There seems to be a lot confusion in the online audio community over file compression.
First, we need to clarify the difference between file compression and audio compression.
Compression in the audio world is an effect used to regulate dynamic levels—generally lowering the dynamic range, making quiet things louder and loud things quieter to provide more even listening levels.
That’s not the type of compression we’re talking about today.
Audio compression is completely different from file compression, which is used to reduce the size of the audio files themselves.
Many folks are of the impression that any sort of compression damages audio quality. (While some even believe that simply uploading/downloading files will damage quality)
This is completely false, but it does have an understandable basis in fact.
All digital audio is merely binary 1s and 0s arranged to recreate sound. File size is determined by the number of bits of data (the 1s and 0s) making up the whole project.
In the early days of the Internet (and more prevalently at the time, private BBS systems), transferring and storing uncompressed audio was not really feasible on any large scale.
An standard 74 minute CD full of audio ran 650MB, which is actually 5452595200 bits of data.
In 1993 my hard drive was roughly the same size, so one’s digital music collecting abilities were limited to say the least, and my 2400 baud modem would have taken roughly 27 days to transfer all that data, in the event it miraculously was able to maintain 100% speed and could last a full month without my mom picking up the phone, getting an earful of static, and the connection dropping.
Out of necessity, the MP3 was created. For 1993, it was actually some amazing technology. Using a psycho-acoustic model of what humans perceive hearing wise, it was possible to keep only the necessary bits of data to recreate a reasonably close sounding facsimile while throwing away everything else, thus greatly reducing filesize, and likewise the storage space and bandwidth necessary to send them back and forth.
Much like the fax machine that bears a similar name (yes, Fax is short for facsimile!), while it’s an acceptable option in a pinch, the resulting product is merely a shadow of its original form.
The fact that the majority of the original data was thrown away made this process known as lossy compression.
For those living through that time, I’d draw an analogy to the difference between a VHS tape set to record to Super Long Play, versus today’s 1080p video.
Yeah, you could store 6 hours of Dr. Quinn Medicine woman on a VHS and watch it back at an acceptable quality, but next to an 1080p show from today it would be laughable.
While we’ve certainly raised our standards exponentially for video quality, audio still uses encoding standards that soon will be old enough to vote in an election.
Some DVDs are even still shipped out with 128kbps MP3 quality audio,synced to exponentially better 480p video data.
Where’s the love for audio quality?
Luckily, almost 20 years later, we’ve come a long way. My home Internet connection is 30mbps, which is 13,107.2 times faster than it was back then (no that number is not made up), and you can pick up 1TB of Hard Drive space for roughly $50 if you find a decent sale.
That 1TB drive can hold 1613.2 uncompressed full 74 minute CDs.
While lossy compression technology has made steps in recent years (M4A, also known to the Apple community as AAC), it’s not really relevant or necessary any more.
While it may allow you to store more songs on your iPhone (or in my case Droid X—get one. They’re awesome.), just as you wouldn’t send a movie you filmed into a video post production studio on SLP VHS tape, what sense does it make to send your audio in for post production work with more than 90% of the original recording missing?
The fact that the EQ, compression, etc. of the mastering process changes what you hear compounds the problem, since the MP3 is designed to only recreate what you hear in the original rendering. Any other data on the original recording that may provide a better listening experience, especially in the high end of the frequency spectrum has fallen into the nether thanks to the MP3’s psycho-acoustic modeling processes.
Thus, the MP3 has earned the disdain of audiophiles across the interwebs, and quite understandably so.
Don’t get me wrong, I listen to MP3s and M4A on a regular basis, but lossy compression certainly has no place in the post production world.
The only time lossy compression should ever be used is to encode a 100% finished final product into something for mass consumption (just as you’d never throw away the original film cells of Star Wars because you had a copy on BetaMax).
While lossy compression still holds the torch as far as reducing file size to the masses, there has to be a way to reduce audio file size while still maintaining 100% quality, right?
Enter Lossless compression!
In recent years, Lossless codecs have come to prominence in the audiophile world.
FLAC (Free Lossless Audio Codec) is the most well known, but many different permutations exist including APE and even Apple’s own version ALAC—I’ll let you guess what that stands for. (Apple always has to be different).
There are even specific encoders for packs of WAV files, most notably the aptly, although not too inspiringly named Wavepack.
While specifically designed to work great on audio files, these methods are at their core are descendants of much older file archiving technologies like ZIP and RAR.
Unlike audio, which can be lossy compressed and retain a similar sound, you can’t really just throw away 90% of computer files and still expect them to work.
It’s still possible to make these files smaller by looking through the data and finding any redundancies, for example noticing “Hey, there are 1000 0s in a row here. Why not just say ‘1000 0s in a row’ rather than storing them all?”
While this is an oversimplified explanation that I’m sure would make a computer scientist cringe and go back to watching his torrent of last week’s episode of Big Bang Theory, it gives the general idea.
In lossless file compression, no data is lost (It’s not just a clever name!) thus after being uncompressed, the file will be exactly the same as it was.
Even better, these methods use Cyclic Redundancy Checks (CRC) to verify the archives contents are 100% perfect.
While a WAV file could theoretically get cut off mid upload and have a portion of the track missing or corrupted, while using an archive any such missing or inaccurate data will immediately throw up red flags, thus ensuring what you sent is 100% what is received, just much smaller!
As lossless compression is based on redundancy, full length individual WAV tracks, which are full of repeating loops and long periods of silence benefit extremely from lossless compression.
It’s not usual to see a large project go from over 3GB of data uncompressed to as little as 200MB once archived. Or a full length track only containing a brief vocal recording or sample go from 100MB to 3-4MB.
While it no longer makes sense to throw away data to save space and bandwidth while doing post production of audio, lossless compression, both for bandwidth and storage reasons is an indispensable tool for producers, musicians, and engineers the world round.
And no, it doesn’t reduce audio quality!
I remember growing up, I was told to make sure not to sneeze near the computer so it would stay virus free.
Much the same, compression equating quality degradation is just a myth, just like Santa Claus, the Easter Bunny, and analogue equipment having some kind of magical fairy dust that just makes everything sound better… (Maybe the last one will need an article of its own?)