Sound source recognizer
Sound source recognition utilizes deep learning method to identify the type/category of recorded audio clips, often one whose sound level exceeds a certain threshold, without human intervention.
The sound source recognizer includes pre-trained deep learning model and audio recording, processing and model recognition units.
The sound source recognizer includes pre-trained deep learning model and audio recording, processing and model recognition units.
- Product Parameters
- Product Quality Control
The sound source recognizer is a module that can be selectively added to our environmental noise monitoring station, to perform the task of automatically identifying the type of the noise. Short audio clips (1 - 6 s, configurable) are recorded once the sound level meter measured a noise level that exceeds a certain preset threshold. Such audio signals are then filtered, noise suppressed and feature extracted, and fed into a pre-trained deep learning network for recognition. The purpose is to automatically identify the major noise sources that exceed local noise threshold, and assess e.g. their occurence pattern and relative contributions to the local noise level.
