As a pair, we wish to develop a Web App that focuses on automatic speech transcription of lecture recordings. We want the application to facilitate uploading of recordings to transcribe in some degree and a way of accessing previous transcribed recordings. We envision having an aesthetically pleasing front end (React, Redux - SPA), and a sufficiently fast backend with data storage capabilities. To improve the accuracy of the transcription, we propose to use machine learning tactics to develop a profile for individual lecturers – for example, during the upload there would be an input to identify the lecturer, allowing the application to retrieve and train the lecturers profile. We also have additional ideas to promote collaboration efforts, such as having the transcribed document being shared across a class. This would enable students to annotate and potentially timestamp important parts of a lecture. We believe this may be too large of a scope, our primary focus is the development of the front end and ML/Speech transcription tool. We are of course open to discussions about scope as we are not completely sure of the bounds on this project. As students, we have found that going through lecture recordings and transcribing what the lecturers have been saying has been extremely important to excelling in university. We believe that many students would see a tangible benefit in automating this process, as we’ve heard complaints from students outside of engineering about the lack of such a tool.
Lab allocations have not been finalised