Use of artificial intelligence in room acoustics prediction using a photograph

Daniel Milne, Lee Davison, Ludovico Ausiello

Research output: Chapter in Book/Report/Conference proceedingConference contribution


Current methods of determining acoustical parameters must adhere to current standards, require necessary training, include high equipment costs, and is time-consuming. The estimation of acoustics through calculation diminishes some issues with acoustics measurement but can be less accurate. It can be difficult for individuals and businesses in the acoustics field to consider RT60 despite its importance as an influential acoustic parameter which arguably cannot be assumed, especially within educational spaces. Artificial Intelligence (AI) is a program capable of performing tasks normally requiring human intelligence (see section 2.2) and has reached a maturity wherein it can now apply to real-world tasks that were not previously feasible within industries. AIs might reduce costs as they can reduce the time needed to complete a task and reduce the number of resources required. AIs require data, and in most cases, data acquisition, such as images, video, or text, only needs to happen once which can then allow a range of applications including image identification. As a proof of concept, this project aims to investigate if it is possible to create an alternative method for determining RT60 using AI.
Original languageEnglish
Title of host publicationProceedings of the Institute of Acoustics: Reproduced Sound 2020
PublisherInstitute of Acoustics
Number of pages12
Volume42, Part 3
ISBN (Print)9781906913380
Publication statusPublished - 30 Nov 2020
EventReproduced Sound 2020: Taking it up a level - Online
Duration: 17 Nov 202019 Nov 2020


ConferenceReproduced Sound 2020
Abbreviated titleRS2020
Internet address


  • artificial intelligence
  • Acoustics
  • data acquisition


Dive into the research topics of 'Use of artificial intelligence in room acoustics prediction using a photograph'. Together they form a unique fingerprint.

Cite this