Compute 1 hour of audio in less than a minute on CPU. Faster performance with DirectML on Windows and CoreML on macOS. Accurate timestamps with Pyannote segmentation. Identify speakers with wespeaker ...