Deep neural network based speech separation optimizing an objective estimator of intelligibility for low latency applications
This work addresses speech separation for hearing aids or real-time systems where low latency is critical, but it is incremental as it builds on existing DNN methods with a modified loss function.
The paper tackled speech separation for low-latency applications by proposing a new loss function based on the ESTOI measure to optimize intelligibility, showing improved or comparable ESTOI scores compared to an MSE baseline, though with some trade-offs in separation performance (SDR).
Mean square error (MSE) has been the preferred choice as loss function in the current deep neural network (DNN) based speech separation techniques. In this paper, we propose a new cost function with the aim of optimizing the extended short time objective intelligibility (ESTOI) measure. We focus on applications where low algorithmic latency ($\leq 10$ ms) is important. We use long short-term memory networks (LSTM) and evaluate our proposed approach on four sets of two-speaker mixtures from extended Danish hearing in noise (HINT) dataset. We show that the proposed loss function can offer improved or at par objective intelligibility (in terms of ESTOI) compared to an MSE optimized baseline while resulting in lower objective separation performance (in terms of the source to distortion ratio (SDR)). We then proceed to propose an approach where the network is first initialized with weights optimized for MSE criterion and then trained with the proposed ESTOI loss criterion. This approach mitigates some of the losses in objective separation performance while preserving the gains in objective intelligibility.