SKYLORE

Skylore is a new mobile MMO RPG featuring both PvP and PvE content, a talent system, vibrant graphics, and a living open world.

The game world combines elements of fantasy, steampunk, and even futuristic details in places. All of this must be taken into account when creating the game’s soundscape. It was crucial to establish distinctive characteristics for magical elements, steam-powered machines, complex technological devices, and the diverse races of characters: humans, Gorgorush, Muar, spirits, elementals, robots, tree-like beings, etc.

Our team handles full sound production for games, including music and gameplay sound design, cutscenes, cinematics, and promotional videos. It was essential for us to have complete control over the in-game sound to free the developers from making constant changes. Skylore was no exception.

In the early stages, it was necessary to choose the right audio engine for implementing the game’s sound in order to solve several tasks at once:

• Complete control over the sound by the AK Audio team (sound systems, content, mix, optimization);
• Reducing the workload for AI Grind programmers on sound integration;
• Increasing the ease, speed, and efficiency of work for all parties involved.

As a result, FMOD by Firelight Technologies was chosen as the main tool to realize the developers’ vision for the game’s sound.

When creating the sound systems and content, our team always consulted with the art team and programmers about the feasibility of implementing certain ideas as well as their alignment with the overall vision and lore of the game. This allowed us to obtain useful constraints that helped steer the project toward the desired result.

Skylore is an extremely expansive game. At present, there are over 100 locations across 14 biomes, all on a mobile screen. It was therefore crucial to create a distinctive sound for each location, while also addressing artistic, technical, and optimization challenges.

Ambient sounds can be divided into three components:

• Global ambient sounds – the primary background layers heard across the location, such as forest, steppe, highland, cave, mysterious rooms, etc.;
• Object sounds – sounds of objects present in the location that need to be spatially positioned. These include trees, vendors, NPCs, fountains, operating towers, animals, etc.;
• Musical layers – additional textures to create the desired atmosphere. Although they are part of the musical component, they play an important role in the overall ambient sound.

Simply voicing all 100+ locations would not have been the best approach. Our team decided that it was far more effective to develop a system and formula by which locations would be grouped, with each group receiving its own sound treatment based on its visual style and prevailing mood.

Each biome was voiced using a combination of ambient layers and spatially positioned sound sources – meaning that an event should always include the core ambient layers (such as wind, distant machinery hum, a subtle drone) along with randomly played characteristic sounds of the biome like birdsong, bat calls, water droplets, etc.

This is how the setup appears in FMOD:

The screenshot shows 5 layers playing simultaneously with complex logic (random modulation of volume, pitch, and pauses between playback, etc.). Implementing all this logic in code for just one such simple event would take a considerable amount of time, not to mention debugging. Now imagine if there were dozens or hundreds of such events.

With audio engines, the programmer only needs to code that when a location is entered, event X is triggered, and it is stopped upon exit.

Thanks to this approach, we managed, together with the programmers, to integrate all ambiances for over 100 locations in just 3 days.

The developers wanted a musical component that could play almost continuously without becoming tiresome for the player.

A decision was made to create a drone system as a complement to the musical system, which we will describe later.

The AK Audio team discussed the sound design with the AI Grind art team, after which we began working on creating drones, categorizing them, and programming their logic in FMOD.

As a result, we divided all drones into two groups – tonal and noise – and further subdivided each group based on their mood and sound characteristics (e.g., gloomy, tense, eerie, mystical, fairy-tale, vocal).

The drone system works so that the sound starts smoothly, plays at random intervals followed by a random period of silence, and then repeats the cycle.

This ensures that the audio does not become monotonous and the player does not get fatigued by constant music.

Drone System in FMOD

When composing music for mobile platforms, it is essential to consider that storage capacity is limited. Creating 4 hours of music as in AAA projects is not feasible for us. Therefore, it was necessary to find a balance between artistic quality and optimization.

Our team conducted research and found that a two-minute composition (~50 MB) after optimization occupied about 1 MB — which is quite a lot, prompting us to seek a more economical solution.

It was then decided to compose main pieces with leitmotifs to set the desired mood for a specific location or situation, and then play short fragments of the composition in random order with brief pauses to maintain the atmosphere without tiring the player with continuous music. In some pieces, the music adapts between exploration and combat modes. This transition is triggered by a parameter, and the switch is handled in the code.

When the musical composition is playing, the drones stop. This system is entirely controlled by our team using snapshots. Snapshot refers to saved mixer settings that allow on-the-fly adjustments to the mix and sound based on the in-game situation. These settings can be triggered either from the audio engine or via code.

In theory, this system could be recreated without audio engines like FMOD or Wwise, but that would take significantly more time, with even more debugging and detail work, while programmers would only need to indicate when track X is played and when a parameter is switched. In the end, this approach improves the overall quality of the game’s sound, while giving programmers more time to focus on other critical tasks.

Musical System in FMOD

Skylore features an incredible variety of magical creatures, monsters, and robots. It was important to create distinctive sound characteristics for each type of being.

We discussed with the art team how we wanted to hear tree-like monsters, spirits, mechanized enemies, Gorgorush, undead, and other inhabitants of this world. We also consulted with the programmers about which in-game sound cues (enemy attacks, damage sounds, death sounds) could be integrated.

After gathering all the necessary information, we compiled lists of mobs sorted by race with all required details, making it easier to create consistent sound profiles. With over 130 mobs in the game, classifying enemies into groups was essential.

Thus, almost all enemy sounds consist of two components: vocalizations and movement sounds. Each layer has several variations, which are triggered with random changes in volume and pitch within specified limits.

In such cases, programmers only need to attach a trigger to initiate the appropriate event at the correct animation frame, without having to code additional logic.

If the team had decided to code each event separately, it would have taken a significant amount of time and increased the risk of errors—since on average 3-5 events per each of the 130+ characters would result in 400-650 sound events and an enormous amount of code that could easily be avoided.

All these aspects are very important, but they are worthless if the player cannot hear and interpret them during intense battles.

To address this challenge, we devised and implemented a dynamic mix and priority system. This system was realized through the combined efforts of the AK Audio team and the talented programmers at AI Grind.

The system was code-named H.A.E.R, which stands for:

• H – hero – the player’s character;
• A – ally – a friendly character (either an NPC or another player);
• E – enemy – an enemy that is not currently attacking us (could be AI or another player);
• R – redenemy – an enemy that is currently attacking the player/hero (could be AI or another player).

The status of each character can change dynamically depending on their role. Accordingly, the sound volume and frequency characteristics will adjust. Thanks to this system, during particularly intense PvP or PvE battles, the player will always be able to distinguish the most critical sounds. For instance, ambient sounds and normal ally attacks will recede into the background, while the sounds of enemies posing the greatest threat will become more prominent.

Like the music system, this system operates in two modes: exploration and combat.

The system comprises 10 priority levels – with 0 being the highest and 10 the lowest. When an event with a higher priority is playing, all lower-level events are attenuated through sidechain compression.

We know that this article turned out to be noticeably longer than others. Nevertheless, it does not cover many other aspects and details that were applied to address the challenges of creating the sound design for Skylore. However, it already gives a general idea of the scope of work required to achieve truly high-quality game sound.