(Patent Enrolled)

Introduction

Sound event detection(SED) can be performed to label the datasets derived from these sounds by predicting sound events in the given audio. SED techniques have been applied to audio surveillance, acoustic monitoring for social care, urban sound analysis, multimedia event detection and bird call detection. Humelo uses this technology to cleanse noisy datasets

Model

We use our own SED model. Our SED technology has world-class performance and the paper is published in ICASSP 2019. We are patent-pending with this technology.

Result

The figure below shows the location of the noise predicted by the model. It is where noise is predicted in black.

F1 score and error rate results for single frame segments