generate a sibling dataset with speech context

harvard-edge / multilingual_kws

Few-shot Keyword Spotting in Any Language and Multilingual Spoken Word Corpus

155 stars 35 forks source link

generate a sibling dataset with speech context #15

Open mmaz opened 2 years ago

mmaz commented 2 years ago

v1: discuss as footnote or addendum in camera ready version, that this is slated for v2 and is part of the embedding, mention the boost in accuracy from training with context as part of the camera ready discussion