The Hernsew Lingster
README.md
sound.voice
The one call the rest of the game makes to be heard.
local Voice = require("sound.voice")
Voice.line("tongue.reedgate.call") -- a written line, spoken
Voice.say("Walk on, then.", "The Lingster") -- free text, spoken if it was baked
Voice.sfx("answer_right") -- a one-shot, flat
Voice.sfx("sfx_splash", { x = 0, y = 0, z = 84 }) -- a one-shot, out there
Voice.ui("click") -- the interface answering
A spoken line always has its text on screen
Every call that speaks also raises hernsew:say with the text, the speaker's
name, and the clip's real length, read off the payload's own header. The
screens slice's caption queue listens on that signal, so the words are up for
exactly as long as the voice is — and a line whose clip was never baked still
goes up, unvoiced. The return value says only whether it was heard.
Everything here is guarded
A clip that was never baked, an anchor that has not spawned yet, a slice that loaded before the mix did — each of them makes this quiet, never an error. A game taken down by a missing noise is a worse failure than a game missing a noise.
Surface
| Call | Does |
|---|---|
line(id, opts?) | speak a line of the written script by id |
say(text, who?) | speak free text; matches it against the script |
sfx(name, at?, opts?) | a one-shot from sound.clips, flat or placed |
ui(kind) | "click" / "hover" / "back" / "page" |
lines() | the whole spoken script, for anybody baking or listing it |
clipFor(id) / lengthOf(id) | the clip and its true length |
isSpeaking() / lastLine() / stop() | what is being said right now |
ready() | whether the mix is up and holding emitters |
__attach / __detach are the mix component's own; nothing else calls them.