search

YuanGongND / ssast

Code for the AAAI 2022 paper "SSAST: Self-Supervised Audio Spectrogram Transformer".
BSD 3-Clause "New" or "Revised" License
362 stars 58 forks source link

sed #22

Open fuguanyu opened 1 year ago

fuguanyu commented 1 year ago

Hello, can this project be used for sound event detection? Input variable-length audio, and output chunk detection results at the frame level

YuanGongND commented 1 year ago

hi there,

Yes and no. We do not provide a function in this repo. But frame-wise embedding can be used as the basis for frame-level prediction. See https://arxiv.org/abs/2202.00874.

-Yuan