alkove.Timbre

class Timbre(models: Models)

Who is speaking: a print from 16 kHz audio (1.5 s or more), compared with others.

Methods

embed()

embed(pcm: Any, *, source: str = 'raw') -> Voiceprint

Windows of 1.5 s, half overlapping, their mean without the 10 % least alike; source says the audio was raw or Focus’s output.

verify()

Timbre.verify(
    a: Voiceprint,
    b: Voiceprint,
    *,
    level: str = 'balanced',
) -> Tuple[bool, float]

a: the known print, b: the voice heard; level lenient, balanced or strict, thresholds set for how much voice each holds.

identify()

Timbre.identify(
    voice: Voiceprint,
    known: dict,
    *,
    level: str = 'balanced',
) -> Optional[Tuple[str, float]]

The known voice that matches best, if any matches.