Performance Based Cost Functions for End-to-End Speech Separation
This work addresses the issue of perceptual discrepancies in speech separation for audio processing applications, but it is incremental as it builds on existing neural network approaches by modifying the cost function.
The paper tackled the problem of source separation in speech by proposing new loss functions based on perceptual metrics like SDR, SIR, SAR, and STOI, instead of using mean squared error (MSE), and found that combinations of these cost functions achieved superior separation performance compared to standalone MSE and SDR costs.
Recent neural network strategies for source separation attempt to model audio signals by processing their waveforms directly. Mean squared error (MSE) that measures the Euclidean distance between waveforms of denoised speech and the ground-truth speech, has been a natural cost-function for these approaches. However, MSE is not a perceptually motivated measure and may result in large perceptual discrepancies. In this paper, we propose and experiment with new loss functions for end-to-end source separation. These loss functions are motivated by BSS\_Eval and perceptual metrics like source to distortion ratio (SDR), source to interference ratio (SIR), source to artifact ratio (SAR) and short-time objective intelligibility ratio (STOI). This enables the flexibility to mix and match these loss functions depending upon the requirements of the task. Subjective listening tests reveal that combinations of the proposed cost functions help achieve superior separation performance as compared to stand-alone MSE and SDR costs.