Home

Welcome to the GigaSpeech wiki!

TAGs in GigaSpeech Dataset:

• a complete punctuation list, that may appear in "text_tn":

<COMMA>
<PERIOD>
<QUESTIONMARK>
<EXCLAMATIONPOINT>

• in dev/test set(human labelling), here is a complete meta tag list:

<SIL> # silence segment
<MUSIC> # music segment
<NOISE> # noise segment
<OTHER> # something else, that human annotators can't tell what it is, i.e. garbage