Function speaker_from_voice 

pub fn speaker_from_voice(
    bundle: &ModelBundle,
    path: &Path,
    cleanup: ReferenceCleanup<'_>,
) -> Result<Vec<f32>, FttsError>

Derive a speaker vector from a voice source: a raw x-vector file, or reference audio.

Passing Some for a cleanup slot opts the reference into that stage and fills the slot with what it measured. Dereverberation runs first: it is a linear operation on the observed signal, so applying it before the noise floor is estimated keeps that estimate from being fitted to a signal the next stage is about to change.

Errors

When the source cannot be read, decoded, resampled, or encoded into a finite x-vector.