intellichirp-snaw

Soundscape Noise Analysis Workbench

Demo Coming Soon

Team Website

Workflow

1. Upload a file
2. Click analyze
3. View visualized results
4. Export results to a .csv

Description

Problem

Various environmental changes affect a range of species around the world and as more species are being affected, proper management and observation are required to understand their response. Traditional field methods require trained observers to determine species presence/absence and are thus expensive and challenging to employ at large scales. Using sound to monitor biodiversity across landscapes is a fairly recent development.

Our clients are working with Soundscapes2Landscapes. They are having a problem with an un-user friendly application that requires manual identification in terabytes of sound files. This manual approach is incredibly time consuming and needs to be automated. We feel confident that we can provide a solution that is user friendly and automates that identification process with machine learning.

Solution

Our envisioned solution is a user friendly web application for use by any researcher or citizen scientist. This application is called the Soundscape Noise Analysis Workbench (S.N.A.W.), and will allow users to analyze sound files with the power of machine learning. The results given to the users include a summary of the audio components in the file, acoustic indices, and an export of the sound file with background noise masked out. Users will gain a better understanding of how various sources of noise in soundscape recordings diminish the ability to detect individual bird species and quantify avian diversity. Using machine learning, instead of the current manual identification process, will drastically speed up the identification of terabytes of acoustic data. This solution will allow users anywhere, anytime, to upload their soundscapes for noise analysis, quickly.

Technology Demo

Here are a couple screenshots from our demo.

Uploading audio for analysis

Visualizing results from analysis

Authors

• Steven Enriquez - Team Lead, Front-End Lead - GitHub
• Michael Ewers - Back-End Lead - GitHub
• Joshua Kruse - Machine Learning Lead - GitHub
• Zhenyu Liei - Testing Lead - GitHub

Acknowledgments

• Colin Quinn. - Client
• Patrick Burns. - Client

School of Informatics, Computing and Cyber Systems Global Earth Observation and Dynamics of Ecosystems Lab

• Soundscapes 2 Landscapes - Website