The project aims to classify environment sounds in real-time, by taking audio input in samples of 5 seconds (i.e. as gap-less as possible) and processing them in parallel. The classification has been done using a support vector machine with "rbf" kernel, trained on acoustic and spectrogram features of many versions of 10 kinds of sounds.
Dataset used for this project was ESC-10.
ESC-10 is a part of the ESC dataset (Environmental Sound Classification).
The ESC dataset is a collection of short environmental recordings available in a unified format (5-second-long clips, 44.1 kHz, single channel, Ogg Vorbis compressed @ 192 kbit/s). All clips have been extracted from public field recordings available through the Freesound.org project. Please see the README files for more details. The dataset is available under the terms of the Creative Commons license - Attribution-NonCommercial.
The dataset has three parts:
ESC-50: a labeled set of 2 000 environmental recordings (50 classes, 40 clips per class), ESC-10: a labeled set of 400 environmental recordings (10 classes, 40 clips per class) (this is a subset of ESC-50 - created initially as a proof-of-concept/standardized selection of easy recordings), ESC-US: an unlabeled dataset of 250 000 environmental recordings (5-second-long clips), suitable for unsupervised pre-training.
Snapshots of Results:
