For my dissertation project, I aimed to answer the question of whether it’s possible to transcribe the speech of others in a VoIP channel (e.g. a Discord voice channel) and pipe it within a game window without any significant latency or UX issues.
The hypothesis was “yes”, however as the research progressed this changed to “maybe”. The end result continued to be a “maybe” as transcription and sending/receiving the transcribed voice data took over two seconds (including the RSA & AES encryption used)- this meant it was suited more towards slower-paced multiplayer games such as RTS titles and did not meet the project’s internal goal of being usable with an FPS.
My dissertation project achieved a mark of 65% and is currently being moderated, the proposal and final report will be made available here when final marks have been released (the system itself will not be released due to proprietary code that may be completed and turned into a commercial product).