Exploring Sound Design Tools: Igniter

Igniter is Krotos’ new engine sound design plugin. They have been kind enough to sent me a license to have a look and see what it can offer. Igniter allows you to virtualize vehicle engines (real, sci-fi or fantastic) combining a granular section, a set of synthetizers and two sample managers. It includes performance controls so you can automate the vehicle RPM, engine load and many FX (including doppler) in order to get a realistic sounding engine. It comes with a big amount of presets including sport and utilitarian cars, planes, helicopters, trucks, motorbikes and sci-fi vehicles.

So here is my in-depth look at the plugin features with some examples here and there. I encourage you to follow along in your own DAW, you can find a full featured demo here.

Interface / UI

The interface is clean and easy to read and you can resize the window which is very nice. A main section (left side) occupies most of the screen and includes all the audio sources we can use. These sources are divided into four different tabs: Granular, Synth, One Shot and Loop and also includes a file browser.

On the right hand side we find the engine master on/off switch and the main revs knob in the middle. This revs knob acts as a gas pedal for the whole plugin. At the top, we find the Mod system, where Igniter’s true power resides, since it allows you to dynamically link any parameter within the plugin to the revs knob using envelopes and LFOs. Lastly, at the bottom right side, we find the FX and mixer sections. Let’s see all these in more detail.

If you need more info, most features are well covered in the manual and on Kroto´s videos. What follows is my own take on the plugin capabilities, plus some wish list features that I would love to see in the future.

Granular Section

This is probably the most complex and important generator. It combines granular synthesis with real recordings to re-create a virtual engine with a revolutions or RPM knob that you can “drive”. Each vehicle includes two mic perspectives: engine and exhaust and we can easily mix between both with an slider.

When I saw this, it occurred to me that it would have been nice to also include an interior perspective as this would be very useful for vehicle scenes like chases. After some looking around, I discovered that all vehicles have a “In-car” preset which solves the problem. But this is not a true recording of the car interior, but a recreation of it using EQ and convolution reverb. Would this be too different or unauthentic compared to a true inside recording? To be honest, I don’t know since I don’t have a huge amount of experience doing car sound design but I suspect these presets will suffice pretty well for most applications and, of course, you can always tweak them to suit your needs or even do your own "in-car” processing outside of Igniter.

In terms of how the granular engine actually works, we can’t see what’s going on under the hood (see what I did there?) but I suppose the plugin is using recordings at different steady RPMs and blending them together as you act on the engine. This is similar to the approach used in middleware like Fmod for use in video games. The result is pretty natural and smooth and driving the RPM feels responsive and clean.

For now, you can’t add your own sounds to the granular section, as they would probably need to be edited in a very specific way for them to work here. On release day, Krotos offers 13 different vehicles that use this granular option but I’m sure more will be added with time or maybe made available for individual purchase in the future.

Driving Modes

As you can see on the interface above, there are basically two ways to control the engine simulation: manual and auto. At the same time, every car comes with a set of three presets, two of them are manual and a third uses auto. So, using the example pictured on the right hand side with the Dacia 1310:

-Dacia 1310: “Free” mode that uses manual driving.
-Dacia 1310 Manual Gears: Uses manual driving but with pre-determined gear shifts on the Revs progression.
-Dacia 1310 Auto Gearbox: Uses the auto mode.

Let’s see what’s the difference between these three:

In general, manual mode allows you to freely change the engine’s RPM an also gives you a “load” knob. This parameter simulates if you are putting pressure on the engine or, in other words, if you are applying pressure on the gas pedal or not and allows us to create more realistic sounding gear shifts and decelerations.

The difference between both manual presets is that on the “free” mode, the relationship between the granular RPM and the master rev knob is completely linear by default, so you have to play with the RPM value yourself to imitate the act of shifting gears. Here is a video of me just doing that with a Revs pass first, followed by a Load pass. As you can see, to achieve a natural result, you need to drive the parameters in a realistic way. It would require a bit of practice to follow onscreen action like this but it feels very easy and responsive.

On the other hand, the other preset type, “Manual Gears”, has the gear shifting already soft coded into the mod section, including on load and off load changes. Of course, you can tweak this as you please but the preset gives you a nice starting point. As you can see, in this mode you don’t need to imitate the engine revs with your automation and you can just use curves to describe how hard you want to accelerate or decelerate.

For the most part, this works quite well when going up on the revs but going down forces you to go through the whole set of gears which doesn’t always feel natural, although sometimes you may want this (Formula 1 cars kind of do this sometimes). I tried different ways to avoid this, like staying within the boundaries of the same gear or jumping fast from a higher to a lower point on the envelope, although this needs to be carefully drawn as automation. A potential solution to this issue would be that the RPM ramps don’t occur when we decelerate, only on our way up on the revs knob.

You can also notice how the load drops are already coded into the revs progression, which is pretty handy and also shows that I was too subtle with it on my free test.

The third preset, Auto Gearbox, uses the auto option which doesn’t allow you to directly control the granular RPM or load and simply gives you an slider called “Power” that we can use to accelerate or brake while the gears shifting is hard coded and can’t be tweaked. This would be similar to driving an automatic car.

Here is an example of me using this mode. Compared to the others, it feels a bit unresponsive at the start but once you get speed it works well, although the gear shifting doesn’t always feel “in the right place”. As long as you don’t need very precise and fast changes in RPM, this mode can be useful to get natural results quickly.

By the way, you may hear some clicks and pops on my examples above. I am not sure a 100% if this is coming from Igniter’s or was a internal audio recording problem but definitely the Audi R8 seems to be a bit more “clicky” on the exhaust than other cars I tried later.

Granular Advanced Controls

Lastly, the granular section also includes some other advanced controls:

-Shuffle Depth controls how thin or wide is the slice that the granular engine uses to select the samples. Higher values can help make the sound more natural and varied. Using the mods, you can, for example, make this value go up as the RPM goes up.

-RPM Smoothing: It slows down the response time to the changes in RPM. You can try increasing this if the engine feels too wild or decreasing it for a more fast response, which could be useful on auto mode.

-Idle Fade: Use this to adjust the fade between the engine on idle and low revs.

-Crossfade: It controls the blending between different grains or audio slices making it more abrupt or smooth.

-Lim Threshold & Kick: The documentation doesn’t cover these two but I suppose they are related to an internal limiter.

Synth Section

It includes 5 oscillators with two different waveforms each that you can blend together. You can also control the frequency and gain of each of the oscillators. There is frequency and amplitude modulation available for each oscillator plus a vibrato option.

And that’s pretty much it. Sounds basic but it is indeed powerful as you are able to link any of these parameters to the master rev knob creating dynamic designs that will grow in intensity and speed as the revs go up. You can also combine synth layers with real engines to create hybrid engines that combine real recordings and synths.

Here are some examples of sci-fi designs I did from scratch. Something that I missed is more options for the noise generator. it would be great to have more noise colours to create textures with or maybe a filter to shape it. The ability to apply separate FX to different oscillators would also be amazing.

One Shot section

This tab allows you to trigger certain individual sounds on specific moments on the rev progression curve. Maybe the most obvious use for this section is to trigger tire skids when we go up on the revs or screeching breaking sounds when we go down. In any case, this section is great to add sweeteners and flavour to the design.

There are four slots where you can drag and drop sounds. Unlike the granular engine, you can use your own sounds here and drag and drop them from finder. Each slot can be monitored independently and there are individual knobs to control volume and pitch. Both of these can also be controlled with an envelope instead of a knob, which offers interesting possibilities.

On top of the sample area there are four “timelines” each of them corresponding to one of the slots. Here is where you can choose when do you want the samples to be triggered but the horizontal axis doesn’t represent time but rev progression. In other words, you get to decide where in the acceleration curve you want some samples to be triggered.

Directionality is also accounted for. You can trigger samples as the revs go up or as the revs go down depending of where the triangle is looking. You can also have a sample that will be triggered both ways (diamond shape) and stop currently playing samples on the slot (square shape).

In general, the system is clever and nice to use but I feel that you’d really need some playlist and randomisation controls to make it really powerful. My idea would be to basically turn each of the slots into something like an fmod event. This way, you could add a playlist of sounds and control how to cycle through them or randomly jump between them.

This will give you a much richer system, where you can use sets of skids, terrain or engine pop sounds to choose from each time the event is triggered. For this to work well, you should be able to choose how deterministic the system is, in case you need predictability. Being able to tweak or re-shuffle the samples that were triggered after a pass would be also a good approach. I know Krotos is working on a run-time, middleware version of Igniter, so maybe something like this is already in mind.

Loop Section

Although the one shot section includes an option to loop its samples, this tab gives us much more power and control of sounds that need to be looped. It can be used in conjunction with the granular system or just by itself to create a completely new vehicle system.

This is pretty powerful. It allows you to have your own responsive car design, provided that you have recordings of steady RPMs to use. You can also use the loop section it to add texture or detail to the granular generator. You can add things like gravel, dirt, snow, clattering, squeaking or engine pops and link their intensity to the master revs knob.

You have four slots for loops and you can control their volume and pitch. The interesting and very handy thing is the section on the upper side. It allows you to customise how you want to blend your four loops together giving you the tools to smooth out both the crossfades and pitch changes between the transitions.

To obtain a good result, you need to make sure you have audio clips that loop cleanly. The Amp section helps when determining the boundaries between the clips but I miss more control on the actual volume of each of the sounds when I need to balance them out. I’ve noticed that actually some of the factory presets use the mod section to control this by using the general gain of the whole looping section but this strikes me as bit left field. Shouldn’t I be able to control the gain of each sample with the amp section? A gain parameter independent of the crossfades is needed here, I think.

On the other hand, the Pitch section is very nice to have and it works well. It would be amazing to actually being able to analyse the pitch of each of the samples and get a “suggested pitch curve”. This could be just an starting point so you can then tweak them by ear later.

The workflow in the loop section is a bit odd since you can’t hear anything unless the main engine switch is on but then if you switch it on, the first loop triggers so you can’t hear what you want to hear in isolation unless you manually mute the first slot. It feels kind of odd. Additionally, when building the loop progression, sometimes a slot doesn’t emit sound and you need to manually hit its play button. Kind of annoying.

So here is an example where I’ve built a Peugeot 307 engine from library recordings. For sure, the result is not as smooth as the granular presets and it sounds a bit “processed”, it’s like you can hear the artificial pitch bending too much. There are also many dropouts in the audio level and I don’t know if this is my fault or if there is a way to remedy that. The factory presets that use the loop system are cleaner than this but I can still hear some dropouts on those so maybe this is a bug?

As for the sound in general, it depends on how you drive the RPM and I assume creating a robust and good sounding vehicle system takes more sample preparation and tinkering than my quick test took. I was also thinking that maybe I chose the incorrect range of RPM loops and I missed having more slots so I can use more RPM states and make the progression smoother.

Browser

It is used to choose and monitor samples for the granular, one shot and looping sections. The tagging system is very nice. Igniter includes a nice selection of different engines and sweeteners, many cars also include recordings of doors, horns or wipers ready to use. I’ve noticed that you can’t drag and drop these sounds from Igniter to Pro Tools, which will probably be my first instinct if I just want a car door sound on the DAW’s timeline. The alternative would be to have the sound on the One shot section and either trigger it via Pro Tools automation or via the timeline system.

Other than the factory sounds, you can also use the “Files” tab to browse around your own computer files, including external drives, which is very nice.

Something I’ve noticed and is a bit counter-intuitive, is that in order to preview a sound on the browser, the engine button needs to be on, maybe that’s the case because that button just mutes the whole plugin internally but it took me a minute to figure it out.

Mods

As I have mentioned before, this is a very powerful and important section of Igniter and probably the one that I liked the most. It reminds me of Propellerhead’s Reason where you can flip the rack and apply envelopes and LFOs to any parameter in the system.

Basically the mod section allows you to link any parameter within Igniter to the master revs knob. You just need to drag the name of the desired parameter and drop it on the mod area. Then, you can edit the envelope that will govern this behaviour and also use an LFO to add some randomness or movement to any of these relations. The range or scale of the change can be adjusted with the sliders that appear to the right of each parameter. There are 8 mod slots so you can create very different envelopes and very complex systems.

By default, the RPM within the Granular section is inked linearly to the master revs and from there, you can link all sorts of other stuff, including FX, to make the engine more dynamic and responsive. Have a look at the presets to get some ideas of what you can do with this, it really allows you to get creative.

I was also thinking that it would be very nice to be able to use the mod section on other things than the RPM. As an experiment, I tried to turn the master revs knob into a distance knob, decoupling it from the granular RPM and linking it in several ways to volume, reverb and EQ.

Why would I want to do this? Because controlling the distance and perspective between shots is probably one of the most time consuming things to do in a vehicle scene. My experiment kind of works although when you do this, you loose the ability to link other stuff to the vehicle RPM. So, for a really powerful, all in one, vehicle design tool, I would love to have 3 master parameters: Revs, Distance and a maybe a third custom one. This is maybe outside of the scope or workflow that Krotos had in mind but that is at least how I would try to design it. Of course, you can create a similar effect just on your DAW but using this method you are able to link many things at once to the “distance knob” like engine/exhaust mix, granular FX, reverb sends, etc, speeding up workflow massively.

FX & Mixer

This section is pretty straight forward, nice to use and clean. You can control the level for each of your audio generators plus you have an FX send and Pan pot. While the sends and FX are pre-fader, the Pan is post fader. Each section has a rack with 5 slots where you can hook up FX. The FX that we can use are:

-EQ: Very nice parametric EQ with everything you need. Works great.
-Compressor: Very good too, with a gain reduction meter and a limiter mode.
-Limiter: Simple and clean dedicated limiter, useful to make sure you don’t saturate the output at high RPMs.
-Saturation: Good for adding some extra nastiness to an engine with extensive controls and colour presets.
-Transient Shaper: An unusual addition to a plugin like this since engine sounds don’t have many transients but it could be cool to use to add or remove dynamics to the granular section or on sweeteners.
-Flanger: Nice for sci-fi designs.
-Noise Gate: I suppose it could be useful if you have a noisy recording on your one-shot section.
-Ring Mod: Pretty cool and alien sounding and a nice addition for creating sci-fi stuff.
-Convolution reverb: Very good to have to recreate distance or an “in-car” sound. The controls are quite simple but you probably don’t need much more. I miss more outdoors IR in the factory library.
-Doppler: Very nice if you need to quickly cover passbys. You can control it independently or attach it to the main Revs knob. Passby presets are already created for each vehicle which is very handy.

General Workflow

In terms of workflow, Igniter allows you to create the engine RPM movements in a very quick and flexible way and of course, you can always come back and tweak the automation to make it work better. Additional passes controlling other parameters (like load) can add extra realism and detail.

The loops are nice to have since you can, for example, make any car go on gravel or dirt, for example, with just adding a loop layer to the granular. The one shots are not that useful, in my opinion, since you can only have five individual sounds and you can’t assign probability or playlists to the triggers, so every time you pass through them on the RPM curve, you would hear the same exact sound. The way it works right now, I think you would be better of just editing sweeteners like skids manually on your DAW the old-fashioned way and use Igniter for the engine itself but I’m open to be wrong about this.

You would probably need two instances of Igniter, one for exterior shots and one for interiors, unless you want to do the interior treatment outside. Once you have the basic RPM behaviour down, you would then need to mix it into the scene with fader work, pan and distance attenuation. That’s why I was thinking that it would be cool to have a dedicated master distance knob so you can tweak this in one go once you find a reverb that works with the scene. With these system I’m imagining, you would do an RPM pass, a distance pass, some tweaks here and there and you would be done for that car. Rinse and repeat.

Lastly, it’s also important to mention that Igniter offers a multi output so you can get an individual signal from each layer and mix them in any way you want in your DAW. This is very much appreciated.

Is Full Tank worth it?

Krotos offers an expanded version called “Igniter Full Tank” which includes all the unprocessed and processed recordings used to build all the presets. You get a lot of coverage for every vehicle in Igniter plus loads foley and sweeteners. The recordings are a great library just by themselves (75 GB of additional audio) and in combination with Igniter will allow you to cover every single detail and sound you may need. To clarify, these extra sounds come as separate audio that you can then browse within Igniter, but they don’t include new presets or vehicles.

Conclusion

I hope both you and me now have a good understanding of how Igniter works and what it can offer. I had a lot of fun testing the plugin, Krotos keeps giving us innovative tools to create custom, unique soundscapes and I feel that with them we can offer much more value to our clients because the result is unique and personal.

Above all, the granular system sounds great and I know how hard is to make interactive engines sound good. I’m sure more content will come for the plugin in the future and maybe some workflow quirks will be fixed with time. As for the features I’ve been suggesting, they are just my own take on how I would improve the software’s workflow and capabilities and since I’m sure some concepts and perspectives have escaped me, I will remain open to new and better ways of using Igniter as it spreads across studios worldwide.

Thanks for reading!

Impressionistic Soundscapes

Claude_Monet,_Le_Grand_Canal.jpg

How your dreams look like

The fascinating thing about Impressionism is that it assumes that a painting is never going to able to recreate reality as accurately as a photograph. Once you leave behind the burden of precision, the artist is free to do what art does best: expressing a feeling, a mood, a state of mind. Impressionism relays more on movement and light than shape and form. The composition is open and the boundary between foreground and background is blurred.

An impressionistic painting doesn’t look like a real place but a distant memory, the impression a place leaves deep in your mind. It looks like the blurry pictures from a dream that linger in your mind just before you forget them.

That’s pretty much how far my artistic knowledge goes but I hope you get an idea. I was thinking that it would be cool to try to translate that approach into sound design by creating soundscapes to go along with some impressionist paintings. But before we do that, we can’t forget that, in a way, this already happened among a very specific sub-section of sound designers. The ones that limit themselves to a narrow amount of defined pitches and timbres: music composers.

How your dreams sound like

I was introduced to Impressionist music by the amazing series Young People's Concerts by Leonard Bernstein which I really can’t recommend enough. If this is the first time you hear about it, just go watch it. There is a whole show about Impressionism.

He does a better job than me in explaining it but, basically, when impressionism is translated into music we are trying to express the feeling, the essence of something in a subtle and seductive way. We are not explaining, we are suggesting. This often results in dreamy melodies (whole tone scales are a staple) and the use of exotic and unresolved harmonies. For the most part, composers limit themselves to traditional instruments but they try to get the most from them in terms of timbre. Piano is probably the instrument of choice for Impressionism, using its large range, dynamics and polyphony (pedals are heavily used).

As an example, here is what musically happened when Manual de Falla, who was born in Cádiz like myself, moved to Paris and met the Impressionists. Maybe is not its most known side, but sometimes flamenco has a dreamy, exotic quality that I think is perfect for this style of music.

And here is a maybe a more canonical example by Debussy. Notice how the melody is usually unresolved. Like in a dream, you don’t really know how you got there and there is no clear conclusion. This music maybe doesn’t sound that different or special to you, as these traits have been assimilated into mainstream music (think jazz) but keep in mind that in that time it was quite a contrast to the musical establishment.

An acoustic impression

If Impressionism doesn’t want to be constrained by shapes, colours or composition, maybe the most logical way to translate this idea into sound would be to forgo concepts like harmony, melody or rhythm. When you do this, only timbre is left and since shaping timbre is kind of my job, it sounds like a perfect fit.

My first approach to a Impressionistic soundscape is simple: just create an auditory complement to the visuals, extending the world within the painting to a new sense. Let’s lay down sounds that could exist in the scene and that go well with the feeling it transmits.

I’m using first the one that gave the style its name (and it was meant as an insult), “Impression, soleil levant” by Claude Monet:

Here is a second one, using “Woman in the Bath” by Edgar Degas:

At first, I thought I would use reverb to blurry sounds together in an analogy of how painters mix colours. But I soon discovered that doesn’t work very well. For the bath painting, I wanted to express a feel of intimacy, a sense of “costumbrismo” which actually was one of the other features of Impressionism: to portray everyday life.

Reverb doesn’t help with this because it creates an unnatural space that doesn’t complement the painting but opposes it. Monet’s Sunrise scene uses more reverb but only enough to match the environment that we are being presented.

One more thing was apparent: it helps to have elements in the scene that suggest motion, since most things that make a sound are moving in some way.

Here is “Effect of Snow on Petit Mountrouge” by Édouard Manet.

Since this painting was created during the franco-prusian war, I decided it could be cool to also tell a little story within the soundscape. I wanted to capture the peaceful calm of a winter snowy day somewhere in Paris. The calm is then broken when distant cannons are heard and the french soldier who is contemplating the scene has to go back to his post.

Finally, here is “Gare Saint Lazare” by Monet again:

I chose this one because I liked the painting from an aesthetics point of view, it has movement and life. And of course trains are a nice sound design opportunity.

Going further

After working on these four soundscapes, l realized I was mostly describing the scene and maybe transmitting some of its essence by choosing certain sounds but not being technically impressionistic. I was basically adding a soundtrack to the painting.

Their relaxing, atmospheric quality goes well with audio that borders on being ASMR. It’s somewhat ironic that the best complement to an impressionist painting is a soundscape that does the opposite: being descriptive, detailed and realistic. Maybe it makes sense in a way. These paintings suggest instead of being explicit so there is room for audio to add to the experience.

Of course this got me thinking about how it would be to create soundscapes for other art styles. Probably the ones that distort reality in different ways like expressionism or cubism could be good candidates. Maybe something worth exploring in the future.

But can we use audio in a way that gets to the core idea of Impressionism? To do this, we would need to go more experimental and abstract. We would need to stop using descriptive sound, forget about what you can see and focus on the feeling the painting creates.

Smearing sounds

I thought about using Paulstretch since if you play with the window size, you can blurry and smear sounds together, like painters mix colours. This worked nicely as Paulstretch tends to sound very dreamy. The following soundscape was created from only one audio sample, this recording of some wind chimes:

I created different layers in Paulstretch playing with the window size, pitch shifting and adding harmonics. I refrained form using any “real” audio. Here is “The Cliff at Étretat after the Storm” by Monet.

As you can hear I’m getting somewhere interesting. I tried to evoke a warm summer feeling although I’m sometimes dangerously close to the line between being dreamy and being unsettling. My first instinct to solve this was to use music tricks, like pitching layers a fifth away from each other but I didn’t want to relay on musicality too much.

Here is another darker example using a fantastic painting, “Winter, Midnight“ by Childe Hassam:

This one was created from a music stinger. If you hear both closely, you can tell it’s the same base sound but in a drone, dream-like state. It works well because the musical impacts are stretched creating some movement in the soundscape and some changes in tone.

And finally the last one turned out quite creepy, maybe too much for the painting but I like the result nevertheless. I used a combination of layers from Paulstretch, using the tonal / atonal slider to remove most of the “musicality” from the sounds (which were kind of musical). Here is “Moonlight, Isle of Shoals” by Childe Hassam.

If this got you interested in learning Paulstretch, I have a blog post about it that goes deep into how it works.

Conclusions

It’s cool to work with the concept of “pure sound design” without the burden of mere description but at times it seems to feel too close to atonal music. That last soundscape got me thinking about Ligeti and Penderecki. But is this something bad? Maybe is atonal music which is too close to “pure sound design”. Maybe they are the same thing but looked from different perspectives.

In any case, both approaches to the creation of a painting soundscape are valid and worth pursuing, I think. Just the idea of using visual art to inspire audio work is a good way to get your creative juices flowing and tackle things in a different way.

Other than that, I was also reminded that sound is not only simple description, it also conveys feelings and can somehow capture the very essence of a place, an action or a character. That’s something to always keep in mind.

Exploring Sound Design Tools: Sound Particles

Sound Particles allows you to create soundscapes and sound design using virtual particles that can be associated with audio files. The results are then rendered using virtual microphones.

If you want to check it out or follow this review along, you can download the demo here. It has all the features of the paid version but is limited for non-commercial projects only.

I won’t explain how to use the software in depth but I will give an over overview and show some practical uses for everyday work in sound design. If you want to get a more in-depth explanation, you can also watch this tutorial.

Sound Particles interface. Nice, clean and responsive.

Features Overview

The heart of the program are the particles. You can basically create them in three different ways:

  • A Particle Group will create any number of particles at the same time in an area or shape of your choice.

  • A Particle Emitter creates particles over time at a particular rate.

  • A single point source is just a single particle.

By default, particles are created as soon as you hit play, although you can also choose to change the start time to delay their creation. Generally, they last as much as the length of audio file attached to them.

You can choose the coordinates used to create your particles and also move the individual particles around the scene to create different effects. Particle emitters can also be moved. The movements that you can apply to the particles stack with each other, giving you an amazing amount of options to create motion. Keyframes can also be used to match any movement to a reference video.

See the video below for an example with the three types of particles:

So in the video you can see:

  • A particle group (red) that generates particles in a square shaped area. These particles are not created at the same time because we have also applied a random delay. They have fireworks sounds attached.

  • A particle emitter (orange) is moving in a circular motion while the particles that creates also have some small random movement. They have magical sounds attached.

  • A single point source (pink) with my voice paulstreched to infinity.

You can also apply audio modifiers to each particle group. These will randomize certain parameters so you obtain more interesting and varied results. If you think about this, this is similar to how audio works in the real world. Each time you take a step, your shoe makes a slightly different sound: pitch, level and timing will be different. Sound Particles lets you randomize the audio from each particle in a similar way. The audio modifiers are:

  • Gain: Basically, audio level.

  • Delay: This determines when the particle is created. It is very useful because usually you don’t want all particles in a group to be created at the start. In the example above, the red particles are being created with a random delay.

  • EQ: It applies different filters and bands of EQ to each particle so they don’t sound exactly the same.

  • Granular: This is kind of a special modifier. It slices the audio file and then plays each slice from a certain particle. You can control how long the slice is or even leave it random. You can also control if the slices are then played in sequence or at a random order.

  • Pitch: It applies a different pitch shifting value to each particle.

For any single parameter that requires randomization, you can choose different probability distributions to get the result that you want. An uniform distribution (all values have the same weight) and a normal distribution (most values will be around the mean) are probably the most useful ones. You can even create a custom distribution which is pretty awesome.

Uniform Distribution

Normal Distribution

Of course, once you have the particles ready, you need a virtual microphone to capture the result. On this area, the amount of options are simply amazing. Not only you can place the microphone anywhere in the scene but you can choose between many configurations including M/S, X/Y and all sorts of surround and ambisonic configurations.

If that wasn´t enough, you can also create several microphones on the same scene and render different stems per microphone. These stems can contain different combinations of particles so you can have more control later on the mix.

Finally, the project settings page allows you to control how Sound Particles is going to manage sound propagation and attenuation from distance. You can change the speed of sound, simulate the delay of far away sounds, change how much sounds attenuate with distance or wether your scene uses the doppler effect.

Microphone configurations can follow a variety of speaker setups

Project Settings

Sound design examples

Enough with the theory, let´s hear some real applications. Since sound particles is much easier to understand when you see the particles in movement, I decided to create a video for every example instead of just audio.

Battlefield soundscape

This is very simple but could be very useful if you need create a soundscape and don´t want to move every single sound into place by hand. As you can see, is very easy and quick to create a randomized soundscape. Something I feel I miss here is a bit more control on which sounds are triggered. When you have different types of sounds, it would be nice to be able to trigger some sounds only occasionally in the same way you can do this in fmod or wwise.

It would also be helpful to be able eliminate a particular particle that moves too close to the mic or at least be able to prevent them to getting too close without using complex custom distributions.

Scifi Interface

Now let’s imagine we are building a somewhat cheesy 80´s computer interface with beeps and blops and some folders flying around the screen.

As you can see, we are using two particle systems at the same time. One of them (blue) creates all the beeps in a circle around the listener while the orange is a particle emitter that throws particles horizontally to simulate things flying by.

Playing with pitch

Let’s explore how we can use the pitch randomization feature to create new, complex sounds from simple ones. On this example, I first use a uniform distribution for a more detuned and unsettling effect. We can also use a discrete distribution so the jumps in pitch are strictly within certain semitones, obtaining a more musical result.

As you can see, just changing the distribution can produce very different results.

We can also automate pitch to create dynamic effect like for example making all the frequencies converge on a central one. The THX deepnote was achieved with a a similar method.

Granular synthesis

This modifier offers many sound design possibilities. You can see an example below of building some sort of alien speech sound step by step.

We can also obtain a “voices in my head” effect by slicing up some speech and distributing it around the listener. As you can see, we can always re-create the particles to obtain a new variations which is very handy for video game work.

Doppler Effect

There are many plugins that recreate a doppler effect but this one for sure offers a unique visual approach. As you can see below, we can create a doppler effect on a single particle or on many.

Conclusion

I hope you found this software interesting, I think is a very good tool to have in your arsenal and I feel I have barely scratched the surface with the sonic possibilities that offers. I believe there is an update coming soon for Sound Particles and I may have another look then and write a new post covering the new features.

You can also have a look at a couple of plugins that Nuno Fonseca, Sound Particles creator has released. They allow you to use the doppler and air absorption simulations that Sound Particles has but in a convenient plugin that you can use in your DAW.

Exploring Sound Design Tools: Mammut

mammut.png

Mammut is a strange and unpredictable piece of software. It basically does a Fast Fourier Transform (FFT) of a sound file but unlike Paulstretch, which uses slices of the sound, Mammut uses the whole thing at once, creating more drastic results.

Mammut is not (in any way) a commercial tool but more of an experimentation one, so I won't go into detail about what is doing under the hood. Instead, I will focus on how it can be used to create interesting and cool sound design. If you want to follow along, you can download it here.

Software Features

Mammut has many processing tabs but I will only cover some of the most interesting ones.

Loading & Playing sounds.

Mammut works as a standalone software. You need to load a sound (using the browse button) to be able to start fooling around. The "Duration Doubling" section adds extra space (technically, FFT Zero Padding) after the sound. This extra space give some of the effects (like stretching) more time to develop and evolve.

Play and stop the sound on the Play section. There is also a timeline of sorts. Now that our sound is loaded, let's see what we can do with it.

Stretch

It creates a non-linear frequency stretching with frequency sweep effects. All frequencies are raised to the power of the selected exponent, so small changes are enough to produce very different results. Because of the frequency sweeps, it sounds quite sci-fi, like the classic star wars blaster sound. Here are some examples at different exponents:

As you can hear, as the values get further away from 1, the effect is more pronounced and it also starts sooner. Here are some results with values higher than 1:

And here are some interesting results with a servo motor sound.

This sounds remind me a bit of japanese anime or video games, maybe this could be one of the steps for achieving that kind of style from scratch.

Wobble

This stretches and contracts the frequency spectrum following a sinusoidal transfer function. You can control the frequency and amplitude of this change.

This one is weird (no surprise) and it doesn't really do what I was expecting. It tends to create sounds that are increasingly dissonant and "white noise like" as you go to more extreme parameters. Here are some examples:

Threshold

Quite cool. Removes all the frequencies below a certain intensity threshold. This means that you can kind of "extract" the fundamental timbre or resonance of a sound. Used on ambiences (3rd example below), it sounds dissonant at first and then, once you remove almost all frequencies, kind of dreamy and relaxing.

Block Swap

This one basically divides the frequency spectrum in chunks and interchanges their halves a given number of times. Hard to wrap your head around but it produces interesting results. First, the number of swaps seems to make the sound more "blurry" and abstract as you can hear:

Then, the block size seems to create different resonances around different frequencies as you increase it.

Mirror

Simple but hard to predict. It reflects the whole spectrum around the specified frequency. The problem with this is that when you flip the spectrum around a low frequency, everything ends up under it and is mostly lost. On the other hand, if you use a higher frequency, too much of the energy ends up on the harsh 5-15 KHz area.

A couple of examples:

Keep Peaks

Screen Shot 2018-06-29 at 15.50.33.png

This one doesn't even have controls or an explanation on the documentation. It seems to extend the core timbre of the sound across time which can be pretty useful. When using this option, the duration doubling function is specially handy.

Conclusions

Mammut is certainly original and unique. Since it only works standalone and is rather unpredictable and unstable, I don't feel it would be very easy to include in someone's workflow. Having said that, is definitely a nice wild card to have whenever you need something different.