Are you interested in improving music listening for people with hearing loss? In the Cadenza project, we are running challenges to do this for hearing aid wearers. We are hosting a short webinar on Friday 13th September at 11am UK time. Please join us. https://t.co/x80oK0u6HM
We've uploaded a preprint of our paper about the CAD1 and ICCASP24 challenges https://t.co/gKe6hJBApp The first Cadenza challenges: using machine learning competitions to improve music for listeners with a hearing loss
If you can't make the webinar, feel free to ask Qs in advance on our google group. https://t.co/v2n3kMybHc We'll be showing how to use the baselines, explaining the rules and introducing the challenges.
We're running a webinar and Q&A to help entrants to the latest @CadenzaChalleng challenges on improving lyric intelligibility and rebalancing classical ensembles. https://t.co/Yn1uGhsCul Fri, 13 Sept, 11am UK UK time. We'll record for those who can't make it.
Cadenza ICASSP24 challenge papers now on @IEEEXplore E.g. overview paper https://t.co/cayE54wBYm
BTW For the currently open CAD2 challenge https://t.co/OGhJw7k4Fk, we're planning to have the challenge workshop at a special session of the ICA/ASA in New Orleans in May 2025.
We're in the latest IEEE Inside Signal Processing Newsletter. https://t.co/Qj4Abkcf96 because our latest machine learning challenge is being kindly supported by IEEE SPS. This means we have cash prizes.
Now launched CAD2: The Second Cadenza Signal Processing Challenge to Improve Music for Those with Hearing Loss. Submission deadline: Jan 2025. CAD2 is part of the IEEE SPS Challenge Program @IEEE_AASP . The SPS are funding cash prizes :) 1/2 https://t.co/mvHfNPmvaV
Our 2nd Machine Learning Challenge CAD2 has launched 🚀. There are two tasks 1. Improving the intelligibility of lyrics. 2. Rebalancing the level of instruments within a classical music ensemble. https://t.co/x8enUWyJ8Q 🎶🦻🏽
Look out for the CAD2 launch tomorrow. As well as this task on improving lyric intelligibility there is also one on rebalancing classical music for small ensembles. #machinelearning#music#challenge
CAD2 Machine Learning challenges to improve music for those with hearing loss. We've been developing a metric to measure lyric intelligibility. It uses the pyClarity hearing loss model + Whisper ASR. Graph is words correct. x: listener scores. y:new metric. 1/2
It also depends on the type of music. Ballads need to be clear so the plot can be followed, other genres (heavy metal?) maybe less so. More on the challenge https://t.co/OGhJw7k4Fk 3/3 Thanks to the focus group participants. #DraftResults
In prep for our CAD2 challenge launching in 2 weeks 😬 we've been working with a focus group of people with hearing loss to define Lyric Intelligibility "Lyric intelligibility refers to how clearly and effortlessly the words in the music can be heard." 1/3
Interesting definition that doesn't include understanding the lyrics, just being able to hear them clearly. The defn challenge is lyrics aren't always clear and that is an artistic choice (e.g. singer whose words are intrinsically unclear). 2/3
We're pleased to announce of second challenge launching in July will be part of the @IEEEsps Challenge Programme. This will mean we will have some cash prizes for the best entries 🏆 https://t.co/OGhJw7k4Fk outlines the tasks, full website going live in a few weeks.
People with hearing loss tend to get fewer words correct as might be expected. But where scores are high (lots of words correct) the hearing loss group seem to be doing better.
These are words correct. Listeners hear a bit of music and say what lyrics they heard. x-axis is the old study. y-axis our new study with people with hearing loss. #preliminary
These are words correct. Listeners hear a bit of music and say what lyrics they heard. x-axis is the old study. y-axis our new study with people with hearing loss. #preliminary
Preps for our next machine learning challenge CAD2 https://t.co/OGhJw7k4Fk in full swing. One track is improving lyric intelligibility for people with hearing loss. We've been replicating a previous study https://t.co/VqG9l1pwLL to refine our metric for lyric intelligibility
🧵
Preps for our next machine learning challenge CAD2 https://t.co/OGhJw7k4Fk in full swing. One track is improving lyric intelligibility for people with hearing loss. We've been replicating a previous study https://t.co/VqG9l1pwLL to refine our metric for lyric intelligibility
🧵
Entering our CAD2 challenge (July launch)? Two panel sessions at the #VCCA2024 next week. What people with and without hearing loss want from music. Future development of AI for enhancing music for listeners with hearing loss https://t.co/zF0ACbEX7K
Thanks to #ICASSP2024 for allow us to run a grand challenge on improving music for those with a hearing loss. We'll be launching our next challenge in July, and preliminary details are now on our website https://t.co/OGhJw7k4Fk