Wwise, Unreal and Volumes

Wwise is an incredible audio system that pairs excellently with Epic’s Unreal Engine and is improved further when using Wwise Acoustics to add propagation and Wwise Reflect to add realtime reflections. However, both Acoustics and Reflect, as well as Wwise in general, really benefit from having well-crafted Volume actors in Unreal. These are regions drawn into maps that tell Wwise about the size and shape of the rooms. In Wwise, you’ll use these volumes to specify reverbs, play sounds and potentially other advanced use cases like specifying states and manipulating RTPCs. 

If using Wwise Acoustics, the volumes are also used to inform Wwise of the geometry of the room in a simplified way so that sounds can correctly propagate throughout spaces, allowing sounds in the next room to correctly funnel through portals on doorways and windows. If using Wwise Reflect, the volumes are also used to replicate the reflections bouncing off walls, creating a sense of immersion and further improving the propagation of the Acoustics system. So volumes are a critical part of game audio development, especially if you want high quality, immersive, and realistic soundscapes.

The most common Wwise volume used in Unreal is the AkSpatialAudioVolume, which can handle triggering of sounds, reverbs, propagation for Acoustics, and reflections for Reflect. There are a few other volume types for specific use cases, such as the newer AkReverbZone, plus you’ll also use AkAcousticPortals, which act as bridges and connections between the different volumes. 

Image 01 - A simple level with Wwise AkSpatialAudioVolues for each room and AkAcousticPortals that connect them together

Volumes must be accurate enough to carefully follow the shape of the accessible areas, without bleeding into adjacent rooms. This can be especially difficult in games with complex shaped rooms made of curved geometry and in maps where rooms fit tightly together like jigsaw pieces. A good example of the challenge are curved glass walls separating different rooms; glass is often very thin, requiring pinpoint accuracy over these curved shapes to ensure there’s no overlap into the next room - fiddly and timeconsuming to deal with.

The other challenge with volumes is maintenance. During development, level designers will improve the experience by adjusting the shape and arrangement of rooms. Later on, level artists will replace the simple geometry with full art assets, further changing the shape of the rooms. As release approaches and the experience takes shape, feedback must be acted on to, say, make traversing through the space faster, improve the visuals or change sizes to accommodate tweaked gameplay. All this requires the audio team to consistently stay on top of changes and updating volume shapes; not only relying on good communication but also being able to quickly and safely adjust the volumes… and of course this maintenance work most often happens at the busiest time for audio, increasing the chances of shipping a title with issues.

Accuracy vs CPU cost

Accuracy is important; the volume must completely fill the space. You also don’t want a volume from next door triggering because it overlaps into your room. In the case of Wwise Acoustics, this can be even more undesirable if it’s a 3rd person game, as the player would enter the other volume but the listener (likely on the camera) is not, causing transmission loss and a large amount of sounds totally cutting out… nightmare scenario! But even ignoring that worst case, getting volumes wrong is going to cause all manner of bugs.

So accuracy is important, but on the flip side, you don’t want to build volumes with too much complexity. Volumes are made up of triangles (even if it looks like they’re made of polygons in Unreal), and the triangle count determines the CPU cost; the more triangles, the more complex, the higher the CPU. We simply cannot use the number of triangles required to exactly represent the shape, so some form of simplification is absolutely necessary to rein in CPU usage. Also it’s worth noting that Wwise Reflect is negatively impacted when volumes are too detailed.

All this means that audio teams have the most demanding of requirements for their volumes. Every space where a sound can exist needs to be encapsulated by the appropriate volume. Volumes usually shouldn’t overlap and need to be carefully portalled and need to accurately reflect the shape, even if concave, curved or complex. Finally, they need to do all of that without using too many triangles, and they need to be easily maintained when changes occur. Phew, quite the list. 

Traditional Methods of building volumes

There are 4 traditional ways to build volumes in Unreal for Wwise:

1. 1. Unreal’s ‘scale’ transform controls (i.e. the main actor widget)
      • Easy and fast to use 
      • Can only build primitive shapes such as cuboids, which makes it unsuitable for most audio volumes
   2. Unreal’s brush editing tools
      • Easy to build simple shapes
      • Requires a lot of extrusion to build non-cuboids, so incredibly fiddly and time consuming to make anything suitable
      • Being deprecated by Epic, replaced by the modelling tools
   3. Unreal’s modelling tools
      • Very powerful featureset with interactive tools
      • Designed more for static mesh manipulation rather than volumes, so tools are geared towards visual accuracy rather than our specific needs (low CPU and accuracy)
      • Very high learning curve for audio people with many irrelevant features
      • Fiddly to use in a fully art’ed map as visibility becomes a huge problem
   4. The Wwise Unreal integration’s ‘Fit To Geometry’ feature
      • Very easy to use as it detects negative space
      • Positioning impacts detection making it inconsistent
      • Can only build convex shapes making it inappropriate for more complex areas

In my experience having worked on about 10 Unreal titles, the modelling tools are the best approach, although the ramp up time is high and manipulating existing volumes for maintenance can be slower than just rebuilding from scratch if the shape of the space has drastically changed. 

Even with staff skilled with the modelling tools, it’s not unusual for a AAA project to have a year’s total time allocated simply to building and maintaining volumes (later on we’ll talk through an example). In today’s development environment, that’s a year too long and frankly is a poor use of skilled audio staff’s time; none of us got into this industry to build boxes and we have more important things to do! 

There has to be a better way to build volumes, and not only that, there’s also an opportunity to be able to test, report and maintain volumes that’s not realised with the current tools. This was my perspective when first looking at this problem over three years ago. Since then I’ve built multiple iterations of tools that can aid this process, ideally getting that year’s work down to as low as possible. At this point, the tools have coalesced down to a single plugin that is not only highly matured and battle-tested, but brings a lot of additional features I never dreamt of when I started this journey.

Volumator: The Procedural Volume System

Volumator is an Unreal plugin that can be integrated into any C++ project using UE 5.5 and 5.6, with 5.7 coming very soon. It is designed to build, test, maintain and automate volumes, and while it can build any volume class and aid all departments, the core goal is to help audio teams build volumes for Wwise and there are many features with this workflow in mind.

The integration process is incredibly easy and the only requirement is Epic’s included Geometry Script plugin, which simply needs enabling if it’s not already. Once integrated, it’s controlled by a single UI widget and a setup wizard runs on the first opening to help get you started. The widget is separated into tabbed panels for different workflows, e.g. Building, Testing, Automating, so you only see tools relevant to your current task.

Image 02 - The Volumator widget showing the Build panel. Switch panels with the tab buttons up top.

Volumator can build volumes extremely quickly, either by physically flying around the area in the editor, or by sending out drones to do the flying for you. Once scanned, the volume is built and can be tweaked with a host of settings, giving you total flexibility over complexity, accuracy, size and shape. It also hosts a number of testing features to check the quality and integrity of your volumes within the space. It can even test to see if the space has changed, requiring adjustments or rebuilds, which can also be done without losing any work on the volumes themselves, so no more lost settings/variables when volumes have to change. 

If you want to go further, you can even fully automate the building, testing and maintenance of volumes across entire levels at the push of a button. Plus there are tons of productivity tools such as reports, selection aids and visibility tools that will make your volume life not only considerably easier but (hopefully) a lot more fun too. All this is done with editor-only functionality; nothing gets packaged into your game, so it’s simply all positives for you the developer.

At this point, it’s probably best to see it in action to get a better idea, so here’s the original v1.0 launch trailer and here’s a deeper overview.

How Does It Build Volumes?

At a high level, there are two distinct phases of volume creation: detection and generation. 

The detection phase involves you physically flying the editor camera around (piloting) to map out the space. Alternatively,  you can send out a drone to do the flying for you, automating the process and carrying it out considerably faster. The time it takes to complete detection will depend on the complexity and size of the space, ranging for a couple of seconds for a small and simple area to maybe 30 seconds for a massive and complex one. Once detected we use that data in the generation phase, but the detection data is also stored into an editor-only component that’s placed on the volume for reuse later, allowing us to regenerate the volume with tweaked settings. 

The generation phase takes the detected data and processes it with a ton of maths to create a volume. This is where the real magic happens and there are a lot of settings that can influence the shape and structure of the volume. Later, if we don’t like the created shape, we can simply change the settings and regenerate the volume with a new shape. 

So detection is carried out once, while generation can be carried out repeatedly with the same data. If the level changes, we can also redetect the space again and update the volume with new detection data to match the changes. 

Any updates to the volumes (regenerations) happen on the same volume actor… so you never lose variables on the volume. If you’ve added AkEvents or reverb sends on, say, AkSpatialAudioVolumes then these will remain intact when regenerating, only the shape of the volume is changed.

Comparison With Other Volume Building Methods

I tried my best to build a volume for the below U-shaped room in the shortest amount of time for the different methods to build a volume. I picked a space that was very simple with straight walls and simplified art which would favour manual building. 

Image 03 - Top down view of the space we’ll be making a volume for

Scale Transform Widget:  124 seconds

As the best primitive to use here is a cuboid and we can only scale it, we are restricted to a simple box, but it still took a long time as the scaling is uniform so you can’t, say, scale only the left side, it always does both. This requires a lot of back and forth with the camera to check both sides when scaling. The result is an inaccurate box and an unpleasant time creating it. 

Image 04 - Volume created with scale transform

Brush Editing: 130 seconds

Using the Brush Editing tools required a lot of extrusions to create the U shape, which is both fiddly and limiting. The result is blocky and inaccurate, but at least it’s better than the scaled box as I could manipulate each face. 

Image 05 - Volume created with Brush Editing mode

Modelling mode: 126 seconds 

A further step up in accuracy, but no improvement in time. The modelling mode is very powerful, but even when very familiar with the tools it can be unintuitive and fiddly to use. Manipulating edges, vertices and faces is incredibly awkward due to art and geometry interfering with selections and snapping. This causes it to be a slow and unpleasant method for volume editing… but at least the output can be good, if you’re willing to put the time in.

Image 06 - Volume created with Modelling Mode tools

Audiokinetic’s Fit To Geometry tool: 0 seconds (instantly)

Unfortunately this tool couldn’t create an appropriate volume to cover the shape. As the Fit To Geometry system fires rays out only from the origin of the volume, they are unable to go around walls, so in the case of the U shape, there was no way for them to correctly capture the whole space. Therefore the volume is a simple cuboid that doesn’t reach the extremities of the shape. It is at least instant, although usually you have to drag the volume around a little to find the optimal detection location.

Image 07 - Volume created with Audiokinetic’s Fit To Geometry feature

Volumator - Pilot: 8 seconds

In piloting mode, we simply fly around the space at least once. This took 8 seconds, including the processing time (Volumator has a handy stopwatch that’ll tell you how long it took). As you can see, the volume is both accurate and hugs the walls better than any previous attempt, and we could improve this with further tweaking of settings or editing points if we choose.

Image 08 - Volume created with Volumator, by piloting

Volumator - Drone: 3 seconds

As it’s a small enclosed space, it was perfect for a drone. This was able to create the volume in an amazing 3 seconds and likely had even better detection data than the pilot version. Due to the more thorough data generated by drones, any tweaking of settings will likely result in even better outcomes if we want to get in and tweak the settings further. 

Image 09 - Volume created with Volumator drone

Volumator - Drone - Multi-Box mode: 10 seconds

This multi-box generation mode shows how easily Volumator can make simple shapes if low triangle count is your priority. The drone took 3 seconds and regenerated a few times with tweaked settings taking it to a total of 10 seconds. This is probably the best volume for the space and yet uses a miniscule number of triangles, only 30.

Image 10 - Volume created with Volumator drone, set to Multi-box mode

Volumator - Drone - Voxel mode: 20 seconds

If we really want to capture that cutout in the U shape, voxel mode is a better choice, although it requires thorough detection data, so we use drone mode with an extended scan, this takes longer to detect. I also had to regenerate a few times with tweaked settings to get the cutout correctly, which took the total time to 20 seconds. 

Image 11 - Volume created with Volumator drone, using Voxel mode

With these examples, the time benefits of Volumator should be pretty clear. Building by hand takes considerably longer and results in worse outcomes, despite this test space being pretty much a best case example for manual building. More complicated environments significantly complicate manual volume building, resulting in considerably longer times and lower quality results by hand. Volumator doesn’t suffer from these problems and typically just requires more piloting or drone time. 

Perfecting Volumes

There are tons of settings to help configure the shape and complexity of the volume. They all serve various purposes and have thorough documentation both online and in the tooltips, but here are some of the most important settings:

• Generation mode - selects the algorithm used to build the volume; has a large impact on the shape/structure and also how the other settings interact. There are five modes ranging from the general purpose Perimeter mode, to an ultra-simplified shape mode, to a voxel-based mode for blocky corridor spaces
• Resolution - the main complexity control, this sets the definition of the volume, in Unreal units, so higher values have lower complexity
• Expand - inflate or shrink the volume; will push walls outwards in the direction of the normal of the walls
• Simplify Concavities - one of the simplification systems that will remove edges that cause small concave sections that are likely the least significant to the shape

Image 12 - A comparison between Resolution settings on a highly curved and concave room. On the left set to 200 and on the right 600. You can see the lower detail and accuracy on the right, along with clearly a lower number of volume faces.

Image 13 - Generation and Advanced settings

Thanks to the regeneration system, you can simply adjust a setting, press regenerate and your volume gets an updated shape. This makes it incredibly easy to tune volumes to your needs and each volume can be regenerated individually with whatever settings suit them.

The settings used during generation are stored on the volume, within the same component used to store the detection data. This means those settings can be later recalled, so you can see what was used to create them at a later date, and then continue tweaking or apply them to another volume.

If you want to make things even easier, head to the Tweak tab. Here you’ll find simple nudge buttons for each of the most important settings. These will add or subtract to the specific settings currently stored on the volume, allowing you to quickly modify values on volumes with a simple button press.

Image 14 - The Tweak panel, pressing any of the small buttons will add or subtract from the volume’s current setting and regenerate

You can even manually modify the volume using Edit Vertices mode, which converts the volume shape into a simple spline that can be manipulated by hand if you need absolute control. 

Image 15 - ‘Edit Vertices’ mode instantly converts the volume into a spline that can be manipulated and regenerated, for super fast manual editing

All of these generation methods allow you to easily manipulate the CPU and accuracy of your volumes, but it’ll also be really handy when using Wwise Reflect, where you’ll want the faces of your volume walls to resemble the reflective surfaces. 

Testing and Reporting

Once you’ve built a Wwise volume in Unreal, there are usually three testing concerns:

• Does the volume successfully cover the area? 
  • Bugs can occur if sounds or the listener go out of the volume while still being in the room
• Does the volume overlap with another volume?
  • If next door’s volume bleeds into this area then other serious audio bugs can occur
• Over time, has the level geometry changed?
  • This is especially relevant later in development, potentially breaking previously good volumes

Image 16 - An example of two rooms, each with a volume. The volume on the left doesn’t correctly hug the curve of the room, causing it to bleed into the room on the right, which would likely result in a nasty audio bug if walking around in the right room.

The usual way to check for these issues is to inspect the volumes or play through the game. Both aren’t desirable on very large or complex games under tight deadlines when audio teams are stretched the most. It’s also an inexact approach to a technical problem and issues are bound to fall through the cracks.

Volumator allows you to not only test for each of these issues but can even automatically fix problems. After testing, any issues are displayed in the map as editor sprites, the status bar gives you a text overview and you can even run detailed reports spanning entire maps. Volumator has two main testing tools: testing with piloting/drone and quick tests. 

Testing With Piloting/Drone

In the Test panel we can pilot the camera around, just like we do when building, to test the area for the created Wwise volumes. It works the same way as when building, but instead of detecting geometry, it is testing for areas that volumes aren’t covering. You can test for any volume of the class, or for specific volumes. Also the same as building, we can fire out a drone to do the flying for us, which can save a lot of time and the drones can be very thorough.

This testing method can also optionally test for locations that are covered by multiple Wwise volumes of the same class, so can clearly show locations in the map where, say, two AkSpatialAudioVolumes overlap. This can help you identify where volumes from next door are bleeding over into the current room. 

Image 17 - After testing, these ‘multi-volume’ sprites show where the two volumes overlap

Quick Tests

Quick tests are carried out instantaneously and can be triggered at any time with the Quick Test button. They’re also automatically triggered after generating or regenerating a volume, so you get instant feedback after a volume has been built or changed. 

Unlike testing with piloting or drone (which detects new testing data), Quick Tests use the original detection data on the volume as its test data for coverage and integrity, this is why they can run instantly.

Image 18 - The results of a quick test, this shows warning sprites at locations where the volume doesn’t quite match the curve of the walls

As it uses historical detection data, quick tests are not great for checking the volume against current geometry. However, they do have the unique ability to check if the original detection locations are still colliding as they were previously, which can imply level changes; these are called Geometry Change warnings.

Image 19 - After this volume was created, the wall was moved to the right, making the volume no longer fit. This caused geometry change warnings that highlight the volume no longer matches the level

Quick Tests will also confirm in the status bar if the location of the volume actor has changed, and a simple overlap test informs you if there are any overlaps with other volumes of the same class.

Image 20 - The quick test status bar readout

Test results show four issue types:

• Warnings - Locations on geometry (collisions) that are not covered by the volume, often not a huge issue as they’re colliding with something but can be a sign of a problem.
• Errors - Locations in open space (not colliding with anything) that are not covered by the volume. These are serious and should be investigated.
• Multi-volume - (only on pilot/drone tests) Optionally can be tested for and can tell you of areas with overlapping volumes, if that’s important to you.
• Geometry Changes - (only on quick tests) Locations that were collisions during detection but are no longer, implying a change in the map since detection took place and are worth investigating.

Image 21 - The different image types in order: warning, error, multi-volume, geometry change. These will appear in the editor at the location detected.

Image 22 - A volume with issues: 19 warnings (detected on geometry) and 2 errors (detected in open space)

So with the above, we can get rapid and thorough testing of our Wwise volumes, as well as information on whether the level content may have changed. This can be vital during late stages of development and help you avoid nasty audio bugs.

We can use this test data to visually guide our volume fixes, and if using the automation system (discussed later) we can even have Volumator automagically fix things for you. 

More Advanced Features

I won't go into all the additional features, but here’s a quick overview of just some of the additional things you can do to volumes:

• Validate the volume; the easy to use Validation system uses quick test results to instantly and iteratively build tweaked volumes aiming to reduce issues, it then picks the best one to output, saving you having to manually fiddle with settings
• Combine multiple volumes together, to create complex compound volumes, these can be separated back out later
• Subtract one volume shape from another; can be helpful to carve out areas, especially in cases of volume overlaps
• Run detailed reports telling you all about volume’s CPU cost, size, shape and all the Volumator data under the hood
• Reseed volumes; scrambles the detection data to give you other shape options
• Change volume colors and opacity for custom visibility and easier visual coherency of content

Image 23 - A report showing volume details (info on what each column represents can be shown by hovering above a column’s title)

Automation

The automation system is designed to further reduce manual work and is based on ‘depot’ actors. These are placed into the center of each area, one for each distinct space you need a volume for. When run, each depot will spawn a builder drone from its location and build the volume. You can run all depots in an entire map with a single button press, so you simply need to drop a depot into each room and press ‘start all depots’ for an entire map’s worth of volumes to be created. 

Image 24 - A depot actor; after building a Wwise volume, the actor lists its associated volume name and the results of the post-build quick test

Once a depot has created its volume, if the depot is run again it’ll spawn a tester drone to test that volume automatically. Not only that, but if it detects enough issues to breach a threshold, it’ll automatically rebuild the shape for the volume (without losing any of your audio settings on the volume of course!). So you just need to periodically run the depots on a map, and Volumator will automatically fix any volumes with issues caused by level changes. 

Depots are automatically assigned bookmarks, allowing you to use them to quickly teleport to each area at the push of a button. You can even set depots to be in groups, which will automatically build combined/compound volumes for the group, for cases where different ‘floors’ have different shapes but the whole area needs to be a single volume. There are also some other handy features like depot reports and the ability to lock depots to ensure nothing gets auto-rebuilt as you approach milestones. 

Barriers

Drones are great for speeding up area detection and the automation system depends on them. But drones don’t understand context the way you do. They can’t tell whether an open doorway is meant to be part of the space or the edge of it. Open transitions such as doorways, windows or ceiling gaps can confuse them, causing them to fly out of bounds and produce bad data. To prevent this we need to block those transitions with Volumator-only barrier geometry. This barrier geometry keeps drones inside the intended space and also stops detection rays from bleeding into adjacent rooms.

Barriers are simple volumes (3d) or planes (2d) that are placed usually on open doorways and windows at transition points. They are designed to be exceptionally quick and easy to implement as their job is simply to block Volumator detection. Barriers are just one-click operations: simply fly to the middle of the doorway, click the ‘Create Barrier’ button and that’s it. Volumator will size and angle the barrier to match the space. 

Volumator comes with its own barrier volume and plane classes, but you can assign any volume class to act as a barrier instead.  When you think about where these areas are and how you build your Wwise volumes, the chances are they are in exactly the same place as where you place AkAcousticPortals. Portals are used to create transition zones between AkSpatialAudioVolumes, so why not use AkAcousticPortals as the fake geometry for Volumator? 

So when using Volumator, you can set the ‘barrier volume class’ to AkAcousticPortal and now any portals you’ve placed in the map will automatically block Volumator drones and detection. An added bonus of doing this is that Volumator will now build AkAcousticPortals when you click create barrier, which you’ll find is faster and easier than manually building them. 

Image 25 - An AkAcousticPortal created by Volumator to also be used as a barrier in a doorway

What about open areas though? How do we ensure drones and detection are restricted to, say, certain areas of an open outdoor space? How do we stop drones from flying off forever in the wrong direction? This is where barrier fences come in. They’re simply chains of barriers and are, as always, very easy to add: just fly around adding a point for each link in the chain to quickly block off the perimeter. Now drones will bounce off the fence and your volume will only extend out this far. 

Image 26 - A barrier fence (in pink) made up of 5 points which took about 10 seconds to create. A drone started from inside the space will be restrained to that area by the fence. 

There are a few features specifically designed for Wwise related to barrier handling, how they impact volume creation and how portals are handled by Volumator. Let’s look at them below:

Barrier Volumes As Planes

During detection this will treat portals as 2d planes at the central mid-line in the middle of the barrier; portals show this line in their visualisation as a coloured slice through the middle of the portal. 

Usually your AkAcousticPortals are quite long in depth to ensure smooth transitions between rooms. By default, Volumator will use the barriers/portals to block any detection and so these long portals that stretch deep into rooms can result in some issues. So instead, we can enable this setting and use only that central mid-line to block detection, which will result in a cleaner ‘cut’ at the center of the portal and likely a better volume. It will depend how you shape your portals though, but it’s easy to try it out.

Image 27 - Barrier volumes as planes disabled, note the volume juts inward where the AkAcousticPortal is

Image 28 - Barrier volumes as planes enabled, note the volume now reaches the center of the AkAcousticPortal 

No Edges On Barriers

When enabled, this will enforce a strict ‘no volume edges on barriers’ rule, so you never get any volume edges or vertices on your AkAcousticPortals. I’ve had some user reports stating edges on portals can be problematic so this setting will ensure these are not allowed. It effectively discards any detected locations that were caused by portals/barriers, making them not contribute to the space at all. 

This setting also plays well with the previously mentioned ‘barrier volumes as planes’ so you can really tailor how portals impact volumes. Unlike that setting though, this one is a generation setting, meaning you can toggle it and regenerate the volume to see the impact it has. If you don’t like it, just toggle and regenerate again. 

Reorienting Barriers

Anyone that’s used AkAcousticPortals will know that the orientation of barriers is critical; they must be facing the correct direction: pointing towards the open area. What happens if the shape of the volume is perfect but it’s oriented in the wrong direction? We have to throw away and start again, as just simply rotating it will cause the shape to also be rotated. 

Volumator can come to the rescue here with its handy Reorient Barrier feature; just a simple click on this button will retain the shape as you see it but rotate the internal orientation of the actor by 90 degrees. This feature was specifically added to aid people with Wwise portals and can alleviate some pain points with portal handling.

Image 29 - On the left, an AkAcousticPortal fits the doorway perfectly but is oriented incorrectly (it’s pointing to the sides). In the center, if we simply rotate the volume, it’s now oriented correctly but is shaped incorrectly and no longer fits the doorway. On the right, after pressing Volumator’s Reorient Barrier button we can see that the portal has been rotated to orient correctly while the shape is retained fitting the doorway

Typical Workflow

Let’s say we have a level we want to populate with AkSpatialAudioVolumes and we’ve decided to use the automate/depot tools in Volumator so we can later test and rebuild volumes automatically. Here’s how we’d go about adding volumes for the whole level:

1. In settings, set volume class to AkSpatialAudioVolume, barrier class to AkAcousticPortal and tailor any other settings to your preference (e.g. enable Validation system). We can save these settings as the default or as a specific preset, so we only need to do this once
2. Fly the camera to each doorway, window or transition area and press ‘Create Barrier Volume’ to drop a portal there
3. If there are any infinite open areas that need cordoning off, add barrier fences to constrain the area
4. Fly to the center of each area to receive a volume and press ‘Create Depot’ to drop a depot into each area
5. Press ‘Start All Depots’ to build a volume for each room
6. After it’s finished, press ‘report’ to see detailed statistics on all the volumes in the map, this will include the results of the quick tests

With this process you can effectively build an entire map’s worth of volumes in a few minutes and have the ability to test and rebuild them easily afterwards by simply pressing ‘Start All Depots’ again.

Changing Wwise Volumes

There are now a few types of Wwise volume that can be placed in Unreal, such as the newer AkReverbZone volume. Each of these have different functionality and purposes. You may also have your own volume classes that are extended versions of the default Wwise ones.

If later in development you decide to change the volume class, up to now you’ve had to build a whole new volume. But in Volumator, you can simply click ‘Change Volume Class’ and a new volume will be created of that class with the same shape as the old volume.

The old volume is kept around, so you can easily copy over any data to the new one before deleting it. This makes porting volumes incredibly easy and makes volume class choice at the time of creation less of an important one.

Also, you can even convert an existing non-Volumator volume to a Volumator one. Simply take your existing Wwise volume, select ‘Rebuild Selected’ and detect the area. It’ll update your old volume with a new Volumator created shape while retaining all your settings (AkEvents, etc), but then allow you to regenerate, test, run reports, attach to depots, etc in the same way you can with any newly created Volumator volume.

Will It Work With My Content?

This is understandably the first question that comes up and the short answer is almost definitely yes. The majority of spaces are handled straight out of the box and even the most difficult environments can be tackled with a little tweaking of settings. These settings can be stored and recalled using the preset system too, so you don’t need to keep configuring it. 

We’ve already covered tweaking complexity to your needs, so it’s more a question of if your content is challenging enough to be problematic for Volumator. The absolute worst scenario for building volumes is highly complicated, curved and concave shapes with razor thin walls and adjacent rooms that need to remain exclusive. I originally developed Volumator for a project that had exactly these environments; it really was a perfect storm of challenges for volume creation and Volumator is considerably more powerful now than when I developed it for that project. I’m yet to find a space that Volumator can’t build effectively, it’s just a question of finding the right settings. 

Collision Types and Geometry

The usual gotcha is handling how collision in your content is set up; this is what Volumator detects. You can set Volumator to detect either simple or complex detection; simple is the default and will result in cleaner volumes provided your content’s simple collision is good. In cases of imperfect simple collision, you can just switch to ‘complex’ and it’ll create a more ‘what you see is what you get’ result.

You can also adjust collision filtering which can further tailor handling of things like glass or cosmetic objects like cables. Here you can select camera, visibility or objects. The latter of which uses a more complex multi-trace system that carries multiple checks and is more thorough, this can be useful for more stubborn content.

Again, you can store any settings as presets, so you typically only need to find the right settings once per project.

If your content has issues and drones are flying out of rooms for unknown reasons, you can even enable ‘Drone Debug’ mode, which will give you a POV from the drone so you can see where it escapes. When pressing stop, it’ll then show you the drone’s flight path as a spline, allowing you to easily trace back to where it escaped, and the offending piece of geometry which likely has a collision bug. In this way, Volumator can be a handy QA tool for art too.

In the unlikely event that even all this isn’t enough and your content is full of holes or imperfections, you can change detection rays into ‘bars’; think of them like fat rays that can’t fit through small gaps. You can even make the drone itself bigger too, so it can only fit through larger areas. These more extreme solutions will solve cases with really horrible collision or weird edge cases, so even bad placeholder content can be handled with a little configuration. 

Lastly, it’s worth noting that while Volumator contains some C++ functions, it is primarily created in Unreal Blueprint. This makes it easy to modify to suit your needs and of course I’d be happy to support you if you need any guidance through its many systems.

Verticality

One specific challenge that is worth calling out is highly angled verticality in walls. Volumator was designed primarily to build shapes with vertical walls, which is by far the vast majority of game content. In cases where you need vertical variety you can of course combine volumes to create complex compound shapes, such as different shapes per-floor. The automation/depot systems can also use groups to build combined volumes too.

However, at the time of writing Volumator cannot build angled vertical walls out of the box. So in cases where this is critical to your content, it’s best to ignore the verticality when building in Volumator and then use the Unreal Modelling tools to simply tweak the vertical sections; this is still considerably easier than trying to build the volume manually and only takes a few moments in rare cases where you need it.

An Example Case

Let’s talk through an example case to determine the benefits of using Volumator. Imagine we’re making a AAA 3rd person Unreal game using Wwise. We want the audio to be high quality and have a great sense of immersion and realism, so we’re using both Wwise Acoustics and Reflect. 

Our game has a play time of 30 hours to complete the linear story mode and there are unique maps for each mission, let’s say there are 30 maps in total. Given there’s an hour of gameplay per map, we can estimate there are 30 volumes per map. That’s a total of 900 Wwise volumes across the whole game, plus around 300 AkAcousticPortals to connect them.

The maps are made up of typical real world urban environments and a few outdoor areas. So indoor spaces are not blocky, unrealistic corridors, but more complex, with a smattering of curves and lots of details. Walls separating rooms are real-world thickness and proper sound occlusion is important. Outdoor spaces are small, not open world, and the player is limited to certain areas so they can’t walk off forever. 

This is a fairly average configuration of content for a game of this scope and quality and we have to build all 900 of these Wwise volumes by hand. As the rooms are much more complicated than simple boxes, and the walls between them are quite thin, our volumes will need to tightly and accurately fit each space; inaccurate volumes bleeding into adjacent rooms will cause bugs. 

Building a volume manually to the sort of accuracy we need takes time, maybe half an hour for complicated spaces. We also need to factor in maintenance, which often involves completely rebuilding volumes, sometimes multiple times as geometry and art changes. So when we include maintenance and bug checking for the duration of development, let’s estimate it at one hour per volume. 

With our 900 volumes, that’s 900 hours of volume work. Let’s also add in 100 hours for those 300 portals too as that’ll involve testing of portal transitions between rooms. 

So in total, we can loosely estimate Wwise volume work at around 1000 hours. 

We’re at a typical AAA studio with average efficiency, so have to contend with the usual meetings, production overhead and general company stuff, so only get 5 solid hours of production time per day (a common average for these estimates). Therefore, 1000 hours of work will equate to 200 full workdays, just shy of the average 230 workdays in a year. 

All that is to say, that Wwise volume work on our typical AAA project will cost nearly a year of dev time. I’ve been Audio Lead/Director on many Unreal/Wwise projects and that actually aligns with my recent estimates too for this scale/configuration of project; in fact on the project that spawned my desire to build Volumator I had loosely estimated a year’s volume work, although the content was configured differently to what we have here.

So let’s compare that to working with Volumator; building a shippable volume will likely be done in under a couple of minutes. If we’re using the depot automation system, we can also automatically rebuild the volume if changes occur, often reducing maintenance down to mere seconds… but let’s be conservative and apply a few minutes on average per volume to account for the more complex cases. So in total, let’s say we need ten minutes per volume on average for building and maintenance.

Now our 900 volumes only take 150 hours. We still need to build those portals, but in Volumator it’s simply a question of flying to the doorway and pressing ‘build’, so again takes seconds. Let’s again be generous and add in another 20 hours for portals, taking our grand total to 170 hours, or 34 days. 

So doing this work manually takes 200 days, versus 34 days with Volumator, making it a bit of a no brainer. My original goal was to reduce it to 10% and in some cases that’s no doubt true, but I’ll still happily take these numbers. 

But not only have we saved over 80% of the time, we’ve also got a whole bunch of other improvements; better quality volumes that’ll result in better audio from Wwise, reports detailing CPU cost and any other things you might care about, automated bug testing and pain-free fixing, class changing, lots of productivity tools (I haven’t even covered here), multi-volume overlap awareness, notification of geometry changes, etc, etc. So you save a huge amount of time and get a bunch of cool superpowers you didn’t have before. From the many messages I’ve received from developers using Volumator, these extra features really make a huge difference to busy audio teams.

How to find out more

If you’ve still reading this far, you’re probably getting a bit tired of hearing about volumes. Frankly so am I; volumes are a chore and we should be spending our valuable time doing the fun, creative and impactful work that we love. That’s why I built Volumator, so you can focus on the more important things. Hopefully you find it useful and can continue making awesome audio using Wwise!

Below are some links to more info about Volumator:

• You can find Volumator on the Epic Fab store
• Read thorough documentation on the website and 
• Watch tutorial videos on youtube
• Also feel free to email me if you have any questions or just want to chat: Volumator@PrecisionSoundDesign.com