next up previous contents index
Next: Closed-set identification Up: Scoring procedures Previous: Scoring procedures

Notation

 

Registered speaker population

 

We will denote as ``X'' a population of m registered speakers:


equation15929

For scoring purposes, we will consider a set of test utterances.   We will use the term genuine test utterances for those which correspond to registered speakers, and the term impostor test utterances for those which are treated as belonging to impostor speakers. Note that the same speech segment can be used as a genuine utterance and as an impostor utterance, in some test configurations. Therefore, different notations can correspond to the same speech utterance.

Each registered speaker tex2html_wrap_inline46933 is supposed to have produced tex2html_wrap_inline46935 genuine test utterances , the set of which will be denoted as ``tex2html_wrap_inline46937'':


equation15942

where superscript k denotes the tex2html_wrap_inline46941 test utterance  of speaker tex2html_wrap_inline46933.

In the rest of this chapter, we will denote as c the total number of genuine test utterances , and as tex2html_wrap_inline46947 the proportion of utterances belonging to speaker tex2html_wrap_inline46933 in the test set , that is:
equation15960

With the convention:
equation15969
we will denote:
equation15979
Integer tex2html_wrap_inline46951 is the number of registered speakers for which there is at least one genuine test utterance .

Finally, we will denote as M the set of male registered speakers, as F the set of female registered speakers and as tex2html_wrap_inline46957 and tex2html_wrap_inline46959 the respective number of male and female registered speakers for which there is at least one genuine test utterance .  

Test impostor population

 

In the most general case, the whole set of impostor test utterances   can be divided in subsets corresponding to one among n impostors tex2html_wrap_inline46963 using the system with a claimed identity tex2html_wrap_inline46933. Hence the general notation


equation16002

to denote the set of tex2html_wrap_inline46967 impostor test utterances  produced by impostor tex2html_wrap_inline46969 claiming he is tex2html_wrap_inline46933.

Similarly to genuine test utterances, we will denote as d the total number of impostor utterances, and as tex2html_wrap_inline46975 the proportion of impostor tests by impostor tex2html_wrap_inline46969 against registered speaker  tex2html_wrap_inline46933, that is:


equation16021

We will also write:


equation16032

When the identity of impostors is not a relevant factor (or is unknown), subscript j can be dropped, and tex2html_wrap_inline46983 denotes the tex2html_wrap_inline46941 impostor attempt against registered speaker  tex2html_wrap_inline46933 (tex2html_wrap_inline46989). Conversely, in open-set speaker identification , impostors do not claim a particular identity; they just try to be identified as one of the registered speakers , whoever this speaker may be. In this case, subscript tex2html_wrap_inline46991 can be dropped, and tex2html_wrap_inline46993 denotes the tex2html_wrap_inline46941 impostor attempt by impostor tex2html_wrap_inline46969 (tex2html_wrap_inline46999). If, moreover, the impostor's identity does not matter, subscript j can also be dropped, and tex2html_wrap_inline47003 simply denotes the tex2html_wrap_inline46941 impostor attempt (tex2html_wrap_inline47007).

With the conventions:

tex2html_wrap_inline47009 if tex2html_wrap_inline47011 and tex2html_wrap_inline47013 otherwise
tex2html_wrap_inline47015 if tex2html_wrap_inline47017 and tex2html_wrap_inline47019 otherwise
tex2html_wrap_inline47021 if tex2html_wrap_inline47023 and tex2html_wrap_inline47025 otherwise

we will write:
equation16094
and
equation16121
Integer tex2html_wrap_inline47027 is the number of impostors for which there is at least one test utterance  against registered speaker  tex2html_wrap_inline46933, integer tex2html_wrap_inline47031 represents the number of registered speakers  against which there is at least one test utterance  from impostor tex2html_wrap_inline46969, tex2html_wrap_inline47035 is the number of impostors from which there is at least one impostor test utterance, tex2html_wrap_inline47037 is the number of registered speakers  against which there is at least one impostor test utterance , and tex2html_wrap_inline47039 denotes the total number of couples tex2html_wrap_inline47041 for which there is at least one impostor test utterance  (from tex2html_wrap_inline46969 against tex2html_wrap_inline46933).

Finally, we will denote as tex2html_wrap_inline47047 the set of male impostor speakers, as tex2html_wrap_inline47049 the set of female impostor speakers, and

tex2html_wrap_inline47051: the number of male registered speakers  against which there is at least one impostor test utterance 
tex2html_wrap_inline47053: the number of female registered speakers  against which there is at least one impostor test utterance 
tex2html_wrap_inline47055: the number of couples tex2html_wrap_inline47057 for which there is at least one impostor test utterance  (from tex2html_wrap_inline46969 against tex2html_wrap_inline46933, tex2html_wrap_inline46933 being a male speaker)
tex2html_wrap_inline47065: the number of couples tex2html_wrap_inline47067 for which there is at least one impostor test utterance (from tex2html_wrap_inline46969 against tex2html_wrap_inline46933, tex2html_wrap_inline46933 being a female speaker)

 



next up previous contents index
Next: Closed-set identification Up: Scoring procedures Previous: Scoring procedures

EAGLES SWLG SoftEdition, May 1997. Get the book...