Welcome to the nerdy side of music!

Let there be an intro

This post will be part of a series of articles. Stay tuned for updates.
PART 2 IS NOW HERE

When I was younger and more inexperienced with the Internet and music as a whole, I’d often end up on many p2p clients to find new content to explore. I’ve seen Napster, WinMx, Audiogalaxy, gNutella, eMule, Kazaa (lite, lite++, etc), iMesh, Morpheus, etc. I tried them all through the years, with connections ranging from 56k to DSL, ADSL and cable.

One thing that I did a lot was trying to find the smallest files possible, because it meant faster download, and I could keep more inside my computer. It took me a few years and speakers worth more than $15 to realize that music at 96kbps or even 128kbps was not that great, after all.

First of all, for those who aren’t too tech savvy, let’s define what the bit rate is. To make things simple, a bit rate can tell how much information is read by the computer every second. The higher it is, the best the quality of the song will be. Also, the higher the bit rate, the bigger the file’s size. With this in mind, you know the basics to understand why choosing VBR (Variable bit rate) may be better than CBR (Constant bit rate).

Choosing a bit rate

When downloading music (may it be legal or not), you’re likely to encounter various formats. The most common formats I can get from the top of my head are .mp3, .flac, .aac, .wma, .m4a/.m4p (you crazy iTunes users), .aiff, .wav, .ogg, and sometimes .ape. While they can guide you on the quality of the songs, it’s not everything: mp3 files can go from extremely shitty to high quality, both in CBR and VBR, while formats like .aiff, .flac and .wav are considered lossless and usually present the best quality of all. However, they’re quite heavy files, and a 40 minutes album can easily weight over 250mb.

So what bit rates should be deemed acceptable, or a good compromise between quality and file size? It’s quite hard to say, because it often depends on people’s ears and their audio equipment. It’s evident that low-end speakers and/or a cheap amplifier won’t make high bit rates sound better than others. So just to compare, here are a few real-life references that could help:

• Human voices can usually be understood at under 1kbps, although it can be hard;
• Telephones have a bit rate from 8kbps to 64kbps depending on the technology and model;
• AM radio has an overall bit rate of 32kbps;
• FM radio stands at 96kbps;
• Compact Discs use bit rates ranging up to 1411kbps (1.4mbps).

Voice is usually easier to understand as it is an evolutive trait: it was a good tool for survival to be able to recognize one’s peers, so this can explain why media centered around the voice usually use lower bit rates (ie.: phones and AM radio). When it comes to music, though, sound quality becomes a lot more important due to our training and how music often acts like candy for the brain. The easier, the better.

So to decide what you likely need in music, a few questions and facts can be useful. As an example, if listening to the radio tires your ears after a while, it’s likely too low quality for you. Otherwise, if you can’t find anything to say against it, then you lose nothing by going in lower qualities.

Usually, people will not be too picky about music at 128kbps, although heavier music like Metal will sound more distorted than it was before. 160 kbps is usually enough for most users, but with the advances in mass storage and file transfers, 192kbps has now become the standard for .mp3 files. If 192kbps is not enough, usually, 256kbps or 320kbps will please you, but at the cost of place. There’s even music compressed at 500kbps-1mbps, with the lossless formats mentioned earlier.

So why would I go for VBR if these are fine?

VBR is like eating all the cake and leaving the pork liver on the side of the dish: VBR will usually encompass rates set by the user over a given range. As an example, a VBR file can have bit rates varying through a single song from 96kbps to 320kbps, depending on how much information is going through.

So say I’m listening to The Wall by Pink Floyd. The Happiest Days of our lives (the song with the helicopter coming close) is the intro. When the song starts, the volume is low, there’s not too much instrumentation. No need to have much bits coming through, right? Well indeed, no need. VBR could then reduce the bit rate there to 128kbps (random number).

Then, when The Wall begins, it starts with someone yelling at the top of their lungs, with heavy percussions, bass kicks in. There, a VBR file might have its quality raised up to 256 kbps, so there’s no loss of quality when more is needed. The rest of the song is more or less guitar solos, singing, so you can assume the quality would hover between 168kbps and 320kbps. The last 30 seconds of the song are nothing but the school principal and kids yelling outside. At this part, the quality could drop to the minimum needed, around 96kbps or even 32kbps if you let the file go this low.

That’s not really interesting, until you compare with CBR music: if I want the 256kbps bit rate I need to have no low quality parts during the song, I’d have to have every single part at 256kbps, taking more place than ever needed. Apply this to any song, really.

This is why VBR is pretty much the best format for music around: it’s the best compromise between quality and size.

What are the downsides?

There’s not a lot, really. The most common cases would be compatibility, but mp3 players that can not support VBR music are getting rarer and rarer (unless you’re a fan of .ogg files). It’s also harder to have a live stream of VBR audio over the Internet, due to the nature of changing bit rates. It also takes longer to compress files to a variable bit rate than a constant one. That’s pretty much it.

Conclusion

There’s more to come. I’m going to write more about other aspects of VBR music in following articles (possibly next week). I’m planning on explaining more detailed stuff like VBR v0, VBR v1, VBR v2, etc., how to compress your own VBR files, etc. You can already read more on these sites, if you’re interested:

• http://www.mp3developments.com/article6.php
• http://www.codinghorror.com/blog/archives/000470.html
• http://en.wikipedia.org/wiki/Audio_data_compression

Part 2.

Category: Theory