Speaker Diarization using Two-pass Leave-One-Out Gaussian PLDA Clustering of DNN Embeddings
This work addresses the problem of speaker diarization for speech processing applications, offering a robust solution that reduces error rates and eliminates the need for parameter retuning across tasks.
The paper tackles speaker diarization by introducing a two-pass system that uses finer time resolution to improve clustering of DNN embeddings, achieving an error rate below 4% on the Callhome corpus without task-dependent tuning.
Many modern systems for speaker diarization, such as the recently-developed VBx approach, rely on clustering of DNN speaker embeddings followed by resegmentation. Two problems with this approach are that the DNN is not directly optimized for this task, and the parameters need significant retuning for different applications. We have recently presented progress in this direction with a Leave-One-Out Gaussian PLDA (LGP) clustering algorithm and an approach to training the DNN such that embeddings directly optimize performance of this scoring method. This paper presents a new two-pass version of this system, where the second pass uses finer time resolution to significantly improve overall performance. For the Callhome corpus, we achieve the first published error rate below 4% without any task-dependent parameter tuning. We also show significant progress towards a robust single solution for multiple diarization tasks.