Signal Processing Challenges to improve hearing devices

[Trevor Cox]

I run two projects that are running open machine learning challenges to improve hearing devices for those who have a hearing loss. The two projects are:

• Clarity https://claritychallenge.org/ is about speech-in-noise
• Cadenza https://cadenzachallenge.org/ is about music

For these challenges to be a success, we need lots of competitors bringing novel approaches, and hence this message. Both projects have launched new challenges this month.

For Clarity, the latest challenge is predicting speech intelligibility for people with a hearing loss listening via hearing aids. The results from this will be presented at a Workshop before Interspeech in Dublin https://claritychallenge.org/clarity2023-workshop/ 

For Cadenza, music might appear to be slightly outside the remit of this group. But the second task we've set is about listening in the car in the presence of noise. There are techniques in near-end speech enhancement that would be really interesting to explore.

Hopefully you’ll be interested in competing, or maybe you know someone else who would be interested.