/BirdWav2Vec-train

Primary LanguagePython

BirdWav2Vec-train

This training code is based on transformers/examples/pytorch/speech-pretraining in https://github.com/huggingface/transformers

  • check.py checks a target model and dataset
  • extract_result.py creates result.pkl, storeing embedding vectors from all audio samples in a target dataset
  • plot_result.py: plots embeding space from result.pkl
  • model_push_to_hub.py: pushes model to huggingface

Training (pretraining)

sh run_birddb.sh

This script performs speech-pretraining for bird songs using run_wav2vec2_pretraining_no_trainer.py.