Bluffer's Guide to Audio Formats

It's only zeros and ones, but in digital music (as in life), the wrapping matters.

Generally speaking, the smaller the file, the more destructive the compression – and the greater the degradation in sound quality. This 'information-richness' is expressed as a 'bit rate' in Kilobits (not Kilobytes) per Second (kbs). A conventional audio CD (16/44.1 Red Book Standard) is just an archive of 1411kbs files.

Why 1411kbs? First, forget the '24-bit' claims advertised on the jewel case: regardless of how the master disc was created, every standard audio CD contains 16-bit files using a sample rate of 44.1kHz. Think of this as a data grid describing one second of music in one channel: 16 bits deep, 44.1 bits wide – 705.6 bits in area. Two channels therefore need 1411.2 kilobits to describe one second of music in stereo. For reasons not transparently clear, as a working rule, one minute of uncompressed 16/44.1 audio occupies about 10Mb.

In the Mac world, AIFF is by far the most common uncompressed lossless format for two channel audio; in Windows, it's WAV. Both formats are effectively wrappers for PCM (Pulse Code Modulation) data. You may have come across Raw PCM streams in BluRay and high end multichannel video applications. All support 24-bit audio at higher sample rates (typically up to 192KHz). AIFF and WAV are the highest quality media for playback.

Those familiar with image file formats will know that TIFFs can optionally be stored with LZW lossless compression, which saves an exact replica of the original file, without any degradation – yet file sizes are reduced by between 30-60%, depending on the dynamic range and homogeneity of detail. iTunes and other media players typically flag these files with bit rates of 300-1100kbs. Similar compressed, lossless, formats are used to store audio files: including ALAC (Apple Lossless or M4A), FLAC, APE (Monkey's Audio), WV (WavPack), WMA Lossless, and MPEG4. Once unpacked and error-checked, the data is truly identical to the original. Most of these formats support 24-bit files.

However, on-the-fly decoding of compressed lossless formats invariably results in audible degradation in the form of jitter: the zeros and ones may be the same, but irregular gaps are introduced between them. These formats are therefore ideal for archival, but flawed for live playback.

At the bottom of the heap, the MP3 is the world's most common – and ugliest sounding – digital music medium. It's the most ubiquitous of an array of compressed and lossy formats designed primarily to optimise file size at the expense of quality: bit rates range from 16-320kbs, with corresponding reductions in file size: up to 90%. However, the compression method differs from the benign, respectful pattern-matching efficiencies deployed in FLAC and Apple Lossless files: lossy compression methods' complete re-encoding creates a degraded simulation, with many subtleties shorn away. Treble becomes peaky and shrill; the mid-range is pushed forward; there is an audible loss of coherence across the frequency range; the soundstage is compressed; there is a timbral hardening, fine detail is smeared or erased, acoustic decays are truncated . . . etc. For the record, AAC and OGG lossy compressed formats typically sound better pound-for-pound than equivalent MP3s. VBR (Variable Bit Rate) encoding assigns less compression to more complex parts of a track (and vice versa) in order to improve its perceived sound quality.

In 2010 the concept of abusively compressing a file to that extent already seems anachronistic: among iPod users there has been a subtle upshifting to higher bitrates now that sufficient storage is commonly available for very large libraries: there's no point leaning that hard on files any more.

On the other hand, in-car hi-fi and inexpensive players - including the iPod, with its dismal output stage - are usually incapable of telling apart the good, the bad and the ugly.

On the other other hand, we may be on the brink of the long-touted era of HD music: 24-bit at large. And 24-bit files at a sample rate of 192khz yields a horrifically unwieldy bit rate of 9216kbs: exhausting the capacity of the largest current iPod Touch in a dozen albums. So maybe all that storage will come in handy after all . . .