Recipe - Changing chords

Key idea

The goal here is to have clips launch out of phase with each other, meaning they don’t all change at the same time. The clips also use probability on the chord notes, so that each time the clip plays we hear a slightly different selection of notes. Across the clips we’re going to try to imply slightly different chords or tonal centres by changing the “root” of the chord in each clip. The combination of these elements creates a shifting pattern of chords and textures.

1. Create three MIDI tracks, each with a virtual synth instrument. For this example we’re going to use long chords, so I’d recommend a pad or arpeggiated sound. One thing that works quite well is to use instrument sounds with rhythmic elements, sounds that evolve or where there is modulation to vary the sound over time.

2. Make three MIDI clips in each track.

3. Prepare a MIDI clip with 8 to 10 notes in key (whatever scale you wish) in each clip within the tracks. Avoiding the 3rd of the scale will make the chords tonally “ambiguous”, and avoiding (major) 7ths will help the sounds produced blend between tracks and clips (see the chapter on Scales Process - Using scales for more discussion of scales in generative music). Use the Scale button in the MIDI clip to only show notes within the scale. The chords suggested are arbitrary, but the recipe will work best if you use scales or chord intervals that are “next door” to each other in the cycle of 5ths - https://en.wikipedia.org/wiki/Circle_of_fifths. See the screenshots below for a visual representation of what I mean.

4. Use random note probability on notes in the chord (see Process - Note-wise probability ). Set the root notes of each clip to 100% probability, and set other notes at about 60-80% probability so that each time the clip triggers, the note choice / chord is different. This means that while the chord might sometimes have 4 notes, it could vary anywhere from 1 to 7 (technically even 0 or 8, though those extremes are very unlikely). Having higher probability on the root notes favours these notes within the chord, but if that probability is slightly less than 100% then you’ll not ALWAYS get that anchoring from the root note.

(Because the notes in the clip could come from a variety of chords, you can use a bass track to imply a shifting tonality by anchoring root notes of chords, not necessarily the “root” note set through 100% probability in the MIDI notes specified).

5. Make sure that the clip lengths vary between tracks (all notes within the clip have the same length). So clip 1 in track 1 could have 16 bars length; clip 1 in track 2 could be 13 bars; clip 1 in track 3 could have 10 bars length etc. The exact clip length isn’t important; what matters is that they vary across tracks.

6. ALSO, vary clip lengths between scenes within tracks. So a track that has 16 bars in clip 1, maybe has 14 bars in clip 2 and 12 bars in clip 3. Similarly, the exact length is unimportant. The important thing is that the clip lengths vary both within tracks and between tracks. AVOID having the same clip lengths in the same scenes. Following the advice above you should wind up with something like this in the first three clip slots of the three tracks:

Track 1 clips:

Track 2 clips:

Track 3 clips:

The clips shown above are just my choices for a slow-moving, textured track. You can play with different choices of notes, shorter lengths of clip, more plucky sounds. Perhaps adding an Arpeggiator or Note Echo to the track would help movement and retain interest (see Tools - MIDI tools for more details on MIDI effects in Ableton Live).

Another option would be to write short melodic phrases and counter-melodies in clips and then have the Follow Actions choose which to play. This could then produce a wide range of interesting movement across the tracks. We’ll look at this in slightly more detail in Recipe - Balancing the unexpected and the predictable.

7. Set up Follow Actions (see the following chapter Process - Follow Actions ) to progress to Next clip at end of each clip with 80% probability, with the other action set to “Play again” with 20% probability. When you launch the scene, each clip will play to the end of the clip, but because individual note probability within each of the stacked chords is set to 60-80%, the chord will sound different each time it plays. Also, the randomness we’re using for note probability means that the same chord clip in each track is also likely to sound different. The 80-20% split of next action - “Play again” or “Next” - will mean that some tracks may progress to the next chord while some will play the same clip again.

If the clips are short, then a 50-50% split of next action chance might be more appropriate. With shorter clips then we might favour repetition so that the listener isn’t bombarded with changes every few beats (see Process - Balancing the unexpected and the predictable for a discussion of balancing repetition with unexpected elements). In this case because the clips are long we want to force movement onwards.

8. Choose sounds for each track that complement each other. You can automate parameters within clips as well to introduce additional movement in the sound. An important point is that the sound shouldn’t be too static.

9. Press play. Listen to what you hear, and refine settings if necessary.

You can hear an example of what this sounds like here…

https://soundcloud.com/mikeksmith/followactions_chords_and_arp/s-CKzkRoOEWrr

MikeKSmith · 1_FollowActions_Chords_and_Arp

(I say “one instance” because this is generative and uses probability. So the next time I render the track it could well sound different. Similar, but different.)