Speaker Diarization with LSTM
This work addresses speaker diarization for audio processing applications, representing an incremental improvement by applying neural network embeddings to an existing task.
The paper tackled speaker diarization by developing a new d-vector based approach using LSTM embeddings and non-parametric clustering, achieving a 12.0% diarization error rate on the NIST SRE 2000 CALLHOME dataset with out-of-domain training data.
For many years, i-vector based audio embedding techniques were the dominant approach for speaker verification and speaker diarization applications. However, mirroring the rise of deep learning in various domains, neural network based audio embeddings, also known as d-vectors, have consistently demonstrated superior speaker verification performance. In this paper, we build on the success of d-vector based speaker verification systems to develop a new d-vector based approach to speaker diarization. Specifically, we combine LSTM-based d-vector audio embeddings with recent work in non-parametric clustering to obtain a state-of-the-art speaker diarization system. Our system is evaluated on three standard public datasets, suggesting that d-vector based diarization systems offer significant advantages over traditional i-vector based systems. We achieved a 12.0% diarization error rate on NIST SRE 2000 CALLHOME, while our model is trained with out-of-domain data from voice search logs.