Evaluating robustness of You Only Hear Once(YOHO) Algorithm on noisy audios in the VOICe Dataset (preprint)
VOICe Dataset Link, click here.
Instructions:
0. Install requirements.txt
.
- Set required parameters like
sample_rate
inconfig.py
. - Download and extract the VOICe dataset in the
data/9dB
anddata/3dB
folder inside theVOICe/
folder. - Run
python process_dataset.py
. - Run
python train.py --backend pytorch -e indoor -alr
for running pytorch code.