search

MontrealCorpusTools / ISCAN

Development repository for Integrated Speech Corpus Analaysis (ISCAN)
MIT License
9 stars 0 forks source link

Datasets using DISC transcriptions should not select 'stressed vowels' subset #223

Closed vannawillerton closed 5 years ago

vannawillerton commented 5 years ago

ISCAN incorrectly selects a 'stressed vowels' subset in the 'phone subset' enrichment for datasets using DISC transcriptions which do not distinguish stressed vowels:

Screenshot from 2019-06-17 12-30-45

Ideally this page would look like it does for Buckeye and would not allow a selection of stressed vowels:

Screenshot from 2019-06-17 12-38-35

The datasets we have which use DISC transcriptions are SOTC, GlasgowBiD & Glaswasian

vannawillerton commented 5 years ago

Additionally, the 'syllabics' subset does not automatically select all the vowels, but a small subset. This should also be fixed.

msonderegger commented 5 years ago

Bizarre. @MichaelGoodale could you investigate/fix?

MichaelGoodale commented 5 years ago

So these default subsets weren't using the SPADE config files(they were hard-coded), but I'm in the process of updating them so that that's the case. From now on it'll use the spade config file, and if that config file is empty (e.g. no stressed vowels), that subset will be empty