Description Usage Arguments Details Value Author(s) References See Also Examples
For averaging detection trial lists that were obtained from different
experimental conditions, the amounts of trials in these conditions
should be weighted. This function computes a contingency table from
an sre
object.
1 | cond.table(x, cond, target = F)
|
x |
An object of class |
cond |
A list of factors making up the various conditions |
target |
Is column |
This function produces a table of the number of trials for each
condition in a trial list. Conditions are specified by all
combinations of all levels of the factors specified in cond
.
Conditions with no trials are left out of the table.
This table can be used in equalizing the influence of different
implicit conditions in a larger set of trials, such as NIST SRE-2008.
In order to do this properly, the analysis needs to be carried out
separately for target and non-target trials, this can be specified by
target=TRUE
.
This function is used by the functions det.sre()
and
plot.cond(), but can be used separately as well.
A data.frame with a columns for each factor, specifying the condition,
and the column Freq
that indicated the trial count.
David van Leeuwen
David A. van Leeuwen, “Overal performance metrics for multi-condition Speaker Recognition Evaluations,” Proc. Interspeech, Brighton, September 2009, pp 908–911.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 | ## a NIST 2008 SRE data set
data(tno.2008)
## Analyse by gender
cond.table(tno.2008, gender)
# gender Freq
#f f 59343
#m m 39433
## Analyse by acoustic condition
cond.table(tno.2008, list(mtype, mmic, ttype, tmic))
# mtype mmic ttype tmic Freq
#1 interview mic interview mic 34181
#4 phonecall phn interview mic 7350
#8 phonecall phn phonecall mic 8454
#13 interview mic phonecall phn 11741
#16 phonecall phn phonecall phn 37050
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.