Speech Processing
by Computer
LAB 9
ISOLATED WORD RECOGNITION
In this lab
session we will test an isolated-word, template based recogniser on a
relatively easy vocabulary, but with the problem of multiple speakers.
1.
Vocabulary design
and capture
a.
Design a
vocabulary of about 20 words. Choose
words with a variety of segmental structure and length: place names or animal
names for example. Add in a few minimal
pairs.
b.
Use the 'learn'
script to record each student saying each word in isolation. When asked to type in the word use the
format <word>.<initials>, e.g. "cat.mh". This way we can keep track over which templates
the system uses to recognise unknown words.
2.
Testing
a.
Use the 'recog'
script to test out the system's performance.
Test with 10 of the vocabulary words chosen randomly by each
student. Record the number of correct
responses and the confusions. Which
words are most often confused? Could
you have predicted this? Does the
system tend to use templates spoken by the same speaker?
3.
Internals
a.
Use the 'recog2'
script to display the table of best-matched distances to an input. Try out some vocabulary words and
investigate the relative size of the smallest distances for the correct and
first incorrect answer. Can you see any
pattern in the ordering of scores?
4.
Architecture
a.
Sketch a
schematic diagram of the components of the recogniser. Look inside the scripts learn.bat and
recog.bat to see how they work.
b.
What do you think are the problems associated with extending this system
to 100, and to 100,000 words?