Recording of phonetically balanced IEEE sentences by two turn-taking speakers in HWU’s OpenAAL Lab, in the presence of multi-talker background noises. This out-of-domain corpus is being extended to test and optimise new and existing baseline AV SE models
Full preprocessed corpus will be available soon