2.1.1

Version `2.1.x` introduces a major overhaul of `pyannote.audio` default speaker diarization pipeline, made of three main stages: 
* neural [speaker segmentation](https://hf.co/pyannote/segmentation) applied to a short sliding window;
* neural [speaker embedding](https://hf.co/speechbrain/spkrec-ecapa-voxceleb) of each (local) speakers;
* (global) agglomerative clustering. 

More details in the attached [technical report](https://github.com/pyannote/pyannote-audio/releases/download/2.1.1/technical_report.pdf).