search

harvard-edge / multilingual_kws

Few-shot Keyword Spotting in Any Language and Multilingual Spoken Word Corpus
155 stars 35 forks source link

support evaluate mode in input_data #6

Closed mmaz closed 2 years ago

mmaz commented 2 years ago

this adds AudioDataset.eval_with_silence_unknown which appends silence and unknown clips to the full set of (test) files