In this project, the temperature of water was to be determined by recording the sound made whilst being poured into a container. The goal was set to achieve an accuracy of at least 80% of measurements should be predicted within ±10°C and a negligible amount outside ±20°C. The approach to this problem was through creating an algorithm based on signal analysis and machine learning, which required a large quantity of audio samples to train. This algorithm would then be put into an application for added versatility. The resulting algorithm was able to make a prediction of 78.4% of measurements within ± 10°C and the ± 20°C interval contained all but 5.4% of the measurements. The app did not work as intended on mobile devices but is functional on a computer. Although the computer version is functional and can predict temperatures, it is not accurate enough to be useful in day to day life.
| Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:uu-477111 |
| Date | January 2022 |
| Creators | Rohlin, Hannes, Thulin, Jesper |
| Source Sets | DiVA Archive at Upsalla University |
| Language | English |
| Detected Language | English |
| Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
| Format | application/pdf |
| Rights | info:eu-repo/semantics/openAccess |
| Relation | MATVET-F ; 22003 |
Page generated in 0.0017 seconds