Flex Your Loops!

Rolling your own loops can breath life into your compositions — especially if they can stretch and shrink on demand.

There can be very few people now who don’t make some use of audio loops in their music. I mean, it’s just so easy isn’t it? Loop libraries abound; you can buy them in various formats for any genre, and most DAWs ship with several gigabytes of such ‘content’. Your DAW also offers various means of editing, quantising and otherwise mangling your loops. We’ve really never had it so good.

Despite the quality of many commercial libraries, though, there’s a real satisfaction to be had from making your own loops. Not only do you get something unique, but you get to determine exactly what material you use in your projects, rather than let the loops dictate what you write. And although there’s a need to invest time in loop creation, you’ll find that you waste far less time auditioning a gazillion unsuitable loops you’ll never use — a process that can sap your creative juices all too quickly.

However, there’s much more to loop creation than tracking a part and snipping out a two- or four-bar measure. You need your loops to transition seamlessly when laid end-to-end, so your edits must be precise. You must work to avoid glitches, pops, crackles and other ugly artifacts. And you’re going to need your loop to be able to stretch or shrink to fit the tempo of any project you wish to use it on. In this article, then, I’ll do a number of things: I’ll explain how to make seamless loops; I’ll discuss the pros and cons of the different approaches to time-stretching; and I’ll then give some pointers to help you use looping tools effectively.

Loop Creation: Prep Work

Whatever tool you’ll eventually use for stretching, the first step is always to make a good loop, so let’s kick off with some tips for preparing ‘loop-friendly’ audio...

Almost all DAWs now have an offline time-stretch facility which allows you to drag a loop to the desired size and usually gives better results than real-time stretching. In Reaper (pictured), for example, you hit Alt and drag — the cursor turns into a pair of hands to denote that you’re in stretching mode. Create a basic loop: your loop could be an audio clip you recorded, a virtual instrument output you’ve bounced, or whatever. One option is to create loops in your DAW. Set up a track or two to which you can play along, then start recording. When you’re done, trim out a section you want to use as a loop. You’ll probably want to turn off snapping, zoom in to see the actual transients, and trim to the start of the first note and the beginning of the first note after the clip (assuming these fall on a beat). I also create loops by dragging in one-shots, like drum hits, to the timeline and then bouncing the results to an audio file.

Beware tempo mismatches: note that where you’re capturing loops from drum machines and keyboards, even if the bpm displays for hardware and software are the same, the actual tempos may differ very slightly. For instance, the theoretical duration of a four-measure 90bpm loop is 10666ms, but a four-measure loop generated by the internal sequencer in one of my keyboards lasted 10693ms. That may not sound like a huge difference, but when working with loops, small things really do add up! The problem’s easy enough to deal with, though:

• With short loops, trim the end to the correct length judged by ear if the difference is only a few milliseconds — don’t worry that the loop is a few milliseconds short or long for the desired tempo.
• Use a time-stretch algorithm and snap-to-grid facility to force the loop to the correct length. Generally, the difference will be small enough that time-stretching won’t alter the sound in any noticeable way. (Be careful, though — it’s rare, but some algorithms create a file that’s actually slightly shorter or longer than you specified in order to improve the fidelity. After processing, check that the loop is the right length.)

Perfect pitch: if you need to play in a specific key, don’t worry, but if you plan to use the loop in other keys, remember that transposing much more than three semitones often leads to a very unnatural sound. A solution is to record the loop in two or more keys: when recording my AdrenaLinn Guitars sample CD, I played each loop in the key of E (to cover the range D to G) and Bb (G# to C#), making it possible to reach any key without having to transpose by more than three semitones. It may not be possible to obtain the same chord voicing in two keys, but you can cheat — use DSP-based transposition and create a loop from the transposed version.

Check your tuning often: with melodic instruments, there’s nothing more frustrating than an otherwise perfect loop that’s a little off-pitch. So check tuning when you start, but also check tuning often. Melodyne Editor can fix tuning problems, but the results aren’t always perfect. It’s better to get the recording right in the first place. Similarly, when recording a drum kit, check that pedals don’t squeak, and so on.

Context is critical: your loops will sound much better if you play or sequence them to a backing track. This helps you get in a playing mood and provides a rhythmic reference. For loops with ‘swing’, it’s pretty much essential to have a rhythm to play against. Make sure you can change the backing track’s tempo!

Another such stretching tool, again using a key modifier and dragging a clip’s beginning or end. The left and middle clips are identical, but the one on the left has been stretched to 111 percent — it’s exactly two measures long and accommodates a slower tempo.In Reason, the Dr OctoRex file player assigns each slice to a MIDI note. Although the sequencer plays these back in order, you can also edit the order to create loop variations.Back to front: Apple Loops Utility, Acid Loops Properties page, and Sonar Loop Construction Window. They all work by placing markers at transients, but Apple Loops include more detailed metadata tags, Acid gives more looping techniques (as well as beat-mapping and mapping transients to a pool), while Sonar folds in loop processing and auditioning options.Slow things down: stretched files sound better when sped up than slowed down, so set the tempo for the right feel and practice until you nail the part. Before hitting record, though, reduce the tempo to, say, 100bpm (if the tempo was below 100bpm, don’t speed it up). You may also find that you can play the part more precisely at the slower tempo. Typically, a properly sliced/edited Acidised, Apple Loop, or REX file (more on these later) can stretch over a range of -15 to +60 percent. That means a 100bpm loop will be viable from about 85-160 bpm. For really down-tempo material, consider recording at 80bpm.

Loop length: I suggest using loops of two or four measures, because single-measure loops don’t ‘breathe’ with tension and release. For example, with two-measure rhythm-guitar loops, make one measure a steady rhythm and the second some kind of variation such as a short run that leads back into the loop’s beginning. That way you can repeat the first part of the loop when you want a steady background, and extend it to the full two measures before going into a change in the song (chorus, solo, next verse, and so on). With four-measure loops, I follow the same strategy — three measures of something steady, perhaps a slight variation in the second measure, and a major one in the fourth. Four-measure loops work well when the sound needs to evolve. Loops with eight or 16 measures are like ‘construction kits’ — you just extract the sections you want.

It’s easy to extend a loop by chopping it up and rearranging the elements. For example, if you create a four-measure loop that builds over four measures but you want it to build over eight measures, you can split the loop, repeat the first two measures twice, then play the full four-measure loop to complete the eight-measure figure.

Swing: swing adds a compound-time quality to a rhythm, by altering the duration of note pairs. (For example, with 50 percent swing, the first 16th note of a pair and the second 16th note have the same duration; with 57 percent swing, the first 16th note’s duration will equal 57 percent of an eighth note, while the second 16th note will equal 43 percent of an eighth note.) Loops recorded without swing may well work with a wider range of material, but they won’t have that cool swing feel! So if a loop wants to swing, let it. Unless it has a significant swing percentage, it will usually play OK against something recorded without swing, unless both loops have percussive attacks on the ‘swung’ beats. Note that many DAWs can quantise audio, and can therefore be used to introduce swing to a part recorded without it.

Keep things simple: the simpler and more elegant the loop, the more easily it will fit into a variety of musical contexts. Complicated runs can come in handy for ‘spotlight’ sections, but they’re not as universally applicable. You can always make a loop more complex by cutting it up and rearranging the various bits, or layering one-shots along with it.

Avoid effects: in the days of slower microprocessors, plug-ins used precious resources and had to be used sparingly. Many loop libraries added processing to loops not just so that they’d sound good ‘out of the box’ but because it meant you wouldn’t need to use plug-ins. Today, this is just not an issue, and an unprocessed loop will work in more contexts. The one major exception is effects that are integral to a loop, such as synchronised delay, vocoder effects or wah-wah.

It’s not always necessary to record long loops. To lengthen a shorter loop, copy and paste pieces of the loop. In this example, the original four-measure loop goes from measures five to nine. But its first two measures have been copied and pasted in measures one and two, as well as measures three and four.The process of quantising a bass note manually, from the beginning (top) to the end (bottom). In the top-most track, the cursor is snapped to the grid at the exact beginning of the kick waveform; the bass note in the next track down lags behind. Cut the space between the cursor and the beginning of the note in the third track, then move the note forward and snap it to the grid, as shown in the bottom track.Make names meaningful: another advantage of playing a part in context is that you should know the backing track’s tempo, and maybe the key. If so, create file names that begin with the tempo, then the key, then a description (‘135_E_technofunk_wah.WAV’ for example). This makes it much easier to browse for suitable loops when composing. Take advantage of long file names, too — be as descriptive as possible.

Time-stretching Basics

Back in the days of tape and early digital systems, pitch and time were inextricably linked: you could speed up playback, making the part shorter and the pitch higher; or you could slow things down, making the part longer but the pitch lower. So, before decent time-stretching became widely available, sampled loops always came as a set of standard audio files, and you’d have to stick with their ‘native’ tempo. To overcome this, some libraries included the same loops recorded at various tempos and/or in various keys. To use such libraries (there are still plenty made that way) you must find loops at the desired tempo and build your track around them. The advantage is that the sound quality is not compromised at all, because the loops aren’t altered in any way. The down side is inflexibility: you have to base your music on the loops, instead of finding loops that serve your music.

Now, to manipulate an audio part’s pitch and time independently we can use digital signal processing (DSP) of some sort. In essence, all DSP time-stretching algorithms either lengthen or shorten the audio by adding or removing data, respectively. Different algorithms achieve this addition and removal in different ways, and what information they add or remove has a huge impact on the quality of result.

A host that can stretch files will guess where markers should go, but optimising this usually requires human intervention. Note that the screenshots show mono files to conserve graphics space, but stereo files work as well.Unfortunately, there isn’t a single ‘best’ algorithm: every process has trade-offs, with different strengths and limitations. In all cases, though, it’s fair to say that the greater the change in tempo, the more challenging it is to create a loop without audible (and usually objectionable) artifacts.

Offline Time-stretching

The highest-quality time-stretching is achieved via offline processing, and most software DAWs include at least one offline DSP algorithm that can stretch or compress the duration of a loop. Often, you need only grab an audio clip and drag its end to shorten or lengthen it. For reasonable amounts of stretching, like -10 to +20 percent, the fidelity can be good enough that on most material you might not even notice the audio has been processed. The main advantage, then, is simplicity, but you might need to put in some work to get the best result. Some stretching algorithms offer different options for different types of audio (mixed, solo, bass, drums and so forth) and, despite the names, it’s often worth experimenting to see which gives you the most pleasing result.

The biggest downside is that your audio won’t follow any tempo changes in your project. You have to re-stretch/shrink-to-fit sections with different tempos, perhaps crossfading to tidy things up. If you want to change the global project tempo later, you’ll need to redo things each time, and dynamic changes, such as speeding up over a measure, are very difficult to do in most DAWs unless you cut the audio into little slices and stretch each one. That introduces its own problems, like audible clicks that require manual crossfading.

Dynamic Time-stretching

Technology marches on, and we now have various file formats and tools that allow audio to stretch with the project tempo. The tools that allow us to do this are based on the approaches pioneered by either Sonic Foundry’s Acid (which was later acquired by Sony) or Propellerhead’s ReCycle. Most DAWs can read Acid and REX (ReCycle) files, but can’t create them. Instead, they include their own versions of both approaches and so allow normal audio files to be forced to follow the tempo.

For example, with Studio One Pro, if you select the option ‘Stretch Audio Files To Song Tempo’ when creating a song, and later change the tempo, the audio obligingly follows suit. This can also be done after the fact with already-recorded audio, and is very much like the ‘varispeed’ control on old multitrack tape recorders, with the added advantage that modern DAWs preserve the pitch. You generally can’t get away with too much stretching, though, and in some cases you’ll hear artifacts, but for reasonable amounts of stretching, or providing useful demo guide-tracks for recordings, it doesn’t get much simpler. Cubase, Logic, Pro Tools and others have similar facilities. In fact, some DAWs — for example, Reaper — actually default to making your audio stretch with the tempo, which can be a blessing or a huge source of frustration, depending how you look at it!

The markers have now been edited. The red ones were placed by Sonar originally, markers with violet heads were moved or created, and the ones with hollow triangles were deleted. The clip, which was recorded at 80bpm, stretches well from 70 to 180 bpm.This sort of dynamically varying DSP facility being embedded within your DAW is probably the wave of the future, but dedicated stretchable file formats (such as REX, Acidised WAV, Groove Clips and Apple Loops) still offer more customisable characteristics that can potentially allow wider stretches, fewer artifacts, more creative options, or a combination of these characteristics. They also allow you to create loops that can be used (by you and others) in various different DAWs on different platforms. And that’s not just important for people making commercial loop libraries: perhaps you’ll jump ship to another DAW in the future and want access to all those loops you’ve worked hard to create over the years!

REX & REX2 Files

Propellerhead’s REX format, originally developed in conjunction with Steinberg, tackles time-stretching through methods other than conventional DSP. The approach is similar to that used in the early days of sampling, where some performers would cut up a phrase inside a sampler, map the samples across a keyboard, and then play them in time with the music. While Pro Tools’ Beat Detective was probably the first computer-based slicing algorithm, Propellerhead’s innovation was to automate the process of triggering the slices it created using MIDI: the REX process cuts the audio at transients to create ‘slices’ and triggers them further apart in time if the tempo slows, or closer together if the tempo speeds up.

Several programs now use similar approaches. FXpansion’s Geist, for example, can slice percussion loops, but has the added bonus that it can apply different filters to the individual slices.

Most DAWs are compatible with REX files: you drag-and-drop or import them, and they’ll automatically sync with the project tempo. Many DAWs also include virtual instruments/samplers which can map the file’s slices to notes or pads, allowing you to trigger hits via MIDI. Furthermore, they almost all offer a similar slicing process, whereby they can detect hits in audio events, and slice at the ‘hitpoint’ to create individual audio events on the project/arrange page, which will remain in sync whatever changes you make to the project tempo. (As these aren’t proper REX files, exporting them is usually not an option.)

REX files and similarly sliced loops have the potential for excellent fidelity because, although the audio is edited, it isn’t actually processed. However, there are important limitations. Each slice has the potential to produce a ‘discontinuity’ in the sound. If the slice occurs at a quiet moment, such as a gap between notes, it’s not an issue, but otherwise you’ll often hear a pop or tick. Careful editing of the slice boundaries can minimise or eliminate clicks, but for sustained sounds, it’s not always possible to hide the slice transition. In this case, the fix often involves creating a slight fade-in/out for each problematic slice.

The more you move the Sensitivity slider to the right, the more sensitive ReCycle is to transients. Go too far, and you’ll see a forest of slice markers. There’s a useful keyboard shortcut: the keyboard’s number keys select different sensitivity levels.What’s more, sustained sounds, including cymbals that ring over other beats, and anything processed with a long reverb tail, don’t lend themselves to slicing if other beats occur during the sound’s decay. Slicing at each beat also slices the cymbal/reverb tail — which again means discontinuity. This time, though, fading can’t help.

Then there’s the potential for unwanted gaps between slices when slowing the tempo. Although ReCycle can use DSP to extend decays, this is only effective if the stretching isn’t drastic. Speeding up is less problematic: the truncation of a slice is often masked by the attack of the next slice; you won’t notice it unless the tempo increase is significant.

Finally, REX optimisation requires ReCycle: creating REX files can be easy or difficult, depending on the source material, and some sound designers are more diligent than others about editing slice points optimally. Although most DAWs include similar functionality, only Propellerhead’s ReCycle lets you create, edit, and optimise REX files themselves.

In summary, REX files and similarly sliced audio files are ideal for stretching percussive loops using sounds with fast decays. Bass lines and staccato rhythm guitar parts are also excellent candidates. But if sustained sounds are involved, it’s usually best to use something like...

Acidised WAVs

Sony’s Acid, Cakewalk’s Groove Clips, Apple Loops and similar use virtually identical technology. They embed pitch and tempo information in the headers for WAV (for the first two) or AIFF (for Apple Loops) files. I’ll focus here on Acidised WAVs, the file format invented by Sony for use in their Acid recording software, but the differences are minor. Note, though, that while many Mac programs can read Acidised WAVs, Windows programs see Apple Loops as standard AIFF loops; they can load them, but I don’t know any Windows host that can interpret the stretching metadata. On the plus side, Apple Loops contain a few additional parameters.

Acid’s approach essentially blends DSP with REX’s slicing. Acidised WAV files mark transients in a file, but instead of slicing and moving the slices, they add or remove samples to lengthen or shorten the file. They then crossfade to smooth transitions in the processed loop. A major advantage is that these loops can follow tempo and key-change markers quite elegantly, and unlike REX files they often work well with sustained sounds. However, like REX files, Acidised WAVs have limitations.

Percussive loops that have been Acidised don’t always sound as ‘punchy’ as the equivalent sliced loop, for example, and the artifacts produced by slowing down Acidised loops can be severe enough to make the loop unusable, or at least unrecognisable. REX files, by contrast, can usually be used for greater tempo changes when slowing down.

What’s more, while playing back Acid loops is easy, creating and editing them takes considerable time and expertise. There are more options than with REX files, but optimising for one tempo range sometimes requires different edits than optimising for another. And while most programs can read Acidised files and follow the header’s instructions, only Sony Acid Pro and Cakewalk Sonar can create, edit and optimise Acidised files.

Another thing to watch out for is that level changes can occur when you change tempo. A loop that doesn’t clip at 120bpm might do so at 140bpm. Or the reverse might be the case — or there may be no change at all. So you need to pay close attention!

Sometimes it’s easier to see transients if you zoom in vertically. Here, the pencil tool is adding a slice at the beginning of a low-level hi-hat hit that was just below the Sensitivity slider’s range.One bonus of both REX and Acidised files is the ability to do some interesting signal processing. REX files can add rhythmic chopping to sustained sounds, and setting ‘wrong’ parameters with Acidised files yields potentially useful results. Acidisation editing can be used to perform fine pitch shifts and even varispeed-style tape effects. Similarly, Sonar allows the changing of level, pan and pitch for individual slices in a Groove Clip.

Other Stretching Options

The stretching options we’ve covered so far are non-proprietary, in the sense that while they might need specific software to create the loops, multiple programs are able to read their metadata and apply the correct amount of time stretching/compression. However, as I’ve mentioned above, other software offers stretching facilities that are unique to those programs.

For example, Pro Tools has a stretching engine called Elastic Audio, Logic has Flex Time and Cubase has Audio Warp. Ableton Live uses several stretching engines to accommodate different types of material; some are more slice-based, some more DSP-based, and some are designed to accommodate tempo variations and either preserve them or smooth them out. Some loop libraries also have their own stretching options, like Ueberschall’s Elastik Audio Engine (based on Zplane’s DSP technology) and ILIO Groove Control.

Acid & REX Optimisation

If you’ve followed all of the above advice, then you should have a loop that will be suitable for stretching over a wide range, and then drop right into a song — the next step is to actually make it stretchable. I’m going to run through ways to convert your loops to Acidised and REX files. Apple Loops are handled the same way as Acidised files, so the techniques are pretty much interchangeable.

There are a few tools available for creating Acidised, REX or Apple Loops files, but the only free option is Apple’s Apple Loops Utility SDK (Software Development Kit). Register as a developer (it’s free), then go to https://developer.apple.com/downloads/index.action. Enter ‘loop’ in the search box and you’ll see the option to download Apple Loops Utility SDK (3.0.1 at the time of writing). Creating REX2 files requires ReCycle. For Acid file creation, any Cakewalk Sonar version can be used (the base version is inexpensive, but you could pick up an older version for even less), and the same goes for Acid (though the Acid Music Studio version doesn’t have the same deep-editing options). Another option is Image Line’s FL Studio. So it’s worth finding out what tools you already have before investing in more!

The Ideal Acidised WAV

When you stretch a poorly Acidised file, where the markers that should define how to stretch the file haven’t been properly positioned, you will hear audible and undesirable sonic artifacts such as doubled attack transients that sound like flamming, fluttering on sustained material, and a loss of some of the ‘ring’ or fullness from percussion and other sounds. It’s hugely important to ensure that every significant transient in the file has a marker identifying it as a transient, and that this marker is exactly at the transient (which, it’s worth stressing, may or may not be the very start of a note — think of reverse cymbals, for instance, where the most critical timing information is near the end of the ‘hit’.). If you plan to reduce the tempo, also include markers within sustained sections, so the DSP that’s applied to stretch the sustained segment doesn’t have to stretch too far.

The upper track is the original 120bpm, two-measure drum loop. The middle track has been ‘sliced’ at each drum transient, and the tempo changed to 100bpm. The interval between triggering successive slices is longer, which reflects the slower tempo. The lower track has DSP stretching applied to fill in the gaps, which in this case also removed any clicks due to transitioning between audio and silence.When a loop repeats, the splice must be seamless, so the start and end levels must be identical. It’s a misconception that these have to be at ‘zero crossings’ — they can be any level that’s equal — but the zero-crossing approach is easiest. Some DAWs allow you to make edits at the nearest zero crossing, but adding a very short fade-in and out puts a zero point exactly where you need it! The optimum fade time is about 2-4ms (100-200 samples at 44.1kHz). If your DAW allows you to specify an automatic fade when splitting a clip, that will save time. But you still need to check that things sound right, so copy the clip several times, placing the copies end-to-end to play back through the transitions. If there are no clicks or pops, no fades are required. If there are, some loops may need a longer fade, or a concave instead of convex fade curve — listen and experiment. Remember to save the fade with your clip by bouncing/rendering it.

Acid Indigestion

The Acidised version of a regular file that plays back without clipping will not distort when played back at its ‘native’ tempo. However, speeding up or slowing down the song may cause volume increases that push the level past the point of overload. This is because the DSP relies a lot on crossfading: if phase-coherent audio sections overlap, the signal peaks can sum and exceed the level of the original clip. This isn’t predictable; a 120bpm clip may distort at 80 and 140 bpm, but not 160. I normalise files destined for Acidisation to -3dB, which leaves enough headroom to prevent this.

Next, I’ll run through some examples using and referring to Sonar, but the same principles apply to Acid and the Apple Loops Utility. Bring the file into Sonar and open the Loop Construction window (equivalent to Acid’s Loop Properties page, or the Apple Loops Utility’s Transients tab). You’ll see a lot of markers as the program attempts to identify the transients, but some markers will be missing. Sometimes the analysis is ‘close enough’, such that some people would hit save and consider their work done. But for the best results, which enable stretching over the greatest possible range, you usually need to tweak the marker positions.

A slow tempo exaggerates problems, so drop the tempo by 20 percent or more and play the loop. Audible flamming usually means a marker is not sitting exactly on a transient. You may also hear notes jam together, as if that section of the loop were sped up. This tells you that a marker is missing at one or more notes.

Some general rules for when moving, adding, and deleting markers in Acid, Sonar or the Apple Loops Utility are:

• All transients should have a marker exactly at the transient’s beginning. Zoom in to ensure precision. As with the loop starts/ends, try to place all markers at zero crossings.
• Secondary transients, such as those you’d find in a delay or flam, should have their own markers.
• With sustained material, add markers at rhythmic intervals such as on every quarter or eighth note. This tells the DSP to create a crossfade to help effect a more seamless transition. Placing this fade on a beat means discontinuities caused by stretching are likely to be masked by other sounds.
• If you hear ‘fluttering’ during sustained notes, add markers in the middle of the notes. This can improve the sound quality at slow tempos, but degrade it at faster tempos.
• Sometimes adding a marker at the end of a note’s decay prevents roughness toward the note end. Again, this can help at slower tempos but work less well at faster ones.
• Remove spurious markers (those not at transients or rhythmic intervals), as they perform no useful function.

Acid offers four different stretching options: Looping Segments (this is the most commonly used option), Non-looping Segments (best for sustaining material), Pitch Shift Segments (shifts pitch to adjust for tempo variations, like conventional resampling), and Sliced Segments (best for highly percussive material). If you find that a file doesn’t stretch properly with Looping Segments, experiment with the other options.

All three stretching programs can add markers at specific rhythmic intervals if transient detection or sensitivity are set to minimum. Increasing transient detection sensitivity places markers at additional detected transients. Not all transients are obvious. For example, a fill and cymbal crash might play simultaneously at the end of a drum loop, so you can’t see the individual transients. Listen to the part: if there’s a hit every 16th note, just place a marker at every 16th. If it’s mostly 16th notes but there are some hits that extend over an eighth, add hits for the 16th notes but omit them for the sections that are longer.

Keep listening and tweaking until everything sounds right before increasing the tempo and listening again. This makes it much easier to hear if there are too many markers. For example, on a sustained section in which there are lots of unnecessary markers, you may hear a roughness or discontinuity. If you can hear this, try removing some markers and decide whether that improves the sound. By alternating between very slow and fast tempos in this way, you’ll find the optimum marker placements for stretching over the greatest range of tempos. Finally, return the loop to its native tempo. Check that it sounds as it did before processing (it should).

With melodic loops, specify a root note if you can. Doing this allows the host DAW to transpose the file if there are key-change markers embedded in the sequence. Note that a host’s DSP transposition algorithms often provide superior sound quality, so sometimes it’s best to set the song to the correct tempo, convert the stretched loop into a standard audio clip, and apply the DSP-based transposition.

Finally, save the file. With Sonar, drag the loop from the Clips pane to a folder or your desktop. It will be saved as an Acidised WAV. With Acid or the Apple Loops Utility, any changes you make are saved when saving the file.

The Ideal REX File

ReCycle saves the digital audio slices and MIDI information in a single file called a REX file. There are older REX file formats, but the most important, and recent, is the REX2 format file which has a ‘.RX2’ suffix. REX files also deploy a lossless compression scheme, and this means that they tend to take up less space than WAV or AIFF equivalents. Most hosts can load and interpret REX files, but can’t extract the MIDI notes. Note that Reason can extract these MIDI files so they can be rearranged to create loop variations, as can Sonar’s RXP REX file player.

Many of the same principles that apply to Acidised WAVs apply to REX files — so, again, you want to find transients. However, whereas for Acidised WAVs it’s usually better to have too many markers than too few, REX files should not have more slices than necessary. (The same also applies to ‘hit points’ in Cubase, Sonar’s ‘Split Beats into Clips’ function with AudioSnap, and similar processes.)

To create a REX file, open the audio file in ReCycle. If asked whether you want to move the left locator to the first slice point, choose Yes (unless there’s supposed to be ‘air’ at the start). Then:

• Set the loop length and enable the Preview Toggle option, which lets you hear the results of adding and auditioning slices.
• Move the Sensitivity slider right. ReCycle will hunt for transients and add slice markers. Adjusting this slider may locate all transients with simple, highly defined drum patterns, but you’ll likely need to edit slices manually (more later). Each slice should have a single hit, whether it’s a discrete sound or layered sounds (kick, snare-plus-kick, bass note, spoken word, hi-hat...).
• Slow the file way down (for instance to 70bpm for a 100bpm file) and play it. You’ll hear each slice as a discrete ‘blob’. For fine-tuning, click within a slice to trigger only that slice.

Problems fall into three categories: clicks or pops at the slice’s beginning or end, missing transients and/or spurious transients. To deal with clicks, move the slice marker associated with the click so it falls on a zero crossing (in both channels of a stereo file). ReCycle will try to ‘snap’ the marker to a zero crossing with stereo files, but may not find the perfect slice point. Most of the time, a slice wants to go as close to the beginning of a transient as possible, but sometimes you’ll get better results if it sits slightly before or after what, intuitively, you’d think would be the optimum point.

You’ll hear if a file needs additional slice markers when listening at the slower tempo. Suppose two 16th notes play but there’s only a slice marker at the beginning of the first one. Because each slice plays back at its original tempo, the notes will appear to hit in quick succession. Placing another slice at the beginning of the second 16th note defines it as a separate note. And if there are too many markers (evidenced by ‘stuttering’ sounds or the splitting of sounds in the middle), click on each offending marker’s head and hit Delete.

Eventually, you’ll nail where all the transients are. With a relatively simple drum part, you may not even need to do any editing at all: there’s an option that allows placing slices at every 16th note, and you’d be amazed how often this is all you need for a lot of EDM! However, bear in mind that some material such as sustained or dense sounds simply isn’t very REX-friendly.

The penultimate step is to slow the file to the slowest reasonable tempo you plan to use, then turn up the Stretch parameter until you don’t hear any gaps. This doesn’t always solve gapping if you’re slowing down a lot, but it can help (and with drums, be virtually transparent). If turning up Stretch causes a big volume swell, locate the marker just after the swell, and move it earlier. Finally, play the loop back at a much faster tempo. If all is well, you’re done. Otherwise, you need to retweak markers.

Incidentally, ReCycle also includes three signal processors. Envelope adds attack and decay times to each slice (globally). Short decays can give percussive effects. The Transient Shaper provides dynamics control, including compression-type effects. EQ has low-cut, high-cut, and two parametric stages. The Pitch control transposes the file (try outrageous amounts for special effects), and Gate Sensitivity adds a type of noise gate. These can not only clean up existing files, but can add very cool special effects.

The Long & The Short Of It

And that’s pretty much it — if you have access to the tools required to make Acid or REX files, you should now know everything that you need to do to create loops that work perfectly over a wide range of tempos. But even if you don’t have those tools, you should have a better grasp of how to make use of the stretching (and shrinking) tools that come bundled with your DAW. And remember, even if your DAW doesn’t have dedicated stretching tools, you can always perform REX-style slicing on the project/arrange page, simply by using your scissors to snip at each and every transient. Assuming the audio track is set to ‘musical’ rather than ‘linear’ timebase the new clips will retain their musical position as you change tempo. It’s a more tedious and less flexible way of slicing (and you won’t be able to build up a decent loop library in this way), but bear in mind the advice I’ve given you for making decent REX loops and you should still be able to achieve good results.