Jonathan Le Roux

jonathanleroux.bsky.social

11 January 2025 13:09

We're sad to bid farewell to Kevin Wilkinghoff at the end of his 6-month stay as a visiting research scientist at MERL🍻 We learned a lot and had a great time🤗

Joan Serrà

serrjoa.bsky.social

21 December 2024 17:02

I'm happy to have two papers accepted at #ICASSP2025! 1) Contrastive learning for audio-video sequences, exploiting the fact that they are *sequences*: arxiv.org/abs/2407.05782 2) Knowledge distillation at *pre-training* time to help generative speech enhancement: arxiv.org/abs/2409.09357

Alex Nichol

unixpickle.bsky.social

18 December 2024 10:33

A funny thing about training NNs on a machine with unified memory is that your run doesn't just OOM, instead it uses a ton of swap before giving up.

Jonathan Le Roux

jonathanleroux.bsky.social

15 December 2024 00:57

@waspaa.com is on Bluesky! Big changes are coming for WASPAA 2025!

Alex Nichol

unixpickle.bsky.social

15 December 2024 00:56

I always forget this about neural networks for audio: if you are going to pad your waveform with 0's, or your audio already has a lot of 0's, you can add a tiny bit of noise to the input to stabilize normalization layers.