Switches inference to a transformers pipeline (+pyannote segments), adds overlapping-chunk processing and CSV timecode export for DaVinci.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
HuggingFace matthijs/svd (torch.jit) model + ffmpeg audio extract + tkinter file picker. Proves the concept in one script.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>