Hi everyone,
I'm working on a project to monitor and analyze poaching activity in a protected area using Audiomooth . I have many hours of audio files and I'm looking for a way to automatically detect and extract gunshot sounds from these recordings.
Could anyone recommend software or tools (preferably open-source, but commercial options are fine too) that are good for identifying gunshots in audio data? Any tips on setting up such a detection pipeline, including noise filtering and training models (if needed), would also be greatly appreciated.
Thanks in advance for your help!
My best Filippo, Italy
Try edgeimpulse.com
Hello Filippo,
I have been doing just this kind of analyses for several of my projects. Originally, we used to rely on the DTD gunshot detection algorithm of Cornell University. Recently however, witht he support of WCS Mongolia a new Python based gunshot detection algorithm was developed. It outperforms considerably the old DTD algorithm, as a) it reduces significantly the number of false positive detections, b) increases the recall rate (i.e. actual proportion of gunshots detected) and c) runs faster. The new gunshot detection algorithm does not require your data to be uploaded to the cloud etc. It has not been officially made publically available, but I assume you can ask to use it. Please send me a message to discuss how I can help you get in touch etc.
Happy to hear that you are using passive acoustic monitoring to estimate hunting/poaching pressure. Do not hesitate to drop me a message if you need to discuss the analysis etc.
Best regards,
Christos