May 4, 2020
An ICASSP like no other
This week, we are (virtually) heading to the prestigious IEEE International Conference on Acoustics, Speech and Signal Processing (aka ICASSP), which is the world’s largest and most comprehensive technical conference focused on signal processing and its applications to audio. Originally planned as a conference in Barcelona, the organisers have turned it into a free virtual conference which runs between the 4th and 8th of May. You can register here.
Alongside other world experts we will be presenting our technical paper “A Framework for the Robust Evaluation of Sound Event Detection.” on Wednesday 6th May, 09:00 – 09:20 (CEST), although the video presentation is also be available on-demand on the ICASSP site.
The presentation introduces Polyphonic Sound Detection Score (PSDS), a new evaluation framework and metric for polyphonic sound recognition systems. In addition to the paper, which describes the metric in full detail, we’ve also made the PSDS source code available on GitHub, enabling researchers to apply the score to their own sound recognition models.
PSDS has also been adopted for Task 4 of the 2020 DCASE challenge.
Throughout the week we will be running our own virtual Q&A sessions, the details of which you can find below. We would love for you to join us.
A dedicated ICASSP event page
We have our own ICASSP page here which will be updated throughout the week with links to relevant PSDS information and the timings of our own ICASSP events.
During the ICASSP week, you can use the live Q&A chat on our ICASSP dedicated webpage to engage with the paper authors.
Live Q&A sessions
We can’t be there to meet face-to-face, so instead please drop in on one of the following open ‘ask our authors’ sessions.
Wednesday 6th May 17:00 – 18:00 (GMT+2) – Ask our authors – We are hosting our own Q&A session where you can ask questions/discuss our paper and find out more about us and our expertise about sound recognition in general. Find out more and join the event here: aud.al/WednesdayPM
Thursday 7th May 11:00 – 12:00 (GMT+2) – Ask our authors – We are hosting our own Q&A session where you can ask questions/discuss our paper and find out more about us and our expertise about sound recognition in general. Find out more and join the event here: aud.al/ThursdayAM
Official ICASSP Q&A
In addition to our own Q&A sessions, there is an official Question and Answer platform live for two weeks from the opening day of our paper (Wednesday 6th May). ICASSP participants will be able to post questions to the authors of the paper.
You can find the full technical paper, along with the video of our ICASSP 2020 presentation and links to our code on GitHub with the Jupyter Notebook, here.