This is a complete redesign of the MoonWorks Audio API.
Voices are the new major concept. All Voices can be configured with volume, pitch, filters, panning and reverb. SourceVoices take in AudioBuffers and use them to play sound. They contain their own playback state.
There are multiple kinds of SourceVoices:
TransientVoice: Used for short sound effects where the client will not be keeping track of a reference over multiple frames.
PersistentVoice: Used when the client needs to hold on to a Voice reference long-term.
StreamingVoice: Used for playing back AudioDataStreamable objects.
SoundSequence: Used to play back a series of AudioBuffers in sequence. They have a callback so that AudioBuffers can be added dynamically by the client.
SourceVoices are intended to be pooled. You can obtain one from the AudioDevice pool by calling AudioDevice.Obtain<T> where T is the type of SourceVoice you wish to obtain. When you call Return on the voice it will be returned to the pool. TransientVoices are automatically returned to the pool when they have finished playing back their AudioBuffer.
SourceVoices can send audio to SubmixVoices. This is a convenient way to manage categories of audio. For example the client could have a MusicSubmix that all music-related voices send to. Then the volume of all music can be changed at once without the client having to manage all the individual music voices.
By default all voices send audio to AudioDevice.MasteringVoice. This is also a SubmixVoice that can be controlled like any other voice.
AudioDataStreamable is used in conjunction with a StreamingVoice to play back streaming audio from an ogg or qoa file.
AudioDataWav, AudioDataOgg, and AudioDataQoa all have a static CreateBuffer method that can be used to create an AudioBuffer from an audio file.
Reviewed-on: #50
VideoPlayer now takes AV1 video instead of Ogg Theora. This brings a significant decode speed improvement. The decoder now also operates in a threaded manner, which should prevent runtime stalls when fetching video frames.
Reviewed-on: #49
- Audio is now processed on a background thread instead of the main thread
- Audio tick rate is now ~200Hz
- MoonWorks.Math.Easings class completely rewritten to be easier to understand and use
- SoundInstance properties no longer call into FAudio unless the value actually changed
- SoundInstance property values can now be interpolated over time (tweens)
- SoundInstance tweens can be delayed
- SoundInstance sets a sane filter frequency default when switching filter type
- StreamingSound classes can be designated to update automatically on the audio thread or manually
- StreamingSound buffer consumption should now set Stopped state in a more sane way
- Added ReverbEffect, which creates a submix voice for a reverb effect
- SoundInstance can apply a ReverbEffect, which enables the Reverb property
- Audio resource tracking improvements
- Some tweaks to VideoPlayer to make its behavior more consistent
Reviewed-on: #47
After months of tweaking and refactoring I have realized that collision is like rendering - it's so fundamental to the structure of your game that making broad decisions about how it should work from a library level is too restrictive and difficult to optimize. Anyone skilled enough to use MoonWorks should be easily able to roll their own collision detection.
Reviewed-on: #46
Whenever the video shaders change, they can be rebuilt with refreshc and distributed alongside the moonlibs.
Co-authored-by: Caleb Cornett <caleb.cornett@outlook.com>
Reviewed-on: #42
Co-authored-by: TheSpydog <thespydog@noreply.example.org>
Co-committed-by: TheSpydog <thespydog@noreply.example.org>
This replaces the Refresh-side check for window claim status in AcquireSwapchainTexture, and adds validation to ensure that BindVertexBuffers is not called before BindGraphicsPipeline.
Co-authored-by: Caleb Cornett <caleb.cornett@outlook.com>
Reviewed-on: #37
Co-authored-by: TheSpydog <thespydog@noreply.example.org>
Co-committed-by: TheSpydog <thespydog@noreply.example.org>
cosmonaut
TheSpydog