ICPR
2006
IEEE
15 years 20 days ago
2006
IEEE
In most approaches to speech recognition, the speech signals are segmented using constant-time segmentation, for example into 25 ms blocks. Constant segmentation risks losing info...