Differentiable Tracking-Based Training of Deep Learning Sound Source Localizers
This work addresses a bottleneck in sound source localization for multi-source scenarios, offering a novel training strategy that is incremental but impactful for the field.
The paper tackles the problem of training deep learning sound source localizers for multi-source scenarios by introducing an end-to-end training technique that optimizes tracking metrics directly, resulting in large improvements in localization error, detection metrics, and tracking capabilities.
Data-based and learning-based sound source localization (SSL) has shown promising results in challenging conditions, and is commonly set as a classification or a regression problem. Regression-based approaches have certain advantages over classification-based, such as continuous direction-of-arrival estimation of static and moving sources. However, multi-source scenarios require multiple regressors without a clear training strategy up-to-date, that does not rely on auxiliary information such as simultaneous sound classification. We investigate end-to-end training of such methods with a technique recently proposed for video object detectors, adapted to the SSL setting. A differentiable network is constructed that can be plugged to the output of the localizer to solve the optimal assignment between predictions and references, optimizing directly the popular CLEAR-MOT tracking metrics. Results indicate large improvements over directly optimizing mean squared errors, in terms of localization error, detection metrics, and tracking capabilities.