Diff. between our simulated train and measured eval dataset:
* Talkers speaking differently.
* Real room acoustics.
* Directivity of interferers.
* Transducer noise.
* Lower order Ambisonics.
#ICASSP2023 Clarity Challenge
---
RT @clarityprojuk
More details on the new eval set for #ICASSP2023 signal processing grand challenge. Including a set of example recordings. Release date 1 Feb 2023.
https://claritychallenge.org/docs/icassp2023/…
https://twitter.com/clarityprojuk/status/1612452018873946116