alkove.Word

class Word(data: dict, warnings: Sequence[dict] = ())

A word Spot listens for: its text, pronunciations and thresholds, bound to this version’s Spot. Store to_json(), read it back with Word.from_json.

Methods

from_json()

Word.from_json(j: Union[dict, str]) -> Optional[Word]

None for a word compiled for another Spot: its thresholds no longer hold, compile its text again.

of()

Word.of(w: Union[Word, dict, str]) -> Word

A word, its JSON or the JSON string, as a Word; AlkoveError (word.outdated) for one compiled for another Spot.

to_json()

to_json() -> dict

The word as plain JSON, to store and read back with Word.from_json.

Attributes

Attribute	Type	Description
text	str	The word as it was typed.
lang	List[str]	The languages it was compiled in; empty for phonemes written between slashes.
phonemes	List[dict]	Its pronunciations, [{ipa, lang, p}], each with its probability.
thresholds	dict	The score that wakes in each context, {anywhere, pause}: log-probabilities, a word said clearly scores near 0.
calibrated	bool	True once enrolled on the user’s voice, False for a typed word (its language’s default thresholds).
chain	str	The audio its thresholds suit: “raw” (the microphone) or “focus” (Focus’s output).
warnings	list	What to fix in the text or the takes, [{code, message, …}] (word.too-short…); empty for a word read from JSON.