Wednesday, October 1, 2014

Mp3 or AAC - Which Is Best?

I'm surprised that so many music sites insist on using mp3 format for their files despite the generally poor quality - even Soundcloud uses only 128kbs mp3 when AAC is a patently superior file format at that same file size/bandwidth. Most noticeable is the retention of transients (eg drum beats) and much better top-end response. AAC sounds almost like CD-quality.

waveform2

Key points
  • mp3 is generally MPEG 1 (layer 3). They did extend it slightly in MPEG 2 to add some lower sample rates and some different channel formats.
  • AAC (Advanced Audio Coding) is MPEG 2 and also in MPEG 4.
  • mp3 only uses 576 blocks to encode audio.
  • AAC uses 960 or 1024 blocks to encode audio.
Note: MPEG stand for Motion Picture Experts Group - the standard-setting body for this sort of stuff.
There's a huge difference in quality between the formats - mainly because they tweaked the hell out of the encoding algorithm after the original mp3, so 128kbps AAC sounds almost like CD-quality, whereas mp3 128kbps sounds phasey and dull.
AAC is also what's generally used nowadays for radio broadcast - they call it MP2 though, to confuse the production people.

So what about MPEG 3 and 4 then?
MPEG 3 only had a couple of tweaks to add to the standard, so rather than having a completely new standard, they just updated good ol' MPEG 2 instead. They added things like more channels for surround-sound files and the like.
MPEG 4 was mainly about metadata (information embedded in the files) and even more channels - basically bundling a variety of video and audio formats (and other stuff like subtitles) into a handy container called "MP4".
If there's only audio inside it, it gets called M4A. Though some M4A files can also contain Apple Lossless format at master quality, but at a larger file size. Lossless formats are more like zip files - they're compressed down to smaller size (not as small as mp3 or AAC) but can be expanded again without any loss.

What's this bit-rate thing?
Since most of these lossy formats were designed to be streamed - either on the internet or off an optical disc, they're measured in how many bits per second they need.

Is that the same as sample-rate?
No - mp3s and AAC files still retain the original sample-rate and bit-depth as the original file. At lower bitrates, obviously more information needs to be trimmed out of the audio to make it smaller - hence the quality loss.

So why do we still use mp3 rather than AAC?
Well, if you're using Apple products, you're likely using AAC way more than you know. Otherwise, mp3 has just been around longer and more software can be guaranteed to play mp3s than AAC (although it's pretty rare that something can't play both now). It's like the old VHS vs Betamax video tape format thing all over again - the lowest common denominator (and cheapest format) usually wins against quality.

So what should you use?
If there's a choice - go with AAC. If not, go with mp3. Simple.
(Or a lossless format if possible!)

Friday, January 4, 2013

Over the Top and Back - Avoiding the Uncanny Valley in Music Production

uncanny reverb valley

One of the dangers of nibbling away at mixing songs - commonly with your mouse rather than a dedicated audio control surface or mixing desk, is that it's easy to be far too conservative when adding effects and the like.

What typically happens is you slowly push the level of an effect up until it starts to sound like it's too much - then you back it down slightly to get a nice balance of "wet" effect vs the "dry" sound source. Ahhhh. Nice.

This is fine - but there are often different contexts that effects work within when the balance of effect vs dry sound change radically - so by using this conservative method, you're always remaining inside the one safe context of the sound balance without ever realising any of the other creative possibilities available.

If you keep pushing the level of the effect past the point where it sounds bad or too much, then you can sometimes get beyond the audio version of the "Uncanny Valley" into a different range of possible sounds.

A simple example is reverb. The first "conservative" range remains within the context of adding a nice subtle tail to a sound to make it blend, or perhaps to give a subtle halo of space around the sound.

As you keep pushing the reverb level up - the sound becomes muddy and cluttered as the dry sound and the reverb fight each other. This is the audio version of the "Uncanny Valley".

If you keep pushing the reverb level even further, you will change the reverb's context completely. The room environment is now dominant, with your instrument or voice existing within it. Of course at this point it will probably also become apparent that there will need to be some tweaking of the reverb to clean it up a bit - adjusting predelay, reverb time and perhaps applying some low-cut EQ to take out some mush.

Reverb is not the only thing that this works with - try it with any applied effect like chorus, flange, distortion, echo etc. Or try going to the extreme and remove the dry sound completely. (Try pre-fade effects sends with the fader pulled right back).

It certainly opens up many more creative possibilities and can help you discover fresh sounds for your mix to make it a little more exciting. Plus it doesn't take much more effort or time to do this and it has zero risk! So make sure you go way over-the-top when applying effects, then just bring it back to where it works best for the song.

Wednesday, November 7, 2012

Moving blog to new site

Hi,

I've decided to set up my own web sites over at zedbrookes.co.nz and BrookesAudioDesign.co.nz and am currently in the process of transferring all the contents of this blog over to those. The Brookes Audio Design site is more audio tech-y, while the Zed Brookes one is more music creativity-based.

Feel free to pop over for a visit!

zed