Editing
Dragon Dictate
(section)
Jump to navigation
Jump to search
Warning:
You are not logged in. Your IP address will be publicly visible if you make any edits. If you
log in
or
create an account
, your edits will be attributed to your username, along with other benefits.
Anti-spam check. Do
not
fill this in!
=== '''Improved Accuracy''' === The Dragon Dictate’ acoustic adaptability is based in a three-level phonology and a stochastic model of production:<ref>[https://doi.org/10.1177/002383099203500218 Mandel, M. A. (1992). A Commercial Large-Vocabulary Discrete Speech Recognition System: DragonDictate. Language and Speech, 35(1–2), 237–246.] </ref> ==== Three-level Phonology ==== This three-level approach alIows Dragon Dictate to adapt quickly to the user’s own speech, approaching the reference speaker’s performance within a few thousand words. * Phonemes: It comprises 24 consonants, three unstressable syllabic resonants, and 17 vowels, including diphthongs and syllabic /r/. * Phonemes in context (PICs): the fundamental unit of Dragon phonology, a PIC represents an augmented triphone, comprising: ** the phoneme before the one being modeled, as context ** the phoneme being modeled ** the phoneme after the one being modeled, as context ** the degree of prepausal lengthening of each phoneme due to its position in the word. * Phonetics elements(PELs): Most instances of English /s/ have a lot in common acoustically, and even a highly context-dependent consonant like /t/ can show significant constancy. We implement this similarity as a slice of a very coarse spectrogram, a steady state called a phonetic element, or PEL. ==== A stochastic model of production ==== It represents pronunciations as [[Hidden Markov Models]] that are built from the speech of a reference speaker according to three-level phonology, which in turn have been restructured into phonemes-in-context (PICs) and segments of PICs . Performance is also improved by use of digram language modeling. Experience with labeling has led to several observations on Germanic resonants and on polyphthongs. ==== Adaptation ==== When a person starts using Dragon Dictate it makes a model of his or her speech: a user model. Initially this is just a copy of its base model, which is derived from the reference speaker’s voice, but with every word spoken it adapts to the user’s pronunciation and usage. For example, Initial recognition performance, which is significantly poorer when the user is of opposite sex to the reference speaker, is improved by adapting the models in advance to a speaker of the user’s own sex through preadaptation.
Summary:
Please note that all contributions to MSc Voice Technology are considered to be released under the Creative Commons Attribution (see
MSc Voice Technology:Copyrights
for details). If you do not want your writing to be edited mercilessly and redistributed at will, then do not submit it here.
You are also promising us that you wrote this yourself, or copied it from a public domain or similar free resource.
Do not submit copyrighted work without permission!
Cancel
Editing help
(opens in new window)
Navigation menu
Personal tools
Not logged in
Talk
Contributions
Create account
Log in
Namespaces
Page
Discussion
English
Views
Read
Edit
Edit source
View history
More
Navigation
Main page
Recent changes
Random page
Help about MediaWiki
Tools
What links here
Related changes
Special pages
Page information