Results

The Clarity ICASSP-2023 Enhancement Challenge is now complete. Results are shown below along with links to the system description papers.

The table below reports the HASPI, HASQI and average (over HASPI and HASQI) scores for all the submitted systems. Results are shown for both Eval 1 (the simulated evaluation set) and Eval 2 (using the real recordings). Where a system ID ends with '_data', additional training data has been used. Those that end with '_hr' has exploitied the head rotation information. Links to the system reports are also provided. The Eval 1 average has been used to rank the systems.

Rank	Team	ID	Paper	Eval 1 (HASPI)	Eval 1 (HASQI)	Eval 1 (Average)	Eval 2 (HASPI)	Eval 2 (HASQI)	Eval 2 (Average)
8	Baseline	Baseline		0.266	0.128	0.197	0.176	0.121	0.149
9	T001	E002	Link	0.179	0.093	0.136	0.101	0.078	0.09
7	T002	E009	Link	0.286	0.161	0.224	0.126	0.108	0.117
5	T003	E014	Link	0.797	0.414	0.606	0.291	0.11	0.201
10	T004	E023	Link	0.117	0.047	0.082	0.026	0.019	0.018
2	T005	E028	Link	0.78	0.526	0.653	0.026	0.019	0.022
1	T005	E028_data	Link	0.816	0.57	0.693	0.249	0.154	0.199
4	T006	E029	Link	0.835	0.393	0.613	0.256	0.104	0.18
3	T006	E029_hr	Link	0.838	0.393	0.616	0.256	0.103	0.18
6	T007	E030	Link	0.729	0.316	0.522	0.284	0.132	0.208