Audio
The Gorilla3D audio building block covers everything a game needs to be heard: a backend independent audio manager, two interchangeable drivers (FMOD and OpenAL), a ready-made playlist component, and two speech modules that sit on top of the stack — Text-to-Speech for talking characters and Speech-to-Text for voice input.
Table of Contents
- Architecture
- The Pages of this Section
- Choosing a Driver
- Platform Support
- Units at a Glance
- Quick Start
- See Also
Architecture
Everything is layered on the abstract TGorillaAudioManager. The speech modules
never talk to FMOD or OpenAL directly — they only use the backend independent
interface, so switching the driver does not touch a line of their code.
TGorillaTextToSpeech TGorillaSpeechToText TGorillaAudioPlaylist
(characters, lipsync) (microphone, commands) (tracks, shuffle)
| | |
+-------------+-------------+-------------+-------------+
| |
TGorillaAudioManager (abstract) — sounds, channels, listener
| |
+-------------+ +-------------+
| |
TGorillaFMODAudioManager TGorillaOpenALManager
(FMOD Core API) (OpenAL Soft + EFX)
Beside this stack there is the old TGorillaLegacyAudioManager, a thread based
wrapper around the FMX TMediaPlayer. It is kept for simple "play a jingle"
scenarios and for projects migrating from earlier Gorilla3D versions — see
Audio Manager.
The Pages of this Section
| Page | Content |
|---|---|
| Audio Manager | The backend independent API: sound items, channels, the 3D listener, Update(), and the legacy manager |
| FMOD Driver | TGorillaFMODAudioManager: channel groups, DSP effects, 3D reverb, geometry occlusion, licensing |
| OpenAL Driver | TGorillaOpenALManager: sources, buffers, EFX effects, distance models, codecs |
| Audio Playlist | TGorillaAudioPlaylist: track list, shuffle, loop, auto-advance |
| Text-to-Speech | TGorillaTextToSpeech with the Piper backend, visemes and lipsync |
| Speech-to-Text | TGorillaSpeechToText with the whisper.cpp and Vosk backends, voice commands |
Choosing a Driver
| FMOD | OpenAL | |
|---|---|---|
| 3D spatialization | Yes | Yes |
| DSP effect chains | Yes (full DSP graph) | EFX effects (reverb, chorus, echo, distortion, flanger, compressor) |
| Channel groups / buses | Yes | Channels grouping several sounds |
| Geometry based occlusion | Yes | No |
| 3D reverb zones | Yes | Reverb effect slots only |
| Streaming | Yes | Yes |
| Built-in codecs | Very broad (MP3, OGG, WAV, FLAC, MOD, MIDI, …) | WAV, plus MP3 and OGG Vorbis through the bundled decoders |
| Licensing | Commercial product, free below a revenue threshold | Open source (LGPL / ISC depending on build) |
| Redistribution | FMOD libraries must be shipped with the app | OpenAL Soft libraries must be shipped or installed |
Rule of thumb: take FMOD when the sound design is a feature of the game — buses, effect chains, occlusion. Take OpenAL when you want positional audio without a commercial dependency.
Both drivers can even be used side by side in one application, since every component that consumes audio takes its manager through a property.
Platform Support
| Platform | Legacy manager | FMOD | OpenAL | TTS (Piper) | STT (whisper.cpp / Vosk) |
|---|---|---|---|---|---|
| Windows | Yes | Yes | Yes | Yes | Yes |
| macOS | Yes | Yes | Yes | Yes | Yes |
| Linux | Limited | — | Yes | Yes | Yes |
| Android | Yes | Yes (via JNI) | — | HTTP mode only | Library mode only |
| iOS | Yes | Yes | — | HTTP mode only | Library mode only |
The mobile restriction of the speech modules comes from process spawning: Piper
and whisper-cli are external executables, which mobile platforms do not allow.
Their library and HTTP modes are unaffected.
Units at a Glance
| Unit | Content |
|---|---|
Gorilla.Audio.Manager |
TGorillaAudioManager, TGorillaSoundItem, TGorillaLegacyAudioManager |
Gorilla.Audio.FMOD* |
FMOD driver, system, sounds, channels, channel groups, DSP, reverb, geometry |
Gorilla.Audio.OpenAL* |
OpenAL driver, loader, API bindings, MP3 and OGG decoders |
Gorilla.Audio.Playlist |
TGorillaAudioPlaylist |
Gorilla.Audio.TTS* |
Speech synthesis, speakers, visemes, phonemizers, Piper backend |
Gorilla.Audio.STT* |
Speech recognition, capture, voice activity detection, whisper.cpp and Vosk backends |
Quick Start
uses
Gorilla.Audio.Manager, Gorilla.Audio.OpenAL;
var
LAudio : TGorillaOpenALManager;
LSound : TGorillaSoundItem;
begin
LAudio := TGorillaOpenALManager.Create(Self);
LAudio.Parent := GorillaViewport1;
LSound := LAudio.LoadSoundItemFromFile('sounds/explosion.wav');
LAudio.PlaySound(LSound);
// call once per frame, drives streaming and 3D updates
LAudio.Update();
end;
See Also
API Reference
- Gorilla.Audio.Manager — Base classes of the audio stack
- Gorilla.Audio.FMOD — FMOD driver
- Gorilla.Audio.OpenAL — OpenAL driver
- Gorilla.Audio.Playlist — Playlist component
- Gorilla.Audio.TTS — Speech synthesis core
- Gorilla.Audio.STT — Speech recognition core
Related Topics
- Installation — Installing the FMOD and OpenAL libraries
- Dialogue System — Dialogue trees that trigger audio and speech
- Camera Controllers — Linking a character controller to audio playback