sstracking_metrics
Date
Formats
Categories
Content
sstracking_metrics provides a Python pipeline for evaluating Sound Source Tracking (SST) systems.
Given a multichannel audio recording, an SST system estimates the time-varying positions of sound sources relatively to the recording microphone array. sstracking_metrics enables the objective evaluation of such systems using standard up-to-date tracking metrics. It also includes utilities to read and process items from the LibriJump SST evaluation dataset, which contains audio recordings of intermittent and moving speakers, i.e., speakers that can move when silent.
The code is available on GitHub under MIT licence.