Query By Vocal Imitation (QBV) Browser

Initialize

• set up environment
  • conda env create -f environment.yml or
  • conda env create -f environment_gpu.yml or
• Place audios to be retrieved into static\items - e.g., esc50
• Clone QBV repository git clone https://github.com/jonathan-greif/qbv.git

Run

python main.py

How to Extend

• Add new backend to retrieval_backends
  • a rank(item_paths, query_path, cache=None) function for ranking
  • a forward_batch(paths) function for caching
• add new backend in retrieval.py in
  • cache_item_embeddings(),
  • rank(), and
  • get_retrieval_backends().