Speaker independent recognition of the eleven steady state vowels of British English using a specified training set of lpc derived log area ratios.
A data frame with 990 observations on the following 12 variables.
yClass label indicating vowel spoken
subseta factor with levels test train
x.1a numeric vector
x.2a numeric vector
x.3a numeric vector
x.4a numeric vector
x.5a numeric vector
x.6a numeric vector
x.7a numeric vector
x.8a numeric vector
x.9a numeric vector
x.10a numeric vector
The speech signals were low pass filtered at 4.7kHz and then digitised to 12 bits with a 10kHz sampling rate. Twelfth order linear predictive analysis was carried out on six 512 sample Hamming windowed segments from the steady part of the vowel. The reflection coefficients were used to calculate 10 log area parameters, giving a 10 dimensional input space. For a general introduction to speech processing and an explanation of this technique see Rabiner and Schafer [RabinerSchafer78].
Each speaker thus yielded six frames of speech from eleven vowels. This gave 528 frames from the eight speakers used to train the networks and 462 frames from the seven speakers used to test the networks.
The eleven vowels, along with words demonstrating their sound, are: i (heed) I (hid) E (head) A (had) a: (hard) Y (hud) O (hod) C: (hoard) U (hood) u: (who'd) 3: (heard)
D. H. Deterding, 1989, University of Cambridge, "Speaker Normalisation for Automatic Speech Recognition", submitted for PhD.