You are here

Quantise Drums With AudioSnap Pool

Sonar Tips & Techniques By Craig Anderton
Published November 2011

If you'd like to make your quantised drums a little more human, dive into Sonar's AudioSnap pool.

The upper track is, again, the human-played track. The track of kicks immediately below it has been quantised to the upper track, while the lower track has not been snapped yet.The upper track is, again, the human-played track. The track of kicks immediately below it has been quantised to the upper track, while the lower track has not been snapped yet.

It's true: when you quantise drums played by a human to a grid, it sucks the life out of the part. Of course, there are exceptions: electro's raison d'être is the metronomic beat, so quantising a human-played drum part makes it fit the music better. But for most music — especially music with a great drummer — the ebb and flow of timing gives 'feel', and provides the rhythmic equivalent of dynamics.

I recently took on a remix project with electronic drums that emphasised the four-on-the-floor kick drum we all know and, uh, love. The female vocalist put major emotion into the song, and the contrast between her and the metronomic drums just didn't work for me. So I reached for the Discrete Drums loop library and grabbed some loops by Nashville drummer Greg Morrow (a powerhouse drummer whose drumming fit in well with the vibe of the vocals), as well as some individual hits.

However, his parts lacked the heavy kick that's essential to much dance music. The tracks I was supplied had two kicks, which suffered from slight timing issues with respect to each other. I'm not a fan of kick flamming, so I locked them to the grid (hey, it's dance music), but then they conflicted with Morrow's drums, so I locked him to the grid too.

Big mistake. All the acoustic drums' power was lost, so I set about quantising the kick tracks to the human, not the other way around. What makes this possible is AudioSnap's ability to designate one track as the master 'pool' of timing transients and snapping other tracks to the pool. We'll investigate doing this with the kicks, but the same principle applies to locking any track to another track's groove.

Candidate Clips

AudioSnap won't work with Groove Clips, so they have to be converted to standard clips first. You can select an Acidised clip and simply invoke Bounce to Clip(s) from the Clip Edit menu, but if the target tempo isn't too far from the clip's original tempo, I prefer turning off Groove Clip looping (select the clip, then type Ctrl-L) and using the iZotope Radius DSP-based, offline time-stretching. To stretch or shrink the clip length, turn on Snap (typically to one measure) and, while holding the Ctrl key, click and drag the clip's right edge to the desired length.

The clip will play back using the preview 'online' stretch algorithm, which is acceptable for auditioning, but not for a final mix. To choose a clip's 'offline' rendering algorithm, select the clip, then type 'A' to open the AudioSnap palette. Choose 'Applies to Clips', then select the Offline rendering algorithm. Radius Mix (the default) works very well, but you may obtain slightly better results with percussion if you choose Percussion for the Online option, then for Offline, choose Same as Online. You can always try both, then decide.

Finally, select the clip and invoke Bounce to Clip(s) from the Clip Edit menu; bouncing occurs using the selected Offline algorithm. Now that we have a standard clip, it's time to start snapping.

Filling The Pool

The primary AudioSnap requirement when creating a pool is to define exactly where transients fall, as this creates the grid to which other transients snap. Ideally, AudioSnap would be a relatively automatic process, and it is with metronomic material containing defined transients. But it's a different story to create a grid for complex parts played by humans. This is somewhat similar to the issues involved in creating Acidised or REX files, where you also need to define transients precisely.

    • To designate a clip as the one that defines the pool, select it and type 'A' to open the AudioSnap panel. The 'power' button should be on.
    • Click on Edit Clip Map to show a map of the transients, then vary the Threshold slider to adjust the sensitivity to transients. For example, in this track the kicks were fairly strong, so I could set a relatively high threshold to prevent interference from other drum hits. Transient markers are visible, but note that they don't all line up precisely with the kicks. We need to line them up to create a grid that other tracks can follow.Transient markers are visible, but note that they don't all line up precisely with the kicks. We need to line them up to create a grid that other tracks can follow.
    • Now the tedium begins. Zoom way in (you want the transient markers to line up as closely as possible), and use the Smart Tool to click and drag on the middle diamond of each transient marker that needs to line up with its associated transient. If there are extraneous transients, you can right-click on them and delete them if you won't be needing them, or disable them if you think they may be necessary at some point.
    • After defining the transients, right-click within the clip and go to Pool / Add Clip to Pool. The transient markers turn from white to turquoise to indicate that they are serving as the 'snap masters'.
    • Now click on the track you want to snap to the pool, and follow the same basic procedure you used with the Pool clip. Type 'A 'to open up the AudioSnap palette, click on Edit Clip Map, and move the threshold slider until the transients have markers.
    • Zoom in to verify that markers indeed indicate transients. If they don't, click on the marker with the Smart Tool, and drag until it lines up. Again, zoom in to make sure the positioning is as precise as possible.
    • Next, right-click within the clip you want to snap to the Pool, and select Pool / Quantize to Pool. A 'Quantize to AudioSnap Pool' window appears that lets you set the quantise strength, which in this case needed to be 100 percent to avoid flamming. You can also set the window within which quantisation occurs; if a transient is too far from the target quantisation resolution, it won't be quantised. The assumption is that, if it's that far off, it's intentionally not on the beat defined by the grid. After setting the window and strength, click on OK.

Make It So

But we're not done yet. Although we've defined the transients and done the proper snapping, to make these changes a permanent part of the snapped clips, we need to render the clips that snapped to the pool.

The choice of rendering options, which are toward the right of the AudioSnap palette under Render Mode (see screen above), is important. The Online option is for previewing, and will use standard Groove Clip (Acidised) stretching. The Offline option default is Radius Mix, which is a fine-sounding time-stretch algorithm. For small tweaks, Radius Mix is outstanding. For larger changes, Groove Clip stretching may sound better and, as mentioned previously, Percussion can work well for purely percussive parts. Note that the Offline pop-up menu gives six different options; the proper choice depends on the material being stretched.

Select the clip, and then, from the Clips Edit menu, choose Bounce to Clip(s). This takes a while — especially with longer files — because of the DSP calculations needed to preserve fidelity. The human drum part (in yellow) is on top, with the electronic kicks (in white) below. In this section, note how Greg Morrow hits the kick very slightly late to add tension: this happens because the kick doesn't hit exactly when expected.The human drum part (in yellow) is on top, with the electronic kicks (in white) below. In this section, note how Greg Morrow hits the kick very slightly late to add tension: this happens because the kick doesn't hit exactly when expected.

After bouncing whatever needed to be bounced, if you don't need the reference clip to serve as the pool, you can turn its track Edit Filter from Audio Transients to Clips. It will now appear like a standard audio clip. Also change the Edit Filters for the snapped tracks to Clips.

Manual Labour

There was one more detail: the bassist was lined up with the electronic kicks. I prefer not to quantise everything to everything, so I left the bass part alone, except for adjusting the notes that hit on the beat. As with the electronic kicks, these tended to be a little ahead of the beat, which interfered with the feel. Setting transient markers in a 'human' track with Audiosnap, to define a timing grid for other tracks to snap to. The transient marker (in yellow) has been placed at the beginning of the transient to its immediate left, and is being lined up with the kick in the upper track. Upon releasing the mouse, the transient will line up with the marker.Setting transient markers in a 'human' track with Audiosnap, to define a timing grid for other tracks to snap to. The transient marker (in yellow) has been placed at the beginning of the transient to its immediate left, and is being lined up with the kick in the upper track. Upon releasing the mouse, the transient will line up with the marker.

Although most people use AudioSnap as an automatic process, you can just as easily turn on AudioSnap and just tweak a few transient markers that sound obviously off (just remember, if it doesn't sound wrong, it's not wrong!). While we don't have space to go into all the ways you can do manual adjustments, the basics are simple: click and drag a transient marker's middle diamond to position the marker, then click and drag on the transient marker line to actually stretch the audio between that marker and the nearest neighbouring markers. If a transient is missing a transient marker, no problem: Alt-click where you want the marker to sit.    

Buy Related Tutorial Videos