The project comprehends:
- TIMIT corpus segmentation frame by frame;
- feature extracting per each frame;
- frame labeling;
- a BLSTM neural network for automatic frame labeling.
This could be used for further vowel detection, therefore automatic syllable segmentation.
Maxim Gaina