Module render 

Headless rendering of a compiled voice under the standard phrase.

Determinism contract: quiver’s thread-local RNG is re-seeded from PhraseSpec::seed immediately before ticking, and the patch is compiled fresh per render, so (term, spec) → bit-identical samples on any thread.

Structs

Audition

A playback-ready audition buffer.

NoteSpan

Where one phrase note sits in the rendered buffer, and what it was — the role information segment-local features key on (crate::audio finds the held note, the highest note and the chord note by property, never by position, so custom test phrases degrade gracefully).

RenderedPhrase

A rendered phrase: mono samples normalized from quiver’s ±5 V audio level to nominal ±1.0.

Functions

render_phrase

Compile tree and render it playing the phrase.

render_playback

Re-derive the audition buffer of an already-featurized term without re-running the loudness analysis, using the gain_db its crate::Features recorded.