Improving speaker turn embedding by crossmodal transfer learning from face embedding
This work addresses the challenge of speaker modeling in audio tasks, particularly for short utterances, by leveraging crossmodal transfer from face embeddings, representing an incremental improvement over existing methods.
The authors tackled the problem of learning speaker turn embeddings by transferring knowledge from face embeddings, assuming shared latent properties like age and gender. Their three transfer learning approaches improved performance on verification and audio clustering tasks, especially for short utterances, achieving promising advances over competitive baselines on two public broadcast corpora.
Learning speaker turn embeddings has shown considerable improvement in situations where conventional speaker modeling approaches fail. However, this improvement is relatively limited when compared to the gain observed in face embedding learning, which has been proven very successful for face verification and clustering tasks. Assuming that face and voices from the same identities share some latent properties (like age, gender, ethnicity), we propose three transfer learning approaches to leverage the knowledge from the face domain (learned from thousands of images and identities) for tasks in the speaker domain. These approaches, namely target embedding transfer, relative distance transfer, and clustering structure transfer, utilize the structure of the source face embedding space at different granularities to regularize the target speaker turn embedding space as optimizing terms. Our methods are evaluated on two public broadcast corpora and yield promising advances over competitive baselines in verification and audio clustering tasks, especially when dealing with short speaker utterances. The analysis of the results also gives insight into characteristics of the embedding spaces and shows their potential applications.