Sound quality metrics have been used in the industry for several years. Vehicle design and manufacturing, but also home appliance design have benefited from the consideration of sound quality issues. The Ecosonic team is exploring ways to apply these metrics to environmental sound. We hypothesize that examining the sound quality based on such psychoacoustic metrics as loudness, brightness, roughness, we can measure the impact that changes in the acoustic ecology may have on humans and animals alike.
The EcoSonic is an art-science project that models correlations between acoustic information (in the form of sound recordings) through psychoacoustic analysis. We have built a mathematical model of the correlation between Mel-frequency cepstral coefficients and local weather data. The model can predict impacts of weather variation on the acoustic properties of the environment and can be applied to similar environments. We hypothesize that acoustic ecology tools, such as EcoSonic, will substantially expedite environmental analysis and the planning of management strategies.
The overall model is based on articulation entropy. It represents the Acoustic Diversity Model through in the kind of sounds present in the environment. This is equated to how diverse an area is in terms of distinct sound categories. Diversity will vary with climate impact. These machine learning techniques can be developed into classifies for identifying broad events categories.
Here is a poster presenting this work 2017 EcoSonic MSC Poster
The EcoSonic is an art-science project that models correlations between acoustic information (in the form of sound recordings) through psychoacoustic analysis. We have built a mathematical model of the correlation between Mel-frequency cepstral coefficients and local weather data. The model can predict impacts of weather variation on the acoustic properties of the environment and can be applied to similar environments. We hypothesize that acoustic ecology tools, such as EcoSonic, will substantially expedite environmental analysis and the planning of management strategies.
The overall model is based on articulation entropy. It represents the Acoustic Diversity Model through in the kind of sounds present in the environment. This is equated to how diverse an area is in terms of distinct sound categories. Diversity will vary with climate impact. These machine learning techniques can be developed into classifies for identifying broad events categories.
Here is a poster presenting this work 2017 EcoSonic MSC Poster