in download folders, itunes folders etc..

Wouldn't it be much smarter to store these songs only once?

You can imagine various solutions for this.

How would you accomplish this?

Some criteria that may help you find an answer:

• It must reduce disk space
• It must remember which music belongs to you (DRM)
• It must use network traffic efficiently

Answer 1

Here's the negative answer: Don't.

The reason I have 60 copies of "Bohemian Rhapsody" is that there are 60 different recordings of it. Live recordings, studio takes, cover versions, remixes. Each one is different, and I want them all _{^{(I want them all, I want them all, and I want them NAO)[*]}}.

• Saving space isn't important to me as a music collector with terabyte drives. It's more important to me to have all the recordings than it is to save disk space.

  Instead, any space savings come from placing those recordings on a file server on the LAN. By playing back over the network from other connected devices, instead of copying the entire collection to those devices, I avoid the need for each playback device to have its own terabyte of storage.

• Duplicates management is fairly simple: there's a Downloads folder and a Music Library folder. In the Music Library, the album is the primary unit. Tracks that are the same recording released on different albums are considered different tracks.

  Everything goes into the Downloads folder first. It is vetted for quality, has any metadata sanitized and verified, and is compared to any existing duplicates in the Music Library proper. Any tracks of equal or lower quality to existing Library tracks are simply deleted from the Downloads folder; any tracks of greater quality replace the existing Library track.

• DRM is simply not allowed. All files are MP3 for device compatibility.

  If tracks are purchased online, they are purchased in DRM-free, Lossless formats. These files converted to MP3, and then the lossless files are archived elsewhere.

• Access is provided via read-only network shares to devices on the local LAN. Devices are not given direct write access to the files, to avoid tag corruption.

  Library server software (Slimserver, Ampache) is used to provide a web interface. Client-side software (Winamp, iTunes, XBMC) can access the shares for playback.

  Access from the internet at large would be provided by VPN if I had a need for that.

• Playlist management is completely up to the client device. Users are given a home directory on the server to which they can save playlists, separated from the actual Library folders.

_{^{[*] It's a Queen joke. Laugh.}}

Answer 2

Spotify is one solution to this - if everyone used it without owning the files themselves.

Answer 3

Theoretically, and assuming a lossless world (FLAC/WV/etc.):

1. If it's DRMed clean it.
2. Use tags as a heuristic to search through existing collection for possible duplicates, or use skillful hash algorithms.
3. Match audio data with more specific file comparison between old & new.
4. Distill the new track by subtracting off the old sample values if they're close enough. Recompress the new track. Store a reference to the old track.
5. When playing back, rebuild the initial input file, or playback both at the same time.

4 + 5 either require a new audio format or a custom tag schema to be recognized by the playback mechanism. This would save space on duplicates, if they were sufficiently close. Can't imagine it would end up being worth the time investment though.

If we're looking at a large ad-hoc network, there's a necessary conflict between file space used and resilience. I'd think the key would be to minimize network load to any individual and therefore would want to spread the files widely. Lots of work done on this kind of problem.