What do the folders speaker-dependence and speaker-independence stand for? about avsr-tf1 HOT 2 CLOSED

Got it, thank you very much

from avsr-tf1.

The directories in master/datasets/tcdtimit/splits contain the partitioning of the TCD-TIMIT dataset proposed by

Harte, N.; Gillen, E., "TCD-TIMIT: An Audio-Visual Corpus of Continuous Speech," Multimedia, IEEE Transactions on , vol.17, no.5, pp.603,615, May 2015 doi: 10.1109/TMM.2015.2407694

Each volunteer in that dataset speaks a number of 98 sentences.
In the speaker dependent partitioning, each volunteer appears in both the train set and the test set, contributing with 67 and 31 sentences respectively. In the speaker independent setup, a volunteer is assigned to either the train partition or to the test one.

Does this answer your question ?

from avsr-tf1.