Digitising my CD Collection

June 2011

Background

I've ripped my CD collection to digital music files on my computer 3 times now. The first attempt started in 2003 when I ripped all my albums to Vorbis OGG files stored on a small notebook hard disk in a USB enclosure. This allowed me to listen to any of my CDs at home and work, and was satisfactory for a time.

In 2006 I was given an iPod nano (2nd gen) for Christmas. It then that I realised that my OGG collection wasn't compatible and it dawned on me that as a format for portable players, the OGG format has failed. This left me doing what most iPod owners do, feeding iTunes all my CDs to produce a library of AAC files which I'd selected to rip at a generous 320kbps.

For a long time my iTunes collection was good until gradually I found my collection no longer fitted on my iPod. At this point iTunes becomes a real bore to use, continually prompting to disable podcast syncing or offering to throw a random selection of files onto the iPod. The only sane way to use iTunes at this point is to select the "only sync checked items" option and fastidiously manage which tracks you want on and off your device while clicking through the various warning boxes iTunes likes to produce.

It was August 2010 when my iPod finally broke. I'd kept it in a little case and treated it with the utmost of care but one day it was struck by a white screen of death failure. Posts on the Internet show this is a common problem and offered various fixes, but none worked for me. Fuming at Apple's fragile device, I decided I'd never buy an iPod again before promptly pre-ordering a new 16GB 6th generation nano which was shortly to be released.

The 16GB nano was initially a good, but as my CD collection has continued to grow, I found that I'm back to messing around with iTunes to try and make everything fit.

It was at this point that I realised that ripping everything at 320kbps was a mistake as and the files were much larger than needed to power the puny earbuds on my iPod. Re-encoding the files at a lower bit rate is possible, but conversion from one lossy format to another would cause the resulting audio to be worse than if just encoded directly at the desired bitrate.

It was at this point I decided to start again, this time for the last time.

Requirements

Having ripped my collection badly a couple of times before, I decided to try and future proof the next version of my music library and came up with some requirements, summarised as follows.

Using an open format means that there should be good options in the future if I wish to re-encode any files in another file format, while using a lossless format ensures that transcoding can be done without any additional loss in quality.

Requiring compression is simply a matter of practicality; a CD full of audio takes up a lot of space, all of which requires a back up. For example, audio CDs record 16-bit samples at a rate of 44100 times a second (44.1KHz). A stereo CD therefore contains 32-bits x 44100 = 1411200 bits = 176400 bytes = 172 Kbytes of data per second with a 60 minute CD having about 605MBytes of raw audio samples.

My final, and perhaps most important requirements are that I can get the music easily onto my iPod with track names and artist details intact. Realistically this means that iTunes must be able to handle the files and ideally will pick up the tags and cover art along the way.

Method

Being particular about these things, I was unable to find a system that satisfied my requirements. I did find lots of good bits and pieces, so I came up with the following setup:

Block diagram of ripping process

This uses a Linux NAS server running CentOS together with two programs I wrote, ripright and aifffffs. Together these programs help move music onto my iPod in two stages. The first is the ripping of CDs to store them on the hard disk of the Linux server using ripright:

This process is largely automatic and only requires user intervention if the CD is not uniquely identified by Musicbrainz or, optionally, if cover art has not been set. In such a case the CD is immediately ejected and Musicbrainz' own Picard tool can then be used from another computer to lookup the CD and enter details or cover art into the Musicbrainz website as needed.

Once ripped, the FLAC files can be loaded onto my iPod using Aifffffs, Samba and iTunes:

Results

Having ripped all my CDs using ripright, the following are some statistics on my collection:

CD TypeCountTracksTotal FLAC (Megabytes)Total AIFF (Megabytes)Ratio
Albums22627537787913005859.8%
Compilations59882243863859463.2%
EP645976160960.7%
Live6571517230565.8%
Other11244060.0%
Remixes8842479429257.7%
Singles88334116861839363.5%
Soundtracks6651270224556.6%

The average album uses 344.6 Megabytes when FLAC encoded. Compression reduces each file to about 60% of it's original file, noting that the files include JPEG cover art and tags which are not subject to compression by FLAC (i.e. the actual FLAC compression ratio will be slightly better).

The following gives the sizes of each album in my collection after compression. A subset of album names are displayed on the y-axis.

Sizes of various FLAC encoded Albums
Bar chart of album sizes
Megabytes

References


This page is maintained by Michael McTernan