Recording and mixing tutorial (sort of)

Making the recording

Audio?
Use a hard disk recorder that allows you to record each track individually before processing. For example, the Alesis HD24. (This has 24 channels on which to record so you have plenty of channels). This records on all channels at 24bit 48khz audio, which is higher than CD quality, which weighs in at 16bit 44.1khz. (After mixing you need to drop the quality to CD quality, haha).

Sensitive microphones will usually require phantom power for the circuitry in them to become active and to therefore pick up sound. With phantom power, 48V is sent down the cable from the preamp to the microphone. A mixer that supports phantom power is therefore needed. You only need the mixer for the instruments/singers that need microphones. The output from each individual channel on the mixer is sent to the HDD recorder.

To record the guitars and bass, take the line out of each amplifier to a direct inject box. If the amp is poorly equipped and does not have proper line out, a DI box sometimes accepts speaker input so you can plug the speaker output of an amplifier into a DI box. The audio from the DI box is sent to the HDD recorder too. With bass, try and get the DI before effects are added to them because bass sounds best without any effects. With guitar, try and get the sound of the amp but not the effects like reverb or delays.

For the live sound on the night, take the appropriate tracks that you want to output to the PA system and add effects to them. This will depend on the size of the venue. For the stuff we play, you normally cannot hear the bass drum or snare which you really need for people to dance and groove. For example:

Channel	Instrument	Method	To PA?
1	Bass drum	Microphone via mixer	Yes with compression and EQ
2	Snare	Microphone via mixer	Yes with compression and EQ
3	Left overhead	Microphone via mixer	No
4	Right overhead	Microphone via mixer	No
5	Smallest tom	Microphone via mixer	Perhaps, with compression and EQ
6	Middle tom	Microphone via mixer	Perhaps, with compression and EQ
7	Floor tom	Microphone via mixer	Perhaps, with compression and EQ
8	Hi hat	Microphone via mixer	No
9	Left vocals	Microphone via mixer	Yes with compression and EQ, perhaps reverb
10	Middle vocals/main singer	Microphone via mixer	Yes with compression and EQ, perhaps reverb
11	Right vocals	Microphone via mixer	Yes with compression and EQ, perhaps reverb
12	Bass before effects and EQ	Direct injection	No - it has it's own massive amp
13	Rhythm guitar 1	Direct injection	No, it has an amp already
14	Lead guitar	Direct injection	No, it has a screaming amp
15	Rhythm guitar 2	Direct injection	No, it has an amp thanks
16	Ambient noise (audience perhaps?)	Microphone via mixer	No way

You might also like to take the PA signal and send it to your foldback monitors so that you can hear what you're singing. Since you probably don't have a road crew, it is best to get all this set up before hand or perhaps during the soundcheck as nobody will have a clue how to adjust your mixer whilst you're playing.

Mixers?
A mixer is actually dead simple. It has three stages:

1. Input, where you set the gain of the input signals (so that they are not clipping and distorting)
2. Processing, where you add effects, equaliser and compression.
3. Output, where you send it to the right outputs on the mixer, and set the overall volume, perhaps adding an overall effect if the mixer supports it.

See below for short details of equalisers, compressors, gates etc.

Mixing the beast

1. Install PlanetCCRMA 4 DVD and get updates from the Internet (good luck getting your wireless network card working with Linux). The FC4 DVD image linked to above is very old and you'll need to download a few gigabytes of updates.
   Alternatively, install Fedora Core 6 (or later) from the net but this doesn't have half as much software with it (notably, lack of LADSPA audio plugins plus other bits and bobs). Having said that, the visual effects you can get with Beryl (and nVidia) on FC6 rival Vista's and Mac OS X's effects, and are better in my opinion because they didn't cost me anything. Follow the instructions for PlanetCCRMA 4 to give system processing priority to audio. ie. Who cares whether the screen is taking ages to refresh as long as the sound doesn't blip or drop out?!
2. Read up on JACK and get your soundcard working without audio dropouts and XRUNs.
3. Buy a decent set of nearfield studio monitors. These are speakers that can be put close to your monitor and have a relatively flat frequency response. Ordinary consumer grade speakers that come with a hifi or surround-sound system put out way too much bass and treble so you don't truly hear what is being sent to them. Monitors range in price from not-bad to mega expensive. You can get some with amplifiers built into them which is great. I use a set of Behringer Truth B2031As. They're massive and very loud, plus cheap! You'll need a good set of stands for them and some accoustic deadening equipment to stop the vibrations going to the floor and driving the neighbours potty.
4. Buy a decent set of studio monitoring headphones. I use Beyerdynamic DT 770s. Heck, they're expensive but just don't go out for a curry for a few nights and you'll have saved up enough cash. They're mega comfy.
5. Buy a decent sound card. Again, consumer grade ones do not have good Digital to Analogue converters (DA) so they are really noisey and if you listen to them with the machine idling, you can hear all sorts of strange noises (such as when you move a mouse, minimise a window etc.). I bought a cheapo Terratec Phase 22 because it has 1/4" jacks on it and RCA input/output plus MIDI and wasn't going to break the bank.
6. Copy all the files from your hard disk recorder to your computer, preferrably doing this overnight whilst you sleep because the 10Mb connection on the machine is so slow. Alternatively, buy a FirePort from Alesis. The Alesis HD24 works as an FTP server so you can connect to it via Windows Explorer if you want to use the super slow Ethernet RJ45 port on it.
7. Open your sound mixing software of choice. I use Muse (which is free) because I don't want to pay for software like ProTools or Cubase and don't want to harm my conscience by stealing software. Other freebies include Rosegarden4 and Ardour. There is a free one for Windows named Music Studio Producer. It isn't pretty but it works.
8. Add a wave track for each track that you have (so you'll need 16 for the above example) and set each track/channel to 0db
9. Import each appropriate wave file you copied across from your HDD to recorder to each track you just added.
10. Add start and stop end points for the song you're working on using the middle and right mouse buttons. You can then loop between these two points.
11. Add 5 groups: 1 for vocals, 1 for instruments, 1 for the bass drum, 1 for drums overall and 1 for the toms. Each of these groups should output to the main output.
12. Route the bass drum channel to the bass drum group.
13. Route the snare, hi hats and two overheads to the 'drums overall' group.
14. Route the three toms to the tom group.
15. Route the main vocals and supporting vocals to the vocal group.
16. Route the stringed instruments to the instruments group.

Now that you have all of these set up, you'll need to work on each channel individually so mute every other channel (or group) to listen to one at a time. Once you're happy with each channel respectively, turn on the others and fine tune the whole thing to make it sound good overall. It really is just a case of adding relevant effects/processors for each channel and group and getting the volumes right.

When adding effects and processors, the order that the processors run in is important. They are ordered top to bottom. For each channel, the overall order of doing things is:

1. Get rid of unwanted sounds either using a gate, or frequency filters (see below for details of lowpass, highpass etc.). In the case of a gate (see below), it'll open when the sound reaches a certain volume and/or includes the frequencies we're interested in. (Great for the bass drum - just listen to hip hop).
2. Add another processor to trim it futher if necessary.
3. Add an equaliser to shape the sound (boost the bass, get rid of the nasally sounding middle, attenuate the ear-splitting treble)
4. In the case of a snare, add a reverb here too because a snare sounds much better with a massive reverb (especially if you love the 1980s)
5. Add other effects if necessary, such as a tube amp emulator, a flanger, chorus (to thicken stuff up) or really dodgy-sounding effects like harmonic generators or spectral accumulators.
6. Add a compressor to boost the now beautiful sound we have to boost the volume and fill out the audio spectrum yet limit how loud the sound is.

On the group we've sent the sound to, further processing can be done:

1. Add a tube preamp emulation to warm it up and a touch of distortion to make it sound like the music is too loud and is leaping out of the speakers.
2. Add a reverb to make it sound realistic. You'll not want to use too many reverbs because these are CPU intensive and you'll run the risk of your computer melting.
3. Add a limiter/compressor to boost the sound/limit how loud it is. Never ever ever go into the 'red' on a VU meter because it'll sound distorted and awful.

Warmth?
When we talk about warmth in songs, this is the main argument for using analogue equipment over digital equipment. Because digital equipment is sampling a sound 48000 times a second (in the case of 48khz) and recording that bit of data using 16 bits, digital sounds are effectively just a load of samples strung together and piped to your speakers. Audiophiles claim that digital sounds shiny and lifeless, which is a bit true - digital recordings do sound really 'clean' and 'shiny'.
Analogue equipment does not have a concept of sampling - sound is sound. Older recordings (such as those on vinyl) sound warmer, software and mellower. Analogue sounds muddier to my ears but then again, all the old recordings originally made on analogue tape (Led Zep, Cream, Deep Purple (Machine Head sounds like the tape is about to break), Rainbow, Bad Company, Free, The Who, The Beatles, early Van Halen, Yes etc.) are now on CD so all I have to listen to is digital versions of analogue stuff so the argument is therefore broken...

Can I see?
To see graphically what is going on, use japa or jaaa and connect it (using qjackctl) to the outputs of your mixing software, turning off irrelevant channels you don't want to hear. You can see which frequency is loudest (due to the graphical waveform) and hence get a better understanding of what to do with your EQ. You might like to listen to a few commercial CDs that you love the drum sounds of and put a drum section through these to see what is going on. Or perhaps sample the bass drum sound of a song you like and put this through it. Remember that they'll have equaliser and compression plus reverb on their sound already but it might be useful anyway.

Processors and Effects

Compression?
Yes, we boost the signal if it quiet, and make it quieter if it is too loud. The threshold is the volume the signal has to be before we work on it. The compression ratio is how strongly we apply the compression. A ratio of 10 means the compressor behaves as a limiter, thereby chopping off the tops of loud sounds that come through it. The decay is how long the compressor takes before it gives up and goes back to idling.

Gate?
Using a gate, we let the signal pass through when we reach a certain volume. Anything below that is not let through. They have attack (how long it takes for the gate to open), hold (how long we keep the gate open), and decay (how long we take to close the gate). Threshold is the volume the signal needs to be to be let through. Range is how much of the signal coming in we doctor.

Reverb?
Reverb is the simulated sound of space. When you drop a box in a hall, the sound of dropping the box bounces off all the walls, hence with your eyes closed you'd know that you're in a massive room. Dropping a box in your tiny bedroom would sound different. To make sounds (vocals, drums, whatever) sound more realistic, we add reverb to them. Effects like this are mixed between wet (fully applied to an incoming signal) and dry (no effect applied). Or you might have to set the wet level and dry level. Fully wet means you'd only hear the sound of the reverb, which isn't that realistic because when you drop a box, you hear the sound of the box dropping, not just the sound bouncing off the walls.

Highpass? Lowpass? Bandpass?
These are filters that either let the high frequencies past (highpass), low frequencies past (lowpass) or a band of frequencies past (bandpass). Each of these has a frequency cutoff . A notch pass is one that cuts out a slither of frequencies. For the bandpass and notch filters, you specify the frequency centre the filter will work at and then the Q value (or range) either side of this frequency.
Interestingly, filters are used to an excessive degree in dance and trance music. When a synth goes from sounding really muddy to really bright over a course of time, all they are doing is increasing a lowpass cut-off frequency gradually. This is used to death but still appears to survive.

For further information on audio effects, buy a book or read wikipedia. Look at the bottom of that page under 'Techniques'.

The End

Everything on and available from this page Copyright © Richard Gibson 2007.