Sound Playback¶
This page describes the sound playback features supported by Kirikiri Z.
Contents¶
Time stretching/pitch shifting
Overview¶
The current Kirikiri Z uses miniaudio as its single sound engine in all builds (the old DirectSound implementation has been removed).
- WINVER — miniaudio owns the output device and outputs to it directly.
- SDL(GENERIC) — miniaudio runs with device I/O disabled (no-device mode) to generate PCM, and SDL3 owns the output device.
In both lines, the core of sound processing — decoding, mixing, effects, and 3D positioning — is the same.
Specifications¶
| Values | Bit depth | Integer: 8-bit to 32-bit Floating point: 32-bit |
| Values | Channels | 1ch, 2ch, 4ch, 6ch (5.1ch) |
| Values | Sampling rate | Depends on the source (44.1 kHz / 48 kHz, etc.) |
| Settable values | Volume | 0 to 100000 0: fully muted 100000: 100% |
| Settable values | Balance (panning) |
-100000 to 0 to 100000 -100000: fully left 0: center 100000: fully right |
| Settable values | Playback position | Milliseconds, sample number |
Supported formats¶
WaveSoundBuffer supports the following formats out of the box.
| Format | Extension | Notes |
|---|---|---|
| WAV (RIFF PCM) | .wav | Supports WAVE_FORMAT_PCM (8- to 32-bit integer), WAVE_FORMAT_IEEE_FLOAT (32-bit float), and WAVE_FORMAT_EXTENSIBLE (PCM / IEEE_FLOAT subtypes). |
| Ogg Vorbis | .ogg | Vorbis decoding. |
| Opus | .opus | Opus decoding. |
Integer PCM is played as is, without reducing the bit depth (8-bit = unsigned / 16-, 24-, and 32-bit = signed; 24-bit is handled in its packed 3-byte form). With WAVE_FORMAT_EXTENSIBLE, if the number of valid bits is smaller than the container width (for example, 24 bits left-aligned in a 32-bit container), samples are read at the container width.
Decoder allocations go through the TLSF pool of SoundAllocator. The pool size can be set with -soundpoolsize; the default is 16MB.
Plugins can add support for more formats.
Seamless looping (.sli)¶
A sound loop information file (.sli file) enables seamless loop playback with no audible seams. .sli files, which contain the loop point settings, are created with the loop tuner.
3D positioning (spatializer)¶
A 3D positioning feature based on miniaudio is provided.
- When
WaveSoundBuffer.use3Dis enabled, the buffer is treated as a 3D sound source. - On the source side, set position, velocity, and directivity with
set3DPosition/set3DVelocity/set3DCone, and so on. - On the listener side, control position, orientation, velocity, and so on with
SoundListener.
The distance attenuation model can be chosen from the following four.
| Model | Description |
|---|---|
| amNone | No distance attenuation |
| amInverse | Inverse attenuation |
| amLinear | Linear attenuation |
| amExponential | Exponential attenuation |
HRTF is not supported, so front and back cannot be distinguished.
Lip-sync/analysis API¶
APIs are provided for analyzing the sound being played. They can be used for lip-sync and audio visualization.
| API | Description |
|---|---|
| getSoundLevel | Gets the RMS / peak level. |
| getSoundSpectrum | Gets the spectrum per frequency band. |
| getVowel | Gets estimated weights for the five Japanese vowels. |
| getVisBuffer | Gets raw PCM around the playback position. |
Time stretching/pitch shifting¶
Time stretching (tempo change) and pitch shifting with a Phase Vocoder are supported. They are applied through WaveSoundBuffer.filters.