Fig. 4From: Introducing phonetic information to speaker embedding for speaker verificationThe c-vector architecture. The x-vector networks using phonetic adaptation and multi-task learning are combined into a unified architectureBack to article page