How a voice note becomes words on a screen
A recording contains sound, not a hidden text file. Whisper turns that sound into a representation a model can use, then builds a transcript piece by piece. The task you choose helps determine what those words should be.