First we use Festival to extract the information we want from the utterance structures
Festival includes a script called dumpfeats that will do what we need: print out the Segment relation, and for each Segment tell us the values of all the other linguistic features that we are interested in.
Edit your copy of scripts/dumpfeats and edit it so that it loads an appropriate configuration for your voice (to select the appropriate phoneset). Find this line near the start of the script, and change it if necessary:
(load "/Volumes/ss/festival/lib.incomplete/voices-multisyn/english/localdir_multisyn-rpx/festvox/localdir_multisyn-rpx.scm")
The dumpfeats script needs some configuration files to tell it what to extract from the Festival utterance files and these are provided in data/resources in the files extra_feats.scm and label.feats
Don’t run this yet – we’ll see a script that does this (and more) in the next step.
