|
|
|
|
|
|
|
|
|
|
| 1. | Map the time-series of (e.g.)
f0 into a point (in a
multi- |
||||||||
| dimensional space). |
|||||||||
| 2. | Make a
cloud of points, one for each |
||||||||
| utterance
or syllable. |
|||||||||
| 3. | Find how
to divide that cloud to best |
||||||||
| reproduce
the human classification. |
|||||||||
| 4. | The
accuracy of the machine classification |
||||||||
| is a
measure of how much information is |
|||||||||
| carried by
each acoustic property. |
|||||||||