Only transcribe parts of audio files that contain a meaningful signal. This comes in handy especially with the new multitrack feature, which otherwise is very expensive to use when you're charged for transcribing all the silence...