In this work, we intend to maximize the speaker information by reconstructing the extracted speaker information in one feature from the other features while at ...

In this paper, we propose the a multi-feature integration method to utilize complementary acoustic features into a single x-vector system, which would extract ...

In this paper, we analyzed the feature engineering based on the x-vector structure, and proposed a multi-feature integration method to further improve the ...

The speaker embeddings are well-designed to train the distributed features throughout the utterance and are modeled to extract the rich information as a fixed- ...

The method uses a two-stage approach: first, estimate celebrity chunks in recordings, then train the extractor using these chunks with standard supervised ...

Oct 21, 2024 · In this work, we propose a multi-level speaker representation approach, from raw features to neural embeddings, to serve as the speaker reference cue.

Aug 30, 2024 · This paper proposes a method for extracting speaker embeddings from multi-speaker recordings using recursive attention weights for pooling ...

Sep 3, 2021 · This paper introduces a novel approach for extracting speaker embeddings from audio mixtures of multiple overlapping voices.

We propose a simple yet effective Hierarchical Speaker Representation Learning (HSRL). The proposed method comprises three modules.

In this paper, we study the technology of multiple acoustic fea- ture integration for the applications of Automatic Speaker Veri-.