Wylde Flowers is a brand new IP published by Apple that launched in 2022 on Apple Arcade for iOS devices, Mac, and Apple TV, and as of September 2022, it's been available on Nintendo Switch and for PC on Steam. It is a cozy game about farming, magic, community and relationships and was developed by a new remote studio with a HQ in Melbourne, Australia called Studio Drydock.
As audio director and composer, I was responsible for evaluating the needs of the project and for selecting Wwise as an appropriate middleware solution. I was responsible for designing and implementing the way that music functions in the game, as well as doing all of the music composition and production. This blog post will focus on how the music works within the game and how the interactive music features available in Wwise were utilised.
First, a brief description of the gameplay style. Wylde Flowers is a quest-based narrative game. The player character, Tara Wylde, moves to a small seaside town called Fairhaven, initially to help her elderly grandmother out on her farm. She then proceeds to learn about farming and magic by interacting with other characters in the story. The game is fully voice acted and has a strong narrative core. Gameplay is divided up into days of the week, with Tara waking up each morning to continue her quests before needing to ‘go to bed’ sometime before 2am at night.
The music in the game falls under the umbrella of 3 different systems. I will describe each in turn because they all function differently from one another within Wwise.
- Narrative Music
- Diegetic Music
- Scene Music
This is a collection of music cues that relate to the overarching narrative as a whole. Key story moments and some of the major characters are given their own theme which get triggered at various points in the game. The creative director and narrative team were responsible for mapping out these key concepts / story moments and I would then go and write demos to present back to the team for feedback and approval. A ‘Narrative’ state group was set up in Wwise and each narrative theme would then be assigned to a state. This state group was utilised in Wwise’s interactive music system and could subsequently be set at any time during gameplay by certain story quests or cutscenes. In addition, a ‘game mode’ state group was set up in the music system. These states were: ‘default’, ‘conversation’, ‘cutscene’ and ‘fishing’ (more on fishing later). For ‘default’, the main narrative theme in question would play at full intensity. For moments where the player is in dialogue with another character, the ‘conversation’ state would be set and the music would then seamlessly cross-fade to an underscore arrangement. The ‘cutscene’ state would usually trigger the underscore arrangement too, (due to typically high volumes of dialogue in the cutscenes), although I was able to manually switch the intensity of the music back up to the full version in the cutscene when needed by using events placed on the cutscene’s timeline to set the state back to ‘default’ where required.
All of the narrative music is performed by a real orchestra consisting of strings, woodwinds, horns, harp, and rhythm section. Due to the pastoral, gentle nature of much of the music, traditional brass instruments such as trumpets and trombones were not used anywhere in the score. The strings and rhythm section were recorded together with woodwinds and horns overdubbed on top. This resulted in getting more mileage out of the cues by mixing to stems and being able to have ‘lite’ versions of themes, which would usually consist of just strings and rhythm section. When spreading a limited amount of orchestral music across a game that is approx 40-60 hours long, having the stems at our disposal to bring in and out for variation purposes proved to be most useful.
Narrative Music: underscore and full intensity
Narrative Music: cutscene
When wandering around Fairhaven, Tara will hear music coming out of various buildings such as the Pub and the Diner. These music sources operated outside of the interactive music system in Wwise and were set up as positional mono emitters placed in the buildings themselves. Music would play from a randomised playlist of tracks, and the volume of the playlist container would be hooked up to the ‘time of day’ RTPC to match the opening hours of the specific building. Likewise, the ‘day of week’ state group would also be hooked up to the music source, as there are some days of the week where the shop or building is not open.
To simulate the effect of occlusion, a low-pass filter was applied to the music source when Tara is outside the building so it sounds suitably muffled and gives the impression that the music is coming from inside. When she walks in, the filter is removed. A simple effect but one which works quite nicely in the context of creating a dynamic ambient world for Fairhaven.
This part of the music score is the most complex in terms of interactivity, randomisation, and how the music responds to other game states and real time parameters. Scene music is what plays in the background as Tara is going about her day-to-day business, when narrative music isn’t playing. (As with the diegetic music, scene music is overridden by narrative music cues as the narrative music takes precedence to communicate key themes or moments in story development).
The scene music is affected by: season of year, location, time of day, weather, and game mode. At the very top of this musical hierarchy is the season. The key, tempo, chord progressions, and overall structure of the scene music are set by whether the player is in Spring, Summer, Fall, or Winter. When Tara wakes up each morning, a randomised seasonal ‘morning sting’ plays, which is a short flourish to signify the start of a new day. After this, the scene music begins to play. Seeing as she always wakes up on the farm, it is ‘farm’ music that is heard first thing in the morning. When Tara moves to a different area or ‘scene’, the music also transitions to a scene-specific variation of the music (each scene variation resides in its own Music Playlist Container) using the ‘same time as playing segment’ rule in Wwise. So, the new music variation follows the same overall structure and harmonic progression as the previous farm piece, but the instrumental arrangement changes.
The ‘Scene’ state group consists of the following states:
Beach / Rural
Ravenwood (magical village)
With one exception, each of these areas contain their own musical pieces that follow the same overall harmonic progression, and the music system is able to smoothly transition from one to the other. The exception is ‘town’. When Tara goes to the town, the scene music fades away as the musical ambience of the town is provided by the diegetic music emanating from buildings as mentioned above in the section on diegetic music.
In addition, each of the scene music pieces respond to the RTPC ‘time of day’. So, on the farm for example, bright woodwind flourishes are heard in the morning with steady acoustic guitar fingerpicking motifs supplying energy and movement. As the day progresses, the high woodwinds give way to a more relaxed arrangement of horn and lower woodwinds with relaxed guitar strums signifying the afternoon. By the evening, the music has become even more gentle and reflects other changes that have happened over the course of the day such as the lighting, birdsong and sound effects, ambience, and weather. Similarly, morning in the forest tends to be characterised by harp arpeggios, gentle choir, and pizzicato strings. By the evening, tremolo strings provide mystery and excitement as the (spoiler alert!) forest is where the witches coven meet each night. This process is achieved by mixing the music to stems and placing the stems on tracks of the Music Segment Editor. The volume of each track is then hooked up to the ‘hour of day’ RTPC and volume is faded in and out at various times of day, as required.
Scene Music responding to time of day
‘Random-Step’ sub-tracks were used a lot in the scene music system to provide variation in the musical arrangement. Sometimes, an entire stem would have a random sub-track to provide either variation or even stem muting, but more often than not, only certain sections of music would contain randomised sub-tracks. Examples of sectional randomisation includes morning woodwind flourishes, afternoon horn melodies, evening low woodwinds and guitar picking patterns. By combining stem randomization with sectional randomization, plus overall stem playback being manipulated by location and time of day, the scene music provides a dynamic ambient musical accompaniment to general gameplay.
The musical arrangement is further stripped back by muting or attenuating various stems when the player is engaged in conversation with another character. Sometimes a filter might be applied too. This technique allows the music to gently step out of the way of dialogue without drawing attention to itself.
When Tara casts her rod, the scene music transitions to Fishing music. This is a gentle arrangement of Tara’s theme that plays (often on electric piano or steel guitar) as she patiently waits for the fish to bite. Each of the instruments plays 1 of a selection of randomised parts utilising random sub-tracks which helps to avoid repetition. Once she has finished fishing, the game mode state reverts back to ‘default’, and the fishing music plays out before transitioning back, syncing to the ‘last exit position’ of the scene music. There is fishing music for each of the 4 seasons so as to match the key and tempo of each season. When a fish is caught, a musical stinger is played using the stinger system in Wwise.
Scene Music Transitions & Fishing Music
Closing Thoughts and Wish List Features
Overall, it was a positive experience using Wwise’s interactive music features for this project. There is one thing I wish I was able to achieve with the scene music which I couldn’t seem to pull off using the music authoring processes I set up. Each ‘scene’ had its own Music Playlist Container and the music would transition from one playlist to another (e.g. FarmAUTUMN music playlist to ForestAUTUMN). This transition rule would be set to ‘same time as playing segment’ so the harmonic progression would continue to make sense. What I wanted to do was set up randomisation of music segments in each playlist using ‘random step’ feature in the Music Playlist Editor. The idea was that each playlist would contain randomised segments of equal length. However, I found that if I set up randomisation in the playlist, then a transition to another playlist would not be able to retain the elapsed time of the initial playlist as a whole, and the music would lose its structural shape across different playlists. I tried setting up some unique transition rules across different groups of music segments but it got very unwieldy very quickly and was not practical to do so.
I hope this look at the music authoring in Wylde Flowers has been useful!