Chapter 17: Speaker Embeddings — Turning a Voice into a Reusable Identity

Voice and AI, Chapter 17: how speaker embeddings represent voice identity separately from content — d-vectors, x-vectors, content/speaker factorization, drift, and why embeddings are sensitive biometric data.

2026-06-03 Last updated on: 2026-06-03 Sho Shimoda

voice ai speaker embeddings d-vector x-vector metric learning biometrics Voice and AI book

Back to Voice and AI