[Ramshackles]

So on this forum I've often heard quite beautiful songs which are held back by poor mixing or a lack of mixing. Of course, there are some wonderfully mixed songs aswell, so this thread is aimed at helping those who might just record and submit, without giving much thought to mixing or without an idea of how to approach it.
Now, I'm by no means a 'professional' and I certainly have much to learn and improve. But thats partly the point - I want to show that its possible to make great sounds in your own home with simple gear.
I also want to shout out that I am always ready to mix songs I like (for free) - just get in touch and have a chat!
So I'm going to start with basics and work up. For many of you this will probably be obvious but I think there are a few that can benefit. If you have tips of your own, please contribute.

I'm working on a song at the moment for my band, Ramshackles. It grows to be a fairly dense song so it's a good example.
Here is the raw, unmixed song. All levels are set to zero, panning is centre and no effects or processing have been applied:
http://snd.sc/n1rvBB

The song has 28 tracks and you can see the DAW setup here:


Assess
I think an important thing to do before anything else is to assess the song - listen through the individual tracks for any problems (pops, cracks, hum etc) and sort them out.
We can already hear a lot of problems that need addressing:
1. The relative levels of all the instruments are bad, especially the drums
2. The lead vocals are often drowned out, either by other instruments or the backing
3. The bass is much too loud
4. 1 drum (a tom) seems to be much much louder than the rest of the kit
5. General levels of single tracks across the entire song are not consistent - sometimes dropping in and out.
6. There is a lot of clipping

And this is just for the levels. You can also pick out tonal clashes, low frequency rumble, some timing errors and a couple of duff notes in the vocal. Each track is also very dry, which is something that needs to be addressed later.

Level Setting
The single most obvious and important part of mixing. It sounds like a trivial thing to do, but I'd spend up to half an hour setting the levels of each individual track. It's important to make sure that they sound good to you on many different systems - listen on your monitors, headphones, earbuds, stereo speakers, car speakers, whatever speakers you can listen on.
At this point I'm also gonna play with some panning. This is basically how much volume of a certain track goes to each speaker. If you pan something hard left, then the track is only played by the left speaker. To have better control over the panning, I generally record in mono - recording in stereo is useful for putting 2 mics on a guitar etc. Even then though, I've split the stereo acoustic guitar (blue tracks) into 2 mono tracks and the same with the bass (purple tracks).

I panned each acoustic track hard left and right, and spread out the drums a little (grey tracks) - I dont want them too wide or they could start to dominate too much (although I'm very amateurish at mixing drums). I kept the vocals, the bass and kick drum panned centre. The piano's and organs (pink-y tracks) I've spread a little left and right (between 30 and 60%)

Result
Ok, lets listen to the result:
http://snd.sc/qU8FwQ

Ok, so we are getting somewhere, although theres still a lot of simple things to address. Point 5 (levels not consistent) is still a problem which I'll address later with automation and compression.  This initial level setting also lets us hear more clearly which frequency imbalances to address, where things already sound good and where we can improve.
I'll add to this thread as I mix this song.

I should also point out that for the basic techniques I'll cover here, I'm not going to use anything more fancy than what you would find bundled with a regular DAW such as Logic, Cubase, Protools or Reaper

[Ramshackles]

The Low End
So, part 1 was a little intro and a reminder that the bare minimum of taking the time to set the levels and panning can do wonders for the sound of your track.
Now, everyone has a different approach but high on your list of things to do should be to address the low end. By that I mean anything around 100hz or below. This would be Bass guitar, Kick drum or anything playing around that range (piano an octave or so below middle C and down, tuba, Upright Bass, etc etc).
This area is difficult to deal with for a number of reasons - the ears are less sensitive to it, small speakers cannot accurately reproduce it, it is heavily affected by room acoustics etc etc.
For these reasons, it's important to have everything going on down there as clear as possible and also to limit the amount of tracks contributing to the low end.
To do this, we make EQ cuts.

EQ'ing
'Equalizing' is the process of boosting or cutting specific frequencies in a track. Every single DAW/Recording software I can think of has, at the bare minimum, a very simple EQ plugin. For the next task I'm just using Cubase's standard parametric EQ.
I've decided I want to leave the low end all for the kick drum and bass guitar. So starting with the drums (leaving out the kick), I make cuts or apply 'high pass filters' on everything, having the cutoff point around 100hz.

Note: for each instrument, the point at which their isn't any interesting/musical sound coming from it varies. E.G on the toms the resonant frequencies were around 130Hz, so I applied my cutoff points here. Everything below was just rumble-y sounds.

You should experiment to find the best places to cut on each thing. Applying these filters can also have the effect that the high end is now more clear - your ear no longer has to listen past all this low frequency noise and rumble to make out other details. You may even find that you want to bring down the volume of other things such as cymbals after making all these EQ cuts.

Here is a clip of the drums before any EQ'ing:
http://snd.sc/rs5vVj
Here is a clip after EQ'ing and applying any new volume changes:
http://snd.sc/mQTCHC

I've also applied a fairly broad cut around 2.5Khz in the whole drum track. I want the drums to sound in the background compared to the rest of the song and the ears are more sensitive around this frequency. So it, stands to reason that anything with a lot going on around there will sound more 'in your face'.

Pretty subtle changes, right? But then apply this across the whole mix - I'm gonna put cuts on every instrument except the bass.

Automation
Ok, I'm going to leave the low end there for now. There are plenty more things you can do to improve it, but I wanna keep these tips simple and quick to implement.
Ok, so next, I'm gonna apply a bit of automation. Automation refers to the process of changing the volume of a track throughout the song. A fade out is a form of automation. Automation can be applied to many more things (EQ, the level of different effects, etc etc). I'm not good enough to really feel confident doing much more than volume automation.
Again, all recording software I know of has pretty advanced systems for implementing automation. At it's 'lowest' form, it is simply moving the volume fader up and down as the track progresses.

You may have noticed that in the chorus section of the song (that begins 'In the darkest of forests' around 2:15) that the vocal is fairly hard to make out - it comes and goes. We need to automate the volume! So, I've gone into to cubase's automation panel and brought up the volume on words that are too quiet, or put it back down on louder parts. Here's a screenshot of the result:


The upper orange track is the waveform and the lower track shows the volume changes throughout. They are of the order of 1-3db. In the context of the rest of the music, these kind of changes are very hard to actually pick out individually, but the overall vocal is much clearer for it.

Now, what I'm going to do is go away and apply automation to all the vocal tracks I think is necessary. For this amount of vocals, it will probably take around an hour, but is quite worth it.
Automation is also useful for places where there are no vocals or not much is going on - for example at the end of a phrase the vocals might drop out, but the drums might do a cool little fill which is buried in the rest of the music. I'd bring up the volume of that little fill to add a little more excitement to the track.

In the next part, I'll see what difference all this automation has made, then go over more EQ'ing and compression - 2 standard but often confusing techniques for new recordists

[Ramshackles]

EQ
Ever listened to commercial songs and wonder how the vocals are so clear, without necessarily being that much louder than everything else going on? Many things come into play to achieve that, but more often than not an engineer would call on EQ to tackle the majority of it.

Think about the human voice and an acoustic guitar - they have very similar frequency spectrums. If you play them both at the same volume, there will be plenty of times where one is masked by the other somewhat. Now start to add all the instruments in a typical band. You can see that pretty soon, our frequency spectrum starts to get overloaded with information. A big part of solving this can be achieved by careful arrangements and recording. Any remaining problems can be solved by EQ. By careful cutting of selected frequencies in selected instruments, we can carve out space for other instruments while not changing the nice sound of the original instrument too much.

EQ isn't just for problem solving. Even the best recorded tracks might use EQ to add some nice 'colour' to a sound or bring out some nice details.
Before splashing EQ around the place, always decide whether it's totally necessary first. Sometimes things just sound great how they are. I like to follow the rule 'less is more'. If I start applying massive cuts or boosts, it's more an indication of doing something wrong in the actual recording. So in order to effectively use EQ, it is (like when I eq'd the low end) often a collection of very subtle adjustments to individual tracks.
Here is an example, part of the vocals before EQ'ing:
http://snd.sc/qMIhF7
and after I've applied my desired EQ:
http://snd.sc/pkFZJ3

In this song, I've applied some EQ to the vocals, acoustic guitar, Drums and bass. Again, I've just used a standard eq supplied with cubase for these purposes.

Compression
This is a tricky one and a technique I'm sure I still don't understand fully. The basics of compression is that it reduces the volume of the signal when it passes a certain threshold. The amount by which it reduces the volume is often changeable and called the ratio. E.G a ratio of 4:1 means that if the input to the compressor is 4 db over the threshold, the output will be 1db over.
You might want to introduce compression in situations where a vocal is fluctuating in volume such that in some phrases it is too loud, whereas in others it is masked by the music - for example when a singer raises their volume to hit a higher note, or going from singing to screaming.  Compression can be used for other things, such as reducing the 'transients' in signals or even making them stand out more (the transient is the attack or inital few ms of a signal - in a strum of a guitar, the transient would be the sound of the pick across the strings).

Different compressors often work in slightly different ways, each giving a different 'colour' to the sound. Most DAW's have at least a couple of different compressors, so you can try using them and listening to how they affect the signal.
In this clip, I've applied compression to the vocal 'Ahh's', the vocals in the break, and the bass all the way through. I've experimented with parallel compression, which is where, instead of putting a compressor as an insert effect on the signal, you setup a seperate FX channel/send channel/group channel which has the effect and send part of the original signal to that channel.
For the vocals and bass I've actually experimented with sending different levels of the signal to 3 different compressors to allow each compressor to add their own subtle colourations to the track.
 
Stereo Effects
One final quick thing which is cool. I've made the guitars sound slightly wider by making use of something called the haas effect. To do it, do this:
1. Add a delay to 1 channel of a stereo track (if your track is mono, double it and pan 1 left and 1 right, then add a delay to one of these)
2. Setup the delay so that it is between 10-30ms. E.G on the acoustic track, I put a delay on the right channel. My delay plugin has other controls such as feedback and filters, which I turned off.
The result should be that you perceive the source as being much wider. You can control this width by adjusting the delay, or the amount that it is mixed in with the original signal. If you put the delay above around 30ms, you ear will start to perceive the delayed signal as a seperate signal in it's own right, and the effect will be lost.

Ok, below is the result of the automation, EQ, compression and delay effects that I've discussed:
http://snd.sc/q2jDqh
Hopefully by going back to the original mix of just the level setting, you can hear the changes this has made. 1 thing should be clear: The changes can be subtle and thats the point. How you record the tracks is the main thing that will shape and guide your mix. The rest of mixing is about finding the right place for those tracks and bringing out their best qualities.

Now, what I might do to finish this mix is add any compression to other parts that need it (the drums probably) and then look into giving everything it's own space and 'air' with reverb.