UXSound

UXSound makes a sound out of a few numbers (a waveform, a pitch and a slide, a length, a volume) instead of loading a recording. It renders the samples in xc, so the synthesis is the same on every backend and is tested headless. The driver only plays the finished buffer. Sounds overlap: each is its own buffer, and the platform mixes them.

#use <UXKit>            // or #import "UXSound.xc"

Overview

UXSound* chime = new UXSound();
chime.note(UXSND_SINE, 880.0, 0.10, 0.07, 0.0, 0.0);    // 880 Hz for 100 ms
chime.note(UXSND_SINE, 1170.0, 0.14, 0.06, 0.0, 0.09);  // a second note, 90 ms in
chime.play();

UXSound* boom = new UXSound();
boom.boom(0.16, 0.0);
boom.play();

Build a sound once and play it as often as you like, including while it is still playing: each play is a copy.

The shapes follow WebAudio, so sounds written for a browser port number for number:

• A note is an oscillator whose frequency can ramp exponentially to slideTo, like exponentialRampToValueAtTime.
• Its envelope rises exponentially from silence to vol in 12 ms and falls exponentially back to silence at secs. A sound has no click at either end.
• A boom is white noise through a lowpass filter whose cutoff falls from 900 Hz to 120 Hz, decaying over half a second.

The samples are 16-bit mono at 44.1 kHz.

Backend	Plays through
macOS	NSSound, from an in-memory WAV
Windows	one waveOut stream per sound
Web	an AudioBuffer source per sound
Linux (GTK)	PulseAudio’s simple API, loaded at run time (PipeWire serves it too)
iOS, Android, GEM	nothing yet: play answers false

play answers false where a sound cannot play: no audio on that backend, no sound server running on Linux, or, in a browser, before the person has clicked or typed in the page. Browsers block audio until then, and UXKit resumes it on the first gesture.

Topics

Building · note · boom Playing · play Samples · samples / frameCount

note

UXSound* note(i32 kind, double hz, double secs, double vol, double slideTo, double at)

Adds a note: the waveform (UXSND_SINE, UXSND_TRIANGLE, UXSND_SQUARE or UXSND_SAWTOOTH), its starting frequency in hertz, its length in seconds, its peak volume from 0 to 1, the frequency it slides to (0 for none), and when it starts, in seconds from the start of the sound. Returns the sound, so notes can be chained. A sound holds up to eight parts.

boom

UXSound* boom(double vol, double at)

Adds a boom (filtered noise, half a second long) at peak volume vol, starting at seconds in. Its noise is seeded, so it is the same boom every time.

play

bool play(void)

Plays the sound now and returns at once. True when it started, false when this backend cannot play it (see above).

samples / frameCount

i16* samples(void)
i32 frameCount(void)

The rendered samples, signed 16-bit mono at 44,100 a second, and how many there are. The sound lasts until its last part ends, plus 20 ms.