Wwise Sounds

type: sound loads WAV files from your pack at runtime and creates real UAkAudioEvent assets. Use these events in ADA messages or other properties that accept a Wwise audio event. Pack authors and players do not need Wwise Authoring installed, and do not generate banks themselves.

welcome.sound.yml

type: sound
sounds:
  - id: AdaWelcome
    file: audio/welcome.wav
    name: ADA welcome volume
    description: Controls playback volume for the welcome message.
    category: Voices
    subCategory: ADA

file starts at the directory containing this pack’s pack.yml, not at the YAML file’s directory. Distribute the WAV alongside the YAML. Absolute paths, paths outside the pack, and symbolic links that resolve outside the pack are rejected. The .sound.yml filename can supply the type when a file contains only one document; every document in a multi-document file needs explicit type.

Accepted audio

Property	Supported value
Container	Little-endian RIFF/WAVE (.wav, case-insensitive extension)
Encoding	Uncompressed signed PCM, 16 bits per sample
Channels	Mono or stereo
Sample rate	8000–192000 Hz
File size	At most 64 MiB per file
Playback	Finite, once, 2D, no spatial attenuation
Volume routing	Pack setting × local Audio option, then Master Audio Bus

PCM16 WAVE_FORMAT_EXTENSIBLE files also work when their valid-bit count and channel mask match mono/stereo PCM. Float WAV, 8/24/32-bit PCM, MP3, OGG, compressed WAV, streaming, loops, effects and spatial attenuation are unsupported. Export recordings as 16-bit PCM WAV; 48000 Hz is a useful choice for voice-over. Sounds are held in memory while loaded, so keep long recordings small.

Field	Required	Meaning
id	Yes	Generated event ID and stable volume-setting identity
file	Yes	Pack-relative PCM16 WAV path
name	No	Exact slider label; defaults to <id> volume
description	No	Slider help text; defaults to generated playback-volume guidance
category	No	Section name; defaults to KDataForge sound events
subCategory	No	Second grouping label; displayed as <category> / <subCategory>

Optional text fields must be non-empty strings when present. Document-level conditions, debug and include use their normal grammar. id uses the same token rules as other generated content and shares the generated asset namespace with images, curves and data assets. Do not reuse one ID for two assets.

Per-event volume

Every sound that passes its document conditions automatically creates two 0–100 controls:

• A pack-wide base-volume slider in KDataForge configuration.
• A local player slider under Options → Audio.

Both controls use the entry’s name, description and grouping. FactoryGame exposes only one section level below Audio, so KDataForge flattens category plus optional subCategory into one heading such as Voices / ADA. Omitting both keeps KDataForge sound events. This preserves Options → Audio as the native top-level category.

The effective amplitude is pack volume × local user volume. With both controls at 100, the authored PCM level is unchanged. One control at 50 is half amplitude (about -6 dB); either at 0 mutes the event.

Saving either control reloads the generated bank, keeps the event UObject path stable, and uses the new level for subsequent playback. Local Audio changes use a short debounce while dragging. A sound already playing may finish with its previous level. Pack values use FactoryGame/Configs/Packs/<PackRef>.config.yml and normal host synchronization/admin rules. Audio values persist locally in Saved/Config/Windows/KDataForgeUserSettings.ini and are not synchronized. Internally generated pack keys live below __sounds; pack-authored config fields must not use that group.

Reflection exports created on a host or client include each active generated slider in indexes/settings.json, including its Audio category placement. Dedicated-server exports omit these local-only options.

Reference an event

Given ref: ADAExample and id: AdaWelcome, the event’s object path is:

/KDataForge/GenAssets/ADAExample.Gen_ADAExample_AdaWelcome

Use AdaWelcome inside the same pack, or the complete path for an explicit cross-pack reference. This also works for soft object references such as FGMessage.mAudioEvent:

type: dataasset
assets:
  - id: WelcomeMessage
    class: /Script/FactoryGame.FGMessage
    properties:
      - path: mTitle
        value: "Welcome"
      - path: mMessage
        value: "Welcome, pioneer."
      - path: mAudioEvent
        value: AdaWelcome

Sounds load in the Assets stage before data-asset property initialization and CDO patches, so sound references do not depend on filenames sorting before message files. Importing a sound does not play it or define an ADA trigger. Configure message triggers, subtitles and queue behavior on the FGMessage separately. See the complete ADA example.

Editor authoring and preview

In the desktop editor, choose File → New document → sound, or use the Wwise sound content wizard. In Visual editor, Sounds provides ID, file, name, description, category and subcategory fields. The wizard creates YAML; place the recording inside the pack yourself. Code editor offers sound completion and hover help; validation checks schema, file containment and the actual WAV header. Go to definition and Rename ID include sound declarations, full event paths and short mAudioEvent references. Runtime simulation, naming audits, dead-content analysis and changelogs also understand sound documents.

In the in-game editor, select sound when creating a pack document. Add a Wwise sound (PCM16 WAV) entry, edit id/file plus optional UI metadata, save, then reload the pack. Preview locally with:

/kdf sound ADAExample AdaWelcome

The game console equivalent is KDF.Sound ADAExample AdaWelcome. Preview returns a real Wwise playing ID and logs both start and end-of-event. It needs a client with initialized audio; a dedicated server cannot produce sound. This preview plays the event directly, independently of ADA’s trigger and subtitle configuration.

Chat preview is available on the local host. Remote players use KDF.Sound in their own client console; a chat request cannot play sound on another machine.

The supported authoring editors are the desktop editor and KDataForge’s in-game editor. Runtime events are generated for the packaged game; KDF.ApplyToAssets does not import them into Unreal Editor assets, and Unreal Editor/PIE audio import is not supported by this runtime path.

Reload, multiplayer and errors

Run /kdf reload after changing YAML or the WAV. The event’s UObject path stays stable. An unchanged recording and volume keep the existing event; changing either updates that object after a replacement bank loaded successfully. A missing, invalid or rejected replacement is reported and leaves the previously imported sound object intact. Removing a declaration also removes its volume slider on reload; close and reopen Options after changing display metadata. Removing a sound requires a session restart to discard the generated event object. Bank cache lives under FactoryGame/Content/KDataForgeRuntimeSounds in the game installation; source audio stays in the pack. The cache directory must be writable. This cache can be cleared while the game is closed and is rebuilt on the next load. Incomplete or damaged cache files are replaced automatically.

Install identical packs and WAV files on all multiplayer clients. Sound data is local and is not transferred by KDataForge’s configuration synchronization. Pack base volumes synchronize like other pack settings; each player’s Options → Audio multiplier remains local. Dedicated servers validate the source and create event references without opening an audio device. A host reload is process-local; clients apply synchronized base values through their own generated banks.

Use /kdf report for diagnostics. Missing files, invalid PCM headers, incomplete frames, unsupported formats, ID collisions and unavailable Wwise audio produce explicit errors. If Wwise rejects a generated bank, inspect LogAkAudio and LogWwiseFileHandler. This importer targets Satisfactory’s Wwise 2025.1 bank format; after a game audio-engine update, use a compatible KDataForge build.

Wwise loads event definitions and media from SoundBanks before events can play; this feature prepares those resources for you. Audiokinetic: Loading Banks