Using the FMOD Scatterer for music

Explore the FMOD Studio Scatterer with Romain Raynal as he describes how it is a tool that he frequently use for all sorts of applications in sound design, but discovers its potentional to unlock new possibilities with music.

Contact Romain at LinkedIn

FMOD has become an essential audio middleware in recent years, especially within the independent gaming scene. Seen as a more cost-effective alternative to its competitor Wwise, FMOD also has a somewhat different philosophy. While Wwise may initially be perceived as a rather "stern" programming tool, FMOD presents itself more like a DAW with programming capabilities. Everything is very visual and intuitive. However, readability diminishes as the project becomes more complex.

FMOD also has some very interesting unique tools, such as the Scatterer Instrument which we are interested in today. To summarise its operation, it allows sounds to be triggered randomly (often short), with equally random parameters: volume, pitch, spatialisation, etc. This is very handy as these parameters can be modified by the player's actions; the possibilities are then enormous. It is a tool that I frequently use for all sorts of applications in sound design. But I had never considered it for music.

In this article on the FMOD website dedicated to Creaks, we discover the work of sound designer Matouš Godík, as well as that of composer Joe Acheson (better known as Hidden Orchestra) who is trying his hand at video game music for the first time. The interview is very interesting as Joe explains his discovery of interactivity through the FMOD tools and his exchanges with the team. He particularly details a musical use of the Scatterer Instrument that I would never have thought of. By aligning the appearance rate with the tempo (which is made all the easier now), we can create generative and random music. It's a very interesting idea that I have decided to explore through this article. You can also listen to the soundtrack of Creaks in an interactive way.

First test: Random percussion

First thing to clarify: since the publication of the interview, FMOD has equipped the Scatterer Instrument with a tempo locking option. We will simply regret the inability to define a range, which forces us to choose the moment of appearance in a deterministic manner. It would be really interesting to be able to select multiple values, or at least to be able to automate them and have them controlled by modulators (which is also currently impossible).

I start by creating a simple musical loop. Then, I create very simple percussive elements with random durations and no recognisable rhythmic pattern.

I export everything from Cubase by adding an extension time to account for the reverb tails (it's much more demanding than Reaper's export system). Here's how I proceed:

I then create a 2D Timeline Event in FMOD, I enter the tempo and looping data in the Logic Track, I import the musical loop and create a Scatterer Instrument in which I import each small piece of percussion.

After entering the tempo in the Logic Track and setting the loop areas, I set the occurrence rate to two bars. To create variation in spatialisation, I configure the Min & Max Distance to 1.00-1.00: the sound is therefore played randomly around the listener with occlusion at approximately one metre away.

• My settings:
• Spawn Rate : In tempo mode, this affects the maximum number of simultaneous events triggered according to the quantisation parameters.
• Spawn Total : We do not use this parameter here.
• Volume Rnd : We could use it, but it is not essential in this case.
• Polyphony : I do not touch this setting, as by fixing the parameter at 2 bars, there is no chance that more than two events will overlap.

This is the result in FMOD :
https://youtu.be/2As2ndMbsgg?si=7Zy75e72nwv9T9wr 

Second test: The random melody

After an initial attempt based on percussion, let's move on to a second test more focused on melody to see if it proves relevant.

I am composing a music loop of the "orchestral action" type and I am using a Santour to generate just under twenty musical phrases:

I then import everything into FMOD:

💡Tip: To maintain stereo sound (and not spatialised), you need to set the Min & Max Distance to 0.00-0.00. The event becomes a 2D Event and the sound is played exactly as it was mixed in stereo.

Quickly, I find myself faced with a first problem: the operation of the Scatterer Instrument within the Loop Region is somewhat enigmatic. As I am using this Loop Region for my music loop (to ensure a smooth transition), I cannot do without it. I therefore find a setting that seems suitable.

Second problem: given the absence of a trigger modifier, if I set the duration to four bars (which is the exact duration of each audio segment), there is no difference from using a Multi-Instrument that would be placed every 4 bars. I then attempt an asymmetric layering approach by setting the "pop" every 3 bars: thus, each phrase overlaps a little with the previous one to create a greater sense of unpredictability.

However, this shifts the timing of the phrases, which are initially intended to start on the first beat. In the end, some phrases sound a bit awkward, while others fare better. This is an element to anticipate if you plan to use this technique in production. Finally, I add compression and a delay to contain the sound and prevent the ends of phrases from sounding too abrupt.

This is the final result :

https://youtu.be/FTHqsgEz1v0?si=S07ZUNIkwbJC9Pxk 

Third test: Double random

So far, I have tested a fixed track accompanied by a random track. Let's see what happens with two random tracks. As it is impossible to anticipate the harmonic and melodic construction, it is musically more "experimental".

I decide to go for a hammered piano accompaniment with eighth notes and "dark jazz" type chords, quite disconnected from each other. For the melody, it is simply the C minor scale. Here is the Cubase session, which I will cut up before exporting everything to FMOD.

Here are the two Scatterer Instruments: at the top the piano that plays the chords, and at the bottom the synth that plays the melody:

The piano is configured so that it never overlaps. Each phrase lasts 4 bars and the appearance rate is set to 4 bars. In this strict configuration, there is again no difference from a simple Multi-Instrument.

To add unpredictability to the melody, the phrases last about 4 bars, but I set the occurrence rate to 3 bars. As in the example above, this allows for overlapping phrases. Since these are single notes, the layering works quite well.

Here is the result: the overlay is convincing, even if the harmonic coherence remains relatively hard to digest:

https://youtu.be/46EN3JcbdpY?si=93KtYKDon5YZ08Qp 

Last test: Out of time

Ultimately, to fully take advantage of the random spawn feature of the Scatterer Instrument, I need to let go of the Tempo approach. Therefore, I am trying a rubato approach with a non-rhythmic drone that loops. On top of that, I place "non-metric" notes, exported and integrated into FMOD.

In the Ableton Live session (the drone at the top, the different notes at the bottom), it is clear that each note has a random duration.

• I keep the event in 2D.
• I set the polyphony option to 2 so that the notes can overlap.
• I add reverb to lengthen the notes and facilitate transitions.

Here is the result when we mix the drone and random notes:

https://youtu.be/9_jUTiWYnms?si=QRlnm6YTgDfrZVjD 

Conclusion: Is it really useful?

If the first three attempts left me a bit wanting (much too close to what one could achieve with a simple Multi-Instrument), the last test opens a path towards something truly interesting. Thanks to a few notes and a relatively short musical loop, I was able to generate movement and avoid repetitiveness.

I think about it in hindsight, but I could have played even more with the Spawn Rate in this last example. As I am not using the Tempo function, increasing the spawn rate allows for more or less frequent random triggers within the chosen time frame. In Tempo mode, it simply increases the layering of phrases at each trigger without any real variation in rhythm.

And this is precisely what poses a problem with the current Tempo mode of the Scatterer Instrument: without the ability to define a range, the tool ends up functioning too rigidly and loses its specificity. But then, why not set a range on the tempo using the Time mode? It is quite easy to calculate a timing in milliseconds. Simply because between the two values, the spawnwould not be based on the rhythmic grid, and the musical phrases could end up completely misaligned with the strong beat. Nevertheless, I remain optimistic: the emergence of the Tempo mode proves that Firelight works on the subject. Who knows, a "quantified time slot" could emerge in one of the upcoming versions.

On my side, I am curious: had you ever heard of this method? Do you use it? Will you use it following this article? I really enjoy exploring different approaches to interactive music and sound design, so if you have any other techniques of this kind, I would be keen to hear them!

Original article written by Romain Raynal (https://www.romainraynal.fr/en/blog/blog-4/using-the-fmod-scatterer-for-music-18)