We have on offer an offline transcription service based on OpenAI Whisper Automatic Speech Recognition (ASR) package to transcribe uploaded audio or video files. The motivation behind offering this service is to allow researchers the ability to transcribe audio and video without their uploaded data leaving New Zealand or being stored / used for business puproses. For this reason, all uploaded data and related transcription files are deleted off of the server the moment the transcription file is downloaded.
This transcription service implements the standalone project written by Purfview. The reason for choosing this package is that it offers a seamless experience if you were to run the package on a local machine. Meaning that a researcher could use it on their local machine and get the same experience as the web app. It offers better time stamping of the audio, and it runs the faster-whisper package by SYSTRAN, which faster and has support for other models, but the current implementation is only running Whisper ASR models.
This Whisper ASR package has been trained using machine learning, but is secure to use at UC. Your data will not be used to train future versions of the app. All files are automatically deleted after closing the browser window. Please ensure that you download the generated transcript file(s) and save it in a secure location. You should also save the original audio or video file in a secure location.
