Once upon a time, music was made in real time by real people who made real mistakes. No two performances, or takes, were ever alike. In the last two decades a new type of musician has emerged. Barring power cuts and software gremlins, the computer can faithfully reproduce a composition with consummate accuracy, and in the late '70s and early '80s the perfect beat of a sequenced dance track sounded fresh. This metronomic precision was partly intentional, and partly a result of limitations inherent in the equipment available at the time.

Quantising has been with us since sequencing began. In fact the first sequencers were step-time recorders offering no choice but strict tempo playback. Before MIDI staged its successful coup of the electro-musical stage, DIN sync kept your 808 in time with your 303, and a knowledge of Gates, Control Voltages and Triggers was de rigeur. (Incidentally, DIN-syncing is currently enjoying a revival in some quarters because an FSK-locked system can sound even more mechanical than an over-quantised MIDI setup. MIDI, it is said, just isn't quite on time!)

Even the first MIDI sequencers offered little real hope of a usable result when attempting a recording in real time. I can remember struggling to program triplets in step time on a reluctant MSQ700 during the recording of 'Wishing I Was Lucky' (Wet Drop Drip). There was a real-time facility on that machine, but it yielded unpredictable results. Some notes turned up just where you didn't want them, while others would appear to be on vacation. As is almost always the way, however, the quirks of the machine could be put to damn good use. For instance, recording in 4/4 and re-quantising to 3/4 left some notes with a their length reduced to a minimum of '1'. Great for producing bass lines with choked notes, and particularly effective with a meaty DX7 patch!

INTO THE GROOVE

The long-anticipated arrival of the MC500 with genuine real time programming and (for then) revolutionary ergonomics finally enabled the sequencer to be treated like a tape recorder. Subject to the (actually quite good) resolution of the machine, any out-of-time playing remained just that, unless you decided to apply quantisation. Akai's APC60 further promoted the reform process; instead of having to quantise to a strict time signature (16 triplets or whatever) it was possible to swing the tempo of a track; the 'Feel' could be defined as x% of 4/4 plus 100-x% of 3/4. Things were beginning to get groovy.

Current music sequencing software offers a host of complex quantising options, such as 'iterative quantise', which can progressively shift notes away from their original positions towards some user-defined quantise value, or 'groove template' where the user can specify where any or all beats should fall. In Cubase it is even possible to record a piece of music in real time and, by digging out the 'Q' tool on the main page, click and drag the 'feel' of that music onto any other track. Groove templates are fascinating.

Cubase, Notator and probably every other half decent sequencing package offers such a facility, and many programmers are drawn to what sounds like a very useful tool. The trouble is, using a groove template does not guarantee groovy results — James Brown's drummer has a pretty cool idea of what a 'feel' is, but do you? For every groovy bit of programming I've heard, there have been a hundred botched attempts that at best made the music merely sound a bit odd. Groove programming is extremely difficult, since not only timing shifts but also dynamics are involved. Both of these parameters are interactive: once a note has been shifted in time its velocity may require adjustment and vice-versa. Furthermore, altering one note affects the way we hear all the others. There are no quick and easy answers. (Fractal Quantise anyone?)

BACK TO LIFE

Despite the programming power of the modern sequencer, dance music in particular over the last few years has relied heavily on sampled loops of real drummers (909s notwithstanding). I have seen programmers painstakingly adding drums on top of a loop, nudging that snare or this bass drum back and forth until the sequenced drums sound 'about right'. Now, not a lot of people know this, but there is a way to capture both the timing and dynamic feel of any piece of music. You will need the following:

• 1. Any old sampler.
• 2. A modern sequencer with groove mapping.
• 3. A Simmons brain or similar velocity sensitive trigger-to-MIDI device.

Item 3 used to be available S/H for a song (sic), but that was prior to this article being published. I see the quicker-witted amongst you have already figured this trick out, and are currently scouring the small ads. For the rest of you, here's how it's done.

Grab a sample off one of your favourite platters and get it rolling along in time with your sequencer. (If you need to be told how to do this you're in the wrong line of work). OK. Nowhere comes the clever bit. Feed the audio output of the sampler into the trigger input of the brain. Adjust the threshold until you get a sensible number of peaks firing (16 in a bar will do nicely). Connect the brain's MIDI out to the input of the sequencer, press record and voila! Instant groove template, dynamics and all.

If you throw a serious equaliser into the mix (TC Electronics do a screamer) it may be possible to isolate individual voices from a loop, and subsequently record them onto separate tracks. In practice I have found it possible using even the basic EQ on my Tascam MM1 (with two shelving frequencies) to isolate kicks from the rest of the drums in a loop. Note that the Simmons brain will react more slowly to low frequencies than to high frequency peaks, and consequently you may have to advance, say, the bass drum relative to the hi-hat.

Figure 1 shows how to convert a mono loop into a multi-channel groove sequence. It is good idea to disable MIDI Thru whilst groove recording, in order to prevent the incoming notes from the 'brain' feeding back into its audio input, causing multiple triggering.

After a bit of fiddling with the input gain of your TMI, and with treble emphasised and bass cut on the loop, you should be able to produce very usable grooves. I've found that due to the complex nature (and in some cases appalling signal-to-noise ratio) of loops, things don't always run smoothly, and some 'cleaning up' of the grooves may be necessary. Take a close look at your groove template in Grid Edit (or whatever); you may have to delete unwanted 'shadow' peaks, or perhaps add a beat that was somehow missed.

I've often found it best to manually program each drum on to a different track of my sequencer, then apply the quantise map, and finally tweak the odd rogue beat. As each voice is successively quantised it appears to merge into the original loop. There seems no good reason not to mix the loop back in with drums quantised in this way, since everything is (or should be) totally synchronous. Applying the groove template needn't be restricted to drum voices: try it out on bass lines, piano vamps etc. You may sometimes find that it's best to map only the timing data, not the dynamics, or perhaps apply only a percentage of the dynamic template. Suck it and see.

You can monitor, in real time, the results of adjustments to the equaliser and to the TMI input sensitivity if the brain's MIDI out is connected to the MIDI input of a drum machine, provided that the brain has several inputs with assignable notes. Most units will have this feature, as does the TMI.

Now, it just so happens that prior to writing this article I bought up the entire world stock of trigger-to-MIDI interfaces, and for a small consideration...