Simple app for sound event detection.
This app requires Python 3 and FFmpeg to run.
Install the requirements and start main.sh
.
$ cd sound-event-detection
$ pip install -r requirements.txt
$ ./main.sh path-to-mp4video
To train a model, start train.py
. It expects a directory dataset
in the same directory as train.py
. Every video in that directory should be around 3 sec long. If in the video is a particular sound which should be detected - name_1.mp4 , otherwise name_0.mp4.
- Experiment with different neural network architectures