Microsoft WavLM Pre-training and TDNN Downstream Code for Speaker Recognition

The code used for Microsoft WavLM pre-training and TDNN Downstream in Speaker Recognition is based on the open source Kaldi speech recognition toolkit. The code includes scripts for training and evaluation of WavLM pre-training and TDNN-based downstream models. It also includes code for creating features for use in speaker recognition tasks. The code is available on GitHub at the following link:

https://github.com/Microsoft/WavLMPretrainAndTDNNDownstream