A CoreML sound classification model based on the ECS-10 Dataset
The CreateML project file used to create the ESC-10-CoreML-Demo, as well as the ESC-10 dataset, is included.
The dataset is split into a testing-data set and a training-data set. The CreateML Project Test Audio Files
folder contains additional test CC-NonAttribution sound files downloaded from freesound.org.
A demo iOS app is included in ESC-10-CoreML-Demo. It includes ten pre-loaded .wav audio files
that the are analyzed using the SoundAnalysis framework and the ESC-10 Sound Classifier.mlmodel.
A video of the app can be found here.
A blog post further detailing the project can be found on my website and on Medium.
The contents of this repo, including the CoreML model, are Licensed under the MIT License. See the License.md file for more details.
Email: nicholasarner (at) gmail.com
Website: www.nickarner.com
Twitter: @nickarner