Next: Combining multiple time-scale features
Up: A multi-stream approach
Previous: Multi-Stream Statistical Model
As a particular case of the multi-stream approach, a new speech recognition
system based on independent processing and recombination of partial
frequency bands was recently developed and tested on several clean and
Experiments were performed on the
NUMBERS'93 database, a continuous speech telephone
database collected by the CSLU at the Oregon Graduate
Institute . It
consists of numbers spoken naturally over telephone lines on the
The subband-based system had four bands and used
subband log-RASTA-PLP features.
Recombination was done at the state level with a
multilayer perceptron with one hidden layer.
Results, reported on Figure 5.5, clearly show that
the multiband approach yields much less degradation than
the classical approach in the case of band limited noise.
Subband-based speech recognition
Error rate for speech + band limited noise in the first frequency
band (first formant) and various SNR levels. Solid line is for the multiband system, dotted line is for the full-band system.