LossyWav Pre-Processor for Increased Lossless Audio Compression Rates

LossyWAV is a variable bit-depth pre-processor for lossless encoders, which takes advantage of the "wasted bits feature" of current lossless encoders such as FLAC, TAK and WavPack, leading to significantly higher rates of compression whilst maintaining transparency.

Although LossyWAV is technically lossy, it's more like a lossless format since it's transcode-safe (you can use it as a transcode source for transform codecs such as MP3, AAC, OGG etc.) and it's future-proof (i.e. once processed with LossyWAV transcoding to other lossless formats is a lossless process, so you can move from TAK to FLAC without any quality loss).

LossyWAV makes significant reductions on almost all music except piano music, for this reason we recommend treating solo piano music as a special case and encoding it losslessly (of the lossless codes we like TAK the best, but for compatibility FLAC is still the way to go).
 

Lossless compression can vary greatly depending on the music

We used a sample of 9 tracks, each 60 seconds in length, to simulate a varied music collection, plus one reference track of white noise. The 9 tracks in the sample set are:

• Classical - Quiet Piano [Mozart's Piano Sonata No.4 in E flat major - I. Adagio by Uchida]
• Classical - Busy Piano [Bach's English Suite No.2 in A minor I. Prelude by Gould]
• Classical - Symphonic, Loud [Mahler's 2nd Symphony - I. Allegro maestoso by Haitink]
• Jazz - Quartet, Soft [Coltrane Quartet - Say It (Over And Over Again)]
• Jazz - Orchestral, Loud [Charlie Mingus - Hobo Ho]
• 60's Pop - Moderate (Original CD Release) [The Doors - Break On Through]
• 60's Pop - Loud (Remastered CD Release) [The Doors - Break On Through]
• Modern Pop - Loud [Songs: Ohia - I've Been Riding With The Ghost]
• Modern Pop - Very Loud [The Flaming Lips - Race For The Prize]

The tracks were encoded with:

• FLAC 1.2.1 (-5)
• TAK 1.1.1 (-p4m)
• LossyWAV 1.0.1.0 (--standard) > TAK 1.1.1 (-p2m -fsl512)
• MP3 LAME 3.98 (-V 2)

Here's how the various encoders fared (click on the graph to enlarge):

Here's the actual bitrates:

So, how much space am I going to need?

A 1000 CD Collection @ 60 mins per CD, uncompressed will require 591 GBs of HDD space (44100Hz x 16bits x 2 channels x 3600 seconds x 1000 CD's). Based on the compression rates of our small sample this is how much space would be required to store a 1000 CD audio collection:

• WAV = 591 GBs [Uncompressed]
• FLAC -5 = 348 GBs [59%]
• TAK -p4m = 332 GBs [56%]
• LossyTAK --standard = 189 GBs [32%]
• MP3, LAME 3.98 -V 2 = 84 GBs [14%]
• MP3, LAME 3.98 -V 5 = 61 GBs [10%]

Now multiply that by 2 to include your backup.