harvard-edge / multilingual_kws

Few-shot Keyword Spotting in Any Language and Multilingual Spoken Word Corpus
163 stars 37 forks source link

TFDS api #23

Open mmaz opened 3 years ago

colbybanbury commented 3 years ago

Similar to common voice or speech commands

mmaz commented 3 years ago

convert from opus to wav on disk as part of the download pipeline (depends on opuslib)

v1 timeline: between nov soft launch and dec public launch