Back in the 1980s, a new audio format was invented at a research facility in Germany called the Fraunhofer Institut. It was called “MPEG-1 Layer III”. A patent was issued for the new format by the German government in 1989 and Franuhofer would receive an American patent for their format in the United States on November 26th, 1996 – by which time the format was simply known as “MP3”. And so time rolled on… until 1997, when programmer named Tomislav Uzelac developed the “AMP MP3 Playback Engine” for his company Advanced Multimedia Projects – get it? Advanced Multimedia Projects… AMP? Anyway, AMP was the first MP3 software player to hit the market, but MP3 had yet to take off. And so time rolled on… until two Americans – Justin Frankel and Tom Pepper – would take part of the AMP code, give it a nice Windows interface and call it WinAMP. And time rolled on yet again – until 1998 when newer versions of WinAMP became more stable. Finally, the MP3 format would become all the rage.
And it’s easy to see why. Back in 1998, many people still had puny 3GB hard drives and slow 28.8k modems. The ability to transfer an entire song to your computer in “only” 30 minutes just seemed plain magical. And the fact that the song would typically be only 3 or 4 megabytes in size – in the days before CD burners and 200GB hard drives – meant that the average person could easily keep a hundred or more songs on his or her PC. Of course, people with access to high-speed networks and huge amounts of storage on servers and networked hard drives – college students and IT workers – could have entire jukeboxes at their disposal.
Of course, times have changed. The average desktop PC can easily have 500GB of storage space or more. Pretty much any American in an urban area can get a high-speed Internet connection at home. Unlike those early days, even the most basic computer for sale in America today can encode an MP3 in seconds, store tens of thousands of them on its hard drive and download them from certain sources in seconds. During this time one thing has not changed though – the MP3 format. Back in the day, most people considered MP3s to be “good enough” given the limited storage space and bandwidth most folks had to deal with. But that has all changed – so why hasn’t our music format?
Granted, of all the human senses hearing might just be the most subjective. After all, what sounds fine to me might sound awful to you. To this day my poor Dad can’t tell the difference between a cassette and CD recording. I honestly believe the only reason he converted to CD in the first place was that he had a difficult time finding cassettes in stores! And my sister is extremely sensitive to low-frequency sounds. Mechanical “humming” that I can barely even hear will drive her insane. The point of me telling you about my relatives is to say that MP3s might sound fine to you. If so, move along… there’s nothing to see here. If, on the other hand, you find MP3s to sound somewhat “hollow” or “tinny” or just “not up to snuff”, then this article is for you.
A Tale of Two Compression Schemes
Just about every piece of music listened to on a PC is compressed. Now, although there are hundreds of specific compression schemes out there for shrinking the size of every kind of data, there are really only two major types of compression out there:
Lossy – a compression scheme where data is discarded as a means of compression.
Lossless – a compression scheme where data is not discarded as a means of compression.
Allow me to give some examples that will illustrate the difference between the two.
Lossy – Like MP3, the popular image format JPG is a lossy format. If you take a large JPG image (say 1024×768 pixels), shrink it to a fraction of its original size (to say, 102×77 pixels) and save it, huge amounts of data will be discarded. The file size might shrink from 500KB to 15KB because most of the original data has been thrown away to save space. If you then take the new, smaller image and blow it up to original size again the picture will look awful – because most of the original data that contained information about the picture has been discarded. Check out this example using the New York City skyline:
Original picture
(click to enlarge)
Shrunken picture
(Actual size)
Shrunken picture, enlarged to original size
(click to enlarge)
MP3s work in exactly the same way. Luckily, most lossy formats can be tweaked to allow for different compression ratios. In the case of MP3, you have a choice about which bitrate you’d like to use. A low bitrate – such as 64kbps – produces files that are half the size of a 128kbps one – with an appreciable loss in sound quality. A file encoded at 256kbps will be twice as large as one encoded at 128kbps – the former “gold standard” for MP3 files traded over the Internet. It will sound much better than the 128kbps, but still, because the MP3 format is lossy, it will never sound as good as the original.
Lossless – Take a Word document – any Word document – and compress it using WinZIP, WinRAR or your own compression program of choice. Notice that the file size has been reduced considerably – typically 80% or more for documents that have no images. Now unzip the document and open it. It’s all still there – every letter, every word, every period. That’s lossless compression.
Lossless Audio Formats
There are several lossless audio formats out there. Because I have chosen to standardize on the FLAC format, this is the lossless compression scheme I will discuss here, although I will mention others at the end of this article. To understand the difference between lossy and lossless compression as it relates to music, it might be helpful to briefly discuss how music gets from your audio CD into any particular format.
Back when MP3s first began to take off, several programs were needed to convert an audio CD into an MP3: a “ripper” (to copy the audio data from the CD ), an “encoder” (to compress the audio data into MP3 format) and a “tagging” program (to add the artist’s name, track name, etc. to the MP3 file). Later on several “all-in-one” applications would come along and combine these into one handy application. However, to this day all MP3 programs work in exactly the same way: audio is ripped from the CD, compressed into MP3 format, then “tagged” with the appropriate information.
When someone says that they are “ripping” a CD, when they mean is that they are copying the information from the disc to their hard drive. Data on audio CDs is contained in “PCM format”, which stands for Pulse Controlled Modulation. To greatly simplify how PCM works on audio CDs, a digital “snapshot” is taken of the music being played. 44,100 of these “snapshots” are taken for each second of audio. When played back at the same rate, these snapshots combine to create music in much the same way that each frame of a movie creates the illusion of movement. Now, none of this is really important to this discussion save the fact that WAV files – Windows native sound format – are also PCM audio. Every sound that Windows makes is in the same format as an audio CD – except that most Windows sounds are much smaller than CD data due to a lower sampling rate (fewer “snapshots” per second) and because most of them are mono (and thus, by definition, require only half the data as their stereo counterparts).
So when a CD is “ripped”, you will have several WAV files on your hard drive that are as close to perfect copies of the audio CD as is possible. If you burn an audio CD at this point using these WAV files, you will have an exact copy of the original audio CD. And by “exact” I mean “a copy which could pass laboratory testing to prove it sounds exactly the same as the original”. The only problem with these WAV files is that they’re huge. Audio in this format averages 10MB per minute, so a 5 minute song would be around 50MB. Back in the day, this would take 35 floppy disks or half a Zip disk to transport and copying it via 28.8 modem would take around 5 hours. So it’s easy to see why MP3 took off back then!
At this point back in 1998, most people would convert the all of the audio CD to WAV files and then convert the files to MP3 (most “all-in-one” MP3 programs people use these days will rip and encode each file in succession to conserve disk space). Describing how MP3 works is far outside the scope of this article, but in a nutshell the encoder discards information that humans have a tough time hearing. For example, the PCM format allows plenty of space to record sounds at frequencies many humans have trouble hearing. By getting rid of audio data that most people can’t readily hear, the MP3 encoder saves space. Note that this also means that your friend that swears that MP3s sound awful isn’t crazy – he or she just has sensitive ears and is able to hear the audio bits that are missing from an MP3 file.
So – MP3s save space by discarding “non-essential” sounds from the audio file – in much the same way as the JPEG samples above delete non-essential data. A problem then happens when those MP3s are converted back to WAV format for burning to audio CD. You will experience the same loss in quality with your sound files that I showed with the New York skyline pictures above. If you were to rip an audio CD to MP3, then burn it back to an audio CD, then convert it to MP3 again, then burn it back to an audio CD you will witness the same type deterioration in quality you’d see in a copy of a copy of a copy of a video tape. Once a song is converted to MP3 you will never again be able to listen to it at the same fidelity as you could with the original WAV files. Or can you?
Enter FLAC. A lossless compression scheme like FLAC doesn’t discard information like MP3 does. How it works is beyond the scope of this article, but it suffices to say that FLAC works in much the same way as compressing a Word document in WinZip. FLAC reduces a WAV file by 50% by default. At that rate, an album that might have taken up 650MB in WAV format will only take up around 325MB in FLAC format. This is still quite large compared to MP3 but the benefit is obvious – although compressed, files in the FLAC format can easily be converted back to WAV files and perfect copies can be burnt to audio CD at any time. But FLAC isn’t just for archiving – the FLAC installation (it’s free!) comes with a plug-in for WinAMP that allows you to listen to FLAC files as transparently as you listen to MP3s. Files compressed with FLAC sound as full and rich as the audio CDs they came from. Of course, this isn’t something you can read – you have to hear it! So why not download and install FLAC (and WinAMP if necessary) then download these samples:
Saint Etienne’s “Statues” in 128kbps MP3 and FLAC format (4.32MB, zipped)
The song above is short (45 seconds) and is perfect for describing what I’m talking about. Find a quiet time and some headphones to listen to the two versions of the song. The song is pretty “quiet” and the MP3 version shows every problem I’ve discussed with the MP3 format… It sounds “muddled” and it lacks the “definition” of the FLAC version. It also sounds “thin”, which I can’t really describe, but know it when I hear it. The FLAC version – which comes in at about half the size of the WAV version – to me sounds exactly as good as the original. Almost as importantly, converting the FLAC files back to WAV is as simple as dragging the FLAC files to the “FLAC front-end” window (click here for a screenshot) and clicking the “decode” button.
So – how do you use FLAC? First you need to go to the FLAC website and download the installer. Next, you need to figure out if your ripping application allows you to rip “uncompressed”. I can’t tell you if yours does or not; you’ll have to search the options in your ripping program to find out for sure. Or better yet, you can download and use the “geek standard” program for ripping audio CDs – Exact Audio Copy. Although EAC is only a “ripper”, it can can easily be configured to create files using the world-class LAME MP3 encoder or FLAC. To use EAC with LAME, simply unzip the LAME ZIP file to some permanent location on your hard drive, then run EAC. You will see a “configuration wizard” that will search your drives for the LAME.EXE file – or you can point it to the executable yourself. For instructions for using EAC with FLAC, click here. I recommend configuring EAC to use LAME, because you can easily have EAC rip audio files to WAV by right-clicking on the files and selecting “Copy > Uncompressed”. At that point, all you would need to do is drag the WAV files to the FLAC front-end and click “encode”.
But I LIKE MP3s!
So you’ve read this whole article and still aren’t convinced. MP3s sound “good enough” to you. That’s fine… in fact, the overwhelming majority of my digital audio files are still in MP3 format too! And MP3 still holds the advantage of having a HUGE support base. There are hundreds of MP3 software players out there, not to mention thousands of portable players, car decks and DVD players that will play the MP3 format. But even if you’re gonna stick with MP3, there are some things you can do to make your music sound better:
Ditch the “All-In-Ones” – Most of the “all-in-one” programs out there – programs like MusicMatch, RealPlayer and Windows Media Player – do a decent job of everything, but a bad job of one thing – encoding audio data. These apps are notorious for creating sub-par MP3s. Using a program tailor-made for such a task means that you’ll get the best available quality for your MP3s. For this, I recommend Exact Audio Copy (an audio ripper) along with the world-famous LAME MP3 encoder (both programs are free). While it’s true that these are two different programs, EAC comes with a simple configuration wizard that closely integrates the two during setup. Once the setup wizard has been run, using EAC is as easy as any of the “all-in-ones” yet produces much better results.
Jack up the bitrate – In the early days of MP3, 128kbps was the “gold standard” for audio files. 128kbps was figured to be the “sweet spot” between audio quality, file size and portability. These days, 128kbps is considered “old school”. You should encode your MP3s at least 192kbps if not 256kbps or higher. Those of you that listen to music with a great dynamic range – classical, jazz, ambient and new age – should seriously consider using 320kbps for your recordings, if not ditching MP3 altogether for FLAC.
Don’t fret – Seriously, if MP3s sound “good enough” to you, then don’t worry about it. Keep ripping and listening. Having said that, you will still have uses for things like FLAC. For example, if a friend brings some super-rare CD over to your house one day, consider ripping it with FLAC and not MP3. This way you can make as many perfect copies of the disc as you wish. You can even burn the FLAC files to a data CD-ROM or DVD for “permanent” storage after encoding them in MP3 format for “everyday” use.
Other Lossless Audio Formats
As promised, here are some links to other lossless audio formats. I have not tested ANY of them, so try them at your own risk. Most of them require WinAMP or Foobar for playback. Although FLAC should arguably be considered the “market leader”, Monkey’s Audio is also popular. Note that the formats are listed below in (roughly) the order of their popularity:
Monkey’s Audio
Lossless Audio
LPAC
OptimFROG
WavPACK
Also, note that both Apple and Microsoft offer “lossless” versions of their own AAC and WMA formats.