Gearheads 101: Free Memory!

August 31, 1998

When I bought my first sampler, it came with 1 Meg of RAM. Later I felt like the gods had taken a liking to me when I got the upgrade, maxing it out at 2 Megs. Since that time, samplers have become my main instrument for composing as well as playing live. Although the price of both samplers and RAM has dropped considerably, let's face it -- we're still talking about a lot of cash. If you couldn't get that high priced sampler with the killer graphic display and the Gazillion Megs of RAM this time, here are some tips and tricks I use that will help you make the most of what you have now.

The first thing to do when trying to free up memory is listen closely to the sounds you're working with. If they are already used in a sequence, listen to them in context and by triggering them manually. Maybe there's a 4 bar loop that is retriggered every 2 bars. Get rid of the extra 2. If you have a multi-sampled bass sound that's only played in one octave, delete the unused zones. (Be sure to save this under a new name, so you don't overwrite the original multi-sample.)

If your sampler has a function for sample-rate conversion, take advantage of it. You'll hear little or no difference in quality when converting most sounds to a lower rate, especially sounds without a lot of high frequency content. Try all the samples one at a time. Start with the lowest rate possible and work your way up until you're satisfied with the results. Be sure not to overwrite the original sample until you're absolutely certain you want to stick with the new, converted sound! This process will take less time as you get more accustomed to the frequencies lost at different sample rates. If your sampler doesn't have a conversion function, it may still be equipped to sample at different rates. Try recording the sound onto DAT (preferred) or a high quality cassette tape, then sample it back into your machine at a lower rate. (Tip: For a lo-fi, grainy sound use a cheap, normal bias cassette. It's great for drum loops that are "too clean" and re-sampling them, using the lower sample rate will use less memory!)

Figure 1: A sample played at its original pitch is shown in (a). (b) shows the same sample played at a lower pitch, which stretches out the audio. This improves your ability to locate the start point of the sample precisely.

When trying to trim dead air from the ends of your samples, big graphic displays can be very helpful -- and very expensive! If you don't have this luxury, this next trick is for you. Let's start with a sample you recorded yourself. It can be any sample. Choose one and load it up. I'll wait ... ready? O.K. Write down the current pitch value of the sample. Next tune the pitch down as far as you can, while having it still be audible. This will not only stretch the length of the audio, but it will stretch the space in front of it too. This makes it a lot easier to get your start point right up against the actual attack of the sound itself.. While repeatedly triggering the sample at it's new low pitch, move the start point up gradually until you hear a click. This is the start of the sample. Now move the start point back slowly, in small increments until the click goes away completely. Congratulations, you've got a new start point. Write this new value down.

Next we'll tackle the end of the sample. This one's easy to overlook because you can't necessarily hear a difference when it's trimmed, but like an unwanted house guest, it's still lingering there and taking up space. In order for this to work, you'll first have to find the control that turns the sample backwards. When you locate it, reverse the sample. Some samplers give you the option of reversing the actual data or just listening to the sample from back to front. For our purposes, either one will work. Next trigger the sample to make sure it sounds backwards. (I've actually ended up using sounds backwards after discovering how cool they sounded while using this technique.) Now we're going to do primarily the same thing we did with the front of the sample. If you've been with me this far, the sample should already be tuned way down. Just move the end point (decrease the value) until you hear the click, then bring it up slowly until it disappears. Although the end point of the sound occurs before the start point, be certain it is the end point you're changing. In case of a screw up, just re-enter the value of the start point. You did write it down didn't you?

To finish this process, all that's left to do is bring the sample back up to it's original pitch and flip it so it plays forward again. I always listen to it at this point to be sure neither the attack nor the end of the sampled sound are cut off, and that there are no audible clicks on either end. This is your last chance to correct any problems. When you're sure everything sounds the way it should, remove your sample's excess baggage by using your sampler's trim or truncate command. This is important. Your sample will sound the same without this step, but the dead air will remain in memory, making all of this almost pointless.

So what are you going to do with the RAM you saved? Each sample by itself probably didn't free up a lot of memory, but by making this a part of your basic sample recording/editing process, it will add up. With many sounds, especially drums and basses, having the sound begin with no delay after you strike the key (pad, drum, string, whatever) will make a noticeable improvement in your grooves. Just wait until you see what you can fit inside that Gazillion Meg sampler when you do get it!

Bill Rettie is a Programmer/keyboardist/sound designer who splits his time between New York and L.A. these days. He's currently working with the band PRONG and has been recognized in music industry for his groundbreaking use of sound collages both on record and on tour. Bill also has credits including The Clay People (Mercury Records) and Shank-456 (Roadrunner Records).