Transcribes audio using Whisper Large V3 with precise word-level timestamps and confidence scores.
This model is cold. You'll get a fast response if the model is warm and already running, and a slower response if the model is cold and starting up.