Whisper within the VRE

Some data needs extra protection because of the content of the data itself or because of the source of the data, like for instance data from children or other sensitive groups. Working with sensitive data can be done securely within the VRE because of the extra security measures in the VRE environment.

The transcription application Whisper is installed within the secure environment of VRE. All computations to make the transcription is contained within the extra secure environment of the VRE.

Currently Whisper is installed as a command line application. We are working on creating a more user friendly interface, but for now you need to create a command-line for Whisper with the source file, the model Whisper needs to use and the output file. The current version is able to run one file at the time and no spaces are allowed in the command.

Make sure that de source file that you want to transcribe is in your home drive H: and not in a sub folder. This is because spaces are not allowed in the command line. Open Notepad or a text editor and create the command line so you can easily paste the whole command with all the components (source file, model and output file) in the command line of Whisper.

The general structure of the command is: whisper [source file] –[model to be used]–output_dir [directory for the output]

Examples or the command are:

whisper H:\opnameAIworkshop.mp3 –model_dir \\hsr\aps\ServerBased\Whisper\LanguageModelfiles –output_dir H:\Whisper_output\

Start Whisper Open the menu and type Whisper. you will find Whisper - CMD

Please refer to the manual of our DCC colleagues on running Whisper on HPC.