No room required

kujamba render, play, and trigger use the instrument without connecting to a server. Only play and trigger require an audio-enabled build and an output device.

Render WAV or Ogg

./zig-out/bin/kujamba render --phrase "kujamba karibu" --out karibu.wav
./zig-out/bin/kujamba render --phrase "kujamba karibu" --out bars.ogg \
  --play loop --pattern 3+1 --bars 4 --bar-ms 500 --seed 42

The extension chooses the container. With no --bar-ms, a bar is as long as the rendered phrase. Render defaults to one bar and pattern 1 unless your config supplies another pattern.

A rest bar is an equal-length stretch of silence, not a missing segment. It freezes the phrase position. That keeps the offline timeline aligned with what the room instrument would play.

The Ogg seed selects the stream serial, not the waveform. The synth seeds its noise from phrase text. Repeatability assumes the same build and inputs; exact synthesized bytes are not promised across optimization modes or platforms.

Audition locally

./zig-out/bin/kujamba play "kujamba karibu" --play once
./zig-out/bin/kujamba play --phrase "po" --device 1

--device accepts a zero-based index or a case-insensitive name substring. Playback uses vendored miniaudio, not the visual app's external audio players. It opens playback only, never a capture device.

A missing device, an invalid device selection, or -Dlive=false produces a clear failure. See audio build options.

Trigger a note map

Create a local preset, for example sampler.toml:

[map]
note60 = "kujamba karibu"
note61 = "shuzi:3"

Then start the sampler:

./zig-out/bin/kujamba trigger --config sampler.toml

Type on 60 or on 61, followed by Enter. off 60 is accepted but currently does not gate or release the sound. q or EOF exits. Notes are numbers in 0..127; unknown notes and malformed lines are warned about and ignored.

For a repeatable input sequence, --script FILE reads the same line protocol. This is a stdin/script sampler, not hardware MIDI support. Do not assume that plugging in a MIDI keyboard supplies those lines automatically.

Next: tune the voice or save a preset.